by Ted Honderich
- The Determinism and Freedom Philosophy Website -

The connection between a mind and a brain is fundamental to the Philosophy of Mind, partly because it is often taken to include the the problem of the nature of a mind -- or, more particularly, the nature of consciousness. What follows here is an inquiry into this connection. It surveys the traditional and still orthodox answers. It is Ch. 2 of Ted Honderich's large work A Theory of Determinism: The Mind, Neuroscience and Life-Hopes -- which chapter is also Ch. 2 of the paperback Mind and Brain.

This inquiry into the relation between mind and brain begins with what obviously is required, an understanding of the nature of consciousness itself, one of the terms of the relation. We need to be in a way realistic about this. The inquiry then considers a half-dozen things that can be meant by saying the relationship is somehow simple -- that the mind just is the brain, that it is identical with the brain. 

Something else is put in place of that tired idea, or rather those tired ideas. It is that there is lawlike or nomic connection between simultaneous neural events and events of consciousness. This Hypothesis of Psychoneural Nomic Correlation makes use of a general account to which you can turn --
an analysis of lawlike or nomic connection in general.

The hypothesis of the connection between mind and brain has been thought to face many objections. These are a host of thoughts about the nature of consciousness, some of them of a Wittgensteinian kind, some owed to thinking about language. These are considered fully in what is in fact the largest part of what follows.

The hypothesis is the foundation but not the whole of this traditional story of the connection between mind and brain. The remainder of the story is set out in Mind and Brain Explanation. For the details of the cited books and articles, go to
References. The sections of the whole inquiry below are as follows:



      2.2.1 Definitions of Mental Events
      2.2.2 Mental Dispositions
      2.2.3 The Mental as Physical


      2.3.1 Dualistic Identity Theory with Neural Causation
      2.3.2 A Variant of Anomalous Monism
      2.3.3 Two-Level Identity Theory
      2.3.4 Local Idealism
      2.3.5 Eliminative Materialism
      2.3.6 Semi-Eliminative Materialism



2.5.1 Atomizing Consciousness
2.5.2 Resemblance of Nomic Correlates
2.5.3 Counting Mental Events
2.5.4 Linguistic Individuation
2.5.5 Conceptualization
2.5.6 Reidentification
2.5.7 Richness and Transience


2.6.1 Mechanism
2.6.2 No Conceptual Necessity
2.6.3 Explaining a Prejudice
2.6.4 The African Lad
2.6.5 Standardly Based Mental Episodes
2.6.6 Twin Worlds


2.7.1 Definitional Precedents
2.7.2 Holism
2.7.3 Indeterminacy of Translation
2.7.4 Domains


Each of the three hypotheses making up the theory of determinism to be expounded in the first part of this book has to do with mental events, which is to say events within consciousness. Thus we need, first, a conception of consciousness.

It is taken by some to be that which falls under statements of certain logico-linguistic kinds. As the story goes, these are statements, first, which have the distinction that their truth-values do not depend on the truth-values of statements they contain. This is so with the statement, about an episode of consciousness, that Freddie thought that the play was to begin at eight. Its truth, if it is true, does not depend on the truth-value of the contained statement that the play is to begin at eight. Compare the statement, containing two others, that he is grey-eyed and his shoes were made in Italy. A second group of statements has the distinction that their truth-values do not depend on whether there does exist a thing called for by a contained referring expression. This is so with the statement that he hoped for a seat in the stalls for under £10. Compare the statement that he sat in a seat in the circle. There is the statement-kind, thirdly, overlapping with the second, such that truth-value may be argued, to my mind unpersuasively, to depend on the inclusion of one particular referring expression, rather than another expression referring to the same thing. Take the statement that he wanted to see the German play of which he had heard some praise, and consider substituting the coreferring description 'the Leftwing play whose inept performance will make him wish he had stayed at home'. Compare the statement that he was present for a play by Brecht.

Attention is paid to this logico-linguistic conception of consciousness, in terms of independent and dependent truth-values, precisely because of its literalness and precision. Those virtues are in short supply in connection with conceptions of consciousness. Of course, in order to have something literal and precise in the desired sense, care must be taken to understand the given conception exactly as it is, without addition. It thus may need to be understood, incidentally, more strictly than its original proponent intended. (Chisholm, 1957) Consciousness is to be identified only in the way mentioned, by way of independent or dependent truthvalues of statements. The conception does not have to do, above all, with the statement's having some otherwise specified content or subjectmatter: that is, say, as being about thoughts, wants, or intentions, or conveying something about persons, minds, subjectivity, experience, or an inner world.

The sad fact, well enough known, is that the given conception of consciousness is open to serious counterexamples. (Heidelberger, 1966; Lycan, 1969; Davidson, 1980e) Not all of consciousness does fall under statements of the given kinds, and certain matters of nonconsciousness do. Consider the statement that I am in a jolly mood, which clearly is not one of the chosen kinds. So with the statement that I have an earache. Consider the statement that good weather made possible the early apearance of my snowdrops. Its truth-value does not depend on whether there actually occurred the happy event referred to by the second referring expression. Such counterexamples have led some philosophers attracted to the logico-linguistic conception to take the audacious course which involves, in part, simply ignoring the very large part of consciousness which does not fall under the conception. (Davidson, 1980e) We shall certainly not do that. Other philosophers have struggled to save the conception in various ways, piling on the epicycles.

There is a further fundamental objection, generally overlooked. Even if it were true that all of consciousness and only that fell under statements of the three kinds, or related kinds, we would by means of this truth get only a wholly uninformative conception of consciousness. We would have nothing like an analysis or other understanding of it. It is true enough to say that we would have no conception of it. In particular, we would not know what it was about consciousness that gave rise to its falling under the given kinds of statement. If a large part of consciousness is ignored in the mentioned audacious way, and the logico-linguistic criterion is used to discriminate the remaining part, it gives us no analysis or understanding of that remaining part. (Cf. Kim, 1971) If we set out to frame hypotheses about the chosen part of consciousness, we do so at our peril, flying in the dark. By way of a quite proper analogy, suppose there were a certain amazing truth, that all and only the members of one species of tree, say Turner's Oak, could be described in a given language by sentences of a certain deep structure, or simply a rare surfacegrammatical sequence: indefinite article, a curious gerund, preposition of a certain sort, and so on. We could then define this splendid species of tree as that which falls under
statements made by sentences of the given grammatical sequence in the given language. The definition would be wholly uninformative, and make error likely.

There are other conceptions of consciousness which also owe their existence to the pursuit of certain virtues. One is the familiar conception in terms of behaviour. It is very possible to sympathize in good part, if not entirely, with the psychologists and philosophers who were sceptical or uncertain of the worth of introspection as a source of knowledge, reluctant to attempt to deal with the unquantifiable, keen to be in accord with certain principles of scientific methodology, resistant to such free speculation as the Freudian kind, and who thus took the step of analysing ascriptions of consciousness into claims about no more than behaviour. It is now generally recognized that the attempt to analyse consciousness in terms of behaviour amounted to flying in the face of the facts. It was doomed to become, as it did, a kind of plodding in the face of the facts. (Maclntyre, 1971c; Mackenzie 1977; P. M. Churchland, 1984; Flanagan, 1984)

Like the logico-linguistic criterion, although in a different way, behaviourism attempts to analyse consciousness by looking elsewhere. If it is claimed that reasons are needed for this verdict, which might be doubted, one has to do with the fact that we do not take all the causes of behaviour to be other behaviour. When we take a belief desire, or intention to have caused an action, as we commonly do, we are not thereby explaining the action by referring to other actions, let alone mere movements. This would be true, evidently, if consciousness consisted in behaviour. Other objections to traditional behaviourism can be transferred from their use with other related conceptions, to which we now turn.

One of these, sometimes itself referred to as 'behaviourism', but better named causalism, takes consciousness to consist in no more than episodes which enter into certain causal relations. These episodes are causes of behaviour, or effects of stimuli, or both. (Smart, 1959; Armstrong, 1968, 1980; Lewis, 1966, 1972) What is distinctive about this view is not that it takes the episodes of consciousness to stand in such causal relations. That at least some of them do so is a platitude. Nor would the view serve any purpose, including the purpose for which it was devised, if it were no more than the idea that particular episodes of consciousness—say desiring—can be identified by their causal roles, where such identification does not give us their nature, or all of their nature.

What is distinctive is the idea that consciousness can be adequately described in terms of causal episodes. That is, the whole of its nature or reality is at least adequately given by this description. Furthermore, such a term as 'causal episode' is to be understood in a way which excludes certain assumptions or implications. A causal episode, in the given sense, is not in itself a physical process, or that particular kind of physical process which is a neural process—a process of the Central Nervous System of a person or other organism. Nor is it in itself what would ordinarily be called a mental process, say a process involving the Self or a process of which the person in question has a unique kind of direct awareness. Causalism, rather, gives a 'topic-neutral' conception of consciousness.

Causalism can be considered together with something else. It lies in the immediate background of another family of views of consciousness which, at the time I write, is propagated with the zeal which once went into behaviourism. This family of views, which derives from many of the same commitments as its predecessors, but owes a great deal to the development of the computer, has among others the rather unenlightening labels functionalism, cognitive science, cognitive psychology, and artificial intelligence—several of which labels obviously have other uses. (Putnam, 1975; Fodor, 1968; Dennett, 1978, 1984; cf. Boden, 1977) To proceed in terms of a kind of model of this family of views, or many of them, what we can call functionalism shares with causalism the idea that mental episodes are to be understood relationally, in terms of their relations to other things. Somewhat less clearly put, mental episodes are to be understood in terms of their roles or functions visavis other things.

Functionalism as we shall understand it differs from causalism, first, in not restricting the relata in question to two categories: external input or stimuli and external output or behaviour. This, certainly, is an improvement. It gives what is in a sense a richer account of a mental episode by including relations with other strictly mental episodes and facts—these too, of course, to be understood relationally. Very roughly, my wanting the window open a moment ago is to be understood not only in terms of the stuffiness of the room (the stimulus), and my subsequently opening the window (behaviour), but also in terms of various beliefs, attitudes, and the like, including certain ordinary causal beliefs about open windows and perhaps attitudes having to do with propriety and the neighbours.

Functionalism, as we shall understand it, also differs from causalism more fundamentally: in its conception of the nature of all the relations in question. They are what can be called logical, conceptual, or formal relations rather than causal relations. They are typified by the relations whose extended statement consists in a machine table, or flow chart, or computer programme. A table, chart, or programme, in this sense, is not a tape or a disc but a sequence of propositions, which is to say abstract objects. (1.1) To speak differently, causalism takes mental episodes to be occupants of causal roles, but functionalism takes them to be logical or Turing Machine states, or logical or computational processes. In the words of two proponents of the view, it is 'the emerging view of the mind as software or programme—as an abstract sort of thing whose identity is independent of any particular physical embodiment'. (Dennett and Hofstadter, 1982, p. 7) It is indeed a consequence of such a conception that consciousness is not necessarily an attribute only of persons and other organisms. Consciousness is not tied to neural structures and events. It is accepted by functionalists that if any computers, or indeed any artifacts or entities whatever, can be said to pass through the requisite states or processes, they must therefore be conscious. Passing through the states or processes is being conscious. Causalism taken by itself has a related consequence, which has been obscured by the addition to it of what are really independent ideas, notably the arbitrary idea that conscious episodes having been conceived causally are then to be identified with neural structures and events.

There is no shortage of objections to both causalism and functionalism. Here is a quick list.

(i) If I am anaesthetized I do not feel the prick of the needle. Suppose, however, that I simulate or even by coincidence pass through whatever sequence of states is taken to be sufficient for the conscious episode of feeling the prick of the needle. It follows, on the causalist or functionalist view in question, that I do feel the prick of the needle. The conclusion, it seems, is intolerable.

(ii) Suppose that my private visual experience is strikingly atypical in that systematically I see green where others see red. That is, I am caused, perhaps by a deformation of my visual cortex, to have the visual experience which others describe as seeing something green in colour when the thing in question is what gives rise to their seeing it as red. This experience of mine, further, thanks to my training, stands in satisfactory relations to other things. I stop at the traffic light when I have a visual experience which others would describe as seeing a green light. I, of course, describe it as red. It follows from causalism and functionalism, seemingly absurdly, that our private visual experience is identical.

(iii) Suppose I do not understand Chinese, but, solely on the basis of the shapes of the letters, and rules connecting them merely as shapes do the right things. For example, I pass through a sequence of states identical to one passed through by a native Chinese speaker in being presented with a question in Chinese and giving the answer in Chinese. It follows, on causalist and functionalist views, that I am understanding Chinese, which again seems absurd.

(iv) Differently, it seems difficult to accept that consciousness is tolerably conceived when it is so conceived that it follows that anything that can be regarded as passing through certain sequences of causal or logical states is conscious. As suggested by one formidable opponent of the views in question (also the author of the objection having to do with understanding Chinese) we do not so understand consciousness that we can believe that it must be possessed by a set of water pipes, paper clips, and old beer cans, provided that they instantiate a certain programme. (Searle, 1980, 1981, and 1984, Lect. 2) That such a thing could instantiate such a programme, as already remarked, necessarily is allowed by the proponents of causalism and functionalism.

(v) To add something unfamiliar, functionalism will need to explain how it avoids the absurd conclusion that consciousness is, so to speak, not merely independent of biology, but entirely independent of all instantiations of programmes and the like, which is to say entirely independent of persons, computers, and all other spatio-temporal things. A sequence of logical or abstract states, in whatever way it exists, exists whether or not instantiated. It does not come into being only when instantiated. Functionalism, having fled the traditional mysteries of the mind, appears to rarify consciousness, to speak quickly, into a Platonic universe, out of space and time. In the enterprise of seeking to understand consciousness as something more manageable and decently scientific than what it calls ghostly stuff, it is understood as yet less than ghostly stuff.

The objections have been very lightly sketched, and, inevitably, there is the possibility of attempting replies. (Abelson et al., 1980; P.M. Churchland, 1984; Flanagan, 1984) I have little doubt, however, that the objections will elicit or reinforce a fundamental conviction in readers, a conviction as good as universal. It is, in sum, that consciousness is not adequately conceived in the given causal and logical ways. Like the logico-linguistic and behaviourist enterprises, causalism and functionalism attempt to define or analyse consciousness by looking elsewhere. There is no doubt that our conceptions of various kinds of conscious episodes do indeed include relational components. That is not to say that we take such episodes to consist in relational facts. (Cf. T. Nagel, 1986, pp. 78; McGinn, 1982, p. 7.) They allow for, and require, a certain realism. (Cf. Dummett, 1978)


It is sometimes supposed, by those who do allow that consciousness is a reality not caught by any of the doctrines we have considered na none the less nothing enlightening can be said of it. It is a kind of given, something of which we do have direct acquaintance but it impossible to give any analysis of it. It is sometimes supposed more extremely, that direct acquaiintance or introspection is so uncertain or fallible that not merely no analysis but nothing of value can be got by means of it . We are in possession of no concept of consciousness at all. The conclusion may be supported by psychological evidence of introspective error (Lackner & Garrett 1973) which evidence seems incapable of supporting any such general conclusion. The conclusion may derive instead from a general scepticism about all our perceptual capabilities, internal and external, and also a certain amount of fast philosophy. An example of the latter is the argument that since we are mistaken about secondary qualities -- the real world does not contain red, sweet, and hot things but only certain electromagnetic stereo chemical and micromechanical properties -- so we can have no faith in our direct awareness of our conscious events, states, and processes (P. M. Churchland, 1984, p. 29) The conclusion that we can get nothing of value, no conception, by direct awareness is yet more excessive than the conclusion that we can get no analysis of consciousness. The best reply to both conclusions consists in providing what it is maintained we cannot have.

Several properly ambitious philosophers attempt to bring consciousness into view by speaking, with respect to a conscious thing of what it is like to be that thing. (T. Nagel, 1979a, 1986, Ch 1; Sprigge, 1971, 1982, 1982; cf. Farrell, 1950) Consciousness, we are invited to say, is a way it is like to be, or, perhaps better, a way of being The consciousness of a bat is what it ls llke to be a bat, that way of being. This perception strikes one as promising, but the impression may be evanescent. It is difficult to avoid the further thought that the given characterization of consciousness is elliptical, and that when it is filled in, as it must be, we are no further ahead, but have our definendum turning up in the definiens. It seems clear, that is, that we are to understand that a way of being is something possessed by or intered into, say, by us, but not by sticks and stones. But then 'a way of being' when filled in, becomes 'a way of being conscious'. Consciousness we find ourselves saying, is a way of being conscious. There is a second consideration. If the understanding of  'a way of being' as elliptical is resisted, as perhaps it can be, and the term 'a way of being' taken to be serviceable as it is, there is another difficulty. To speak of my way of being, or what it is like to be me, seems to be to speak of what distinguishes my conscious life from the conscious lives of others. That is, it is not to characterize consciousness generally. But that is precisely our aim.

What is offered may owe something to Brentano's familiar characterization of consciousness. (Beyond doubt that characterization lies behind the logico-linguistic criterion considered earlier.) Consciousness is said by him to be, fundamentally, activity which has reference to a content, or activity which is directed upon an object. (1973 (1874))

A content is not a spatio-temporal state of affairs, such as the play's beginning in the Olivier Theatre at eight. If someone had the right thought, that the play did begin there at eight, then their thought did have a content that was true, but that content was not identical with the actual state of affairs. The content was precisely the same sort of thing as the play's having begun on Saturn with Helen of Troy in the audience, which never happened. The objects in question, similarly, do not exist in the spatio-temporal way of ordinary things. Even in the cases where what we want are things which exist in the ordinary way, these things are not to be identified with the mentioned objects. What lies behind these conceptions of content and object is the true idea that the nature of consciousness itself may not be different at all between the times when we think truly and when we think falsely, or when we want what exists and when we want what does not exist.

As for the mentioned activity, an attempt must be made to understand it in a way consonant with what has been said of contents and objects, that they are not states of affairs or ordinary things, and also with a further fact, that there are various modes of consciousness. With respect to the first point, none of the activity—no instances— can be identified with ordinary perception, plain seeing, which consists in activity in a plainer sense, having to do with ordinary things. The activity posited by Brentano, secondly, is common not only to seeing and other perception, thinking in the sense of deliberating or judging, and also remembering, intending, and deciding, all of which are ordinarily thought to be in different ways active, but also pain and other sensations, dreaming, experiencing an emotion, and being subject to a mood, all of which are thought to be in different ways passive. As for the ideas of having reference to and being directed upon, they too must be given sense consonant with the nature of the given contents and objects and the various modes of consciousness.

It has often been objected to this account that it does not fit all of consciousness, that some of consciousness refutes it. (K. Campbell, 1970, p. 23) It is allowed that it is enlightening or tolerable to speak of hoping, perceiving, thinking, and a good deal else as having objects or contents in the given sense. What, however, of states of feeling, perhaps being depressed or feeling good about nothing in particular? What of sensations, perhaps the warmth one feels while sitting in a sunny window? Surely these cannot be said to have objects or contents in the given sense.

Anyone who takes this line confidently must surely understand more about the mentioned objects and contents than was conveyed above. He must understand something more precise than Brentano explains. It is true, certainly, that feeling good for no particular reason, and the sensation of warmth, do not involve objects or contents that can be said to be determinate in a certain sense. There is a great difference between the feeling and the sensation and, on the other hand, having the thought that one's father wrote a pamphlet about the consistency of Christianity and Communism. Still, we are not debarred from attempting to conceive of an indeterminate object or content. Nor are we debarred from using 'content' rather than 'object' where the former is more natural.

A good deal of philosophical writing about what is called intentionality suggests that it is likely that this line of objection, that objects or contents are sometimes missing, has another wholly different root, in a certain confusion. Brentano evidently has in mind a relation between something not explicitly mentioned, that which is active, and, on the other hand, contents and objects. This is the relation of 'reference' or 'direction'. There is another matter. Some of the contents and objects, we can take it, are distinguished from the rest by being representative. That is, they involve a second and wholly different relation, a semantic or intentional relation between themselves and whatever they represent. It is precisely true of hoping, perceiving, thinking, and a good deal else that it can be conceived in terms of such representative objects or contents. It is precisely untrue of certain states of feeling and sensations. However, the absence of an object or content of a representative kind must obviously not be confused with the absence of an object or content. Again, the absence of the representing relation is not the absence of the more fundamental relation.

My present aim, in fact, is not to rely on Brentano's conception of consciousness, but to use it to introduce my own somewhat related one. It, like his, has the recommendation of being a realist conception, which is to say one which does not withdraw from its subject-matter.
Realism about the mental, to speak freely, seems to me essential, and the opposed lookingelsewhere, whatever its roots, a kind of fastidiousness or want of nerve which is futile. (Cf. Searle, 1984, Lect. 1.) My conception is related, also, in that both conceptions derive in a direct way from our pre-theoretical, firstperson grasp of consciousness, as cannot be said of any of behaviourism, causalism, or functionalism, nor really of the logico-linguistic criterion, derived though it is from Brentano. My alternative account, a minimal one, seeks among other things to give fewer hostages than Brentano's to philosophical fortune, or anyway to philosophical doubt.

To think of any of one's conscious episodes in the moment after it has happened, is to think of a certain duality, one which has nothing to do with dualistic doctrines of mind and body. Think of feeling a sensation in one's knee—or noticing a cup, having the usual inattentive visual experience of a room, feeling good or depressed about nothing specific, being struck by the fact of a recent death, wanting to go to bed, deciding not to, momentarily intending to watch the news on television, picturing a face, thinking a question or a sentence, writing one, having a dream. Certainly there is great diversity here, and causal and other relations enter into our various conceptions of these things. It is also indubitable, and a fact of which philosophers and psychologists of different inclinations have made different things, that to think of any episode of experience is to think of two elements, two elements within the experience.

One of these elements turns up in each of one's conscious episodes, or is of a kind such that each of one's conscious episodes contains an instance of the kind. That is, there is that which is common, or of one kind, in each of the experiences of feeling a sensation, being depressed, thinking a question, taking a decision, and so on. This is not to assert, certainly, that there exists some entity, outside an experience or episode, which experiences or possesses it. The second element in each of one's conscious episodes is almost always different—different from its counterparts in all other episodes. The first element can be referred to as the subject, the second as the content.

The terms are of no great importance, and do little more than mark what is the incontrovertible fact of the duality within conscious episodes as we recall them. They mark our perception of that fact, and are not to be taken as carrying theory with them—not even so much theory, perhaps, as is suggested by William James in speaking of a 'dualism' of 'mind knowing and thing known'. (1890, p. 214) The subjectobject relation is not to be characterized as the subject's attending to the content, or anything of the kind. To characterize it in this way would be, it seems, to introduce the definiendum into the definiens. Certainly, to repeat, the terms 'subject' and 'object' are not to be taken as carrying with them the Cartesian theory of a subject in the sense of a simple indivisible mental substance whose identity over time is primitive and irreducible. (McGinn, 1982, pp. 1212) Even acute defences of Descartes' argument for an ego (B. Williams, 1978, pp. 95100), prove to be ineffective (Parfit, 1984, pp. 2256), but their failure does not affect the present view.

It is to be admitted that any particular conscious episode, in the moment of experiencing of it, seems to consist at least largely in but one thing—as, say, when the address '4 Keats Grove' comes to mind. One's experience is certainly in some sense not of two things. No doubt this is the truth on which Hume relied in denying the existence of the Self as commonly conceived. However, to think afterward of the having of the thought is to think of two things, both somehow integral to it. Having an idea of having had an idea is having an idea of two things. This is as much a truth about such episodes as unfocused depression. It is not as if one's depression could be captured in recollection by the idea of bare content. Nor is it possible to capture one's experience by the idea of a bare subject. That, surely, is inconceivable.

In any case, it is not quite true that the subject can be perceived, so to speak, only in recollection. Certainly the subject, sought from within a particular conscious episode, is peculiarly recessive. There is a difference, evidently, between the manner of the occurrence or contents in consciousness and the manner of the occurrence of the subject. Contents are at least to the fore. However, it seems impossible to deny that each of us also has a sense of a self, or a centre of awareness, somehow within and integral to our ongoing experience, and such as to give a kind of unity to it.

The duality is consistent with the mentioned diversity of contents. They are indeed of different characters, (i) In the case of visual and other perception, they are bound up with ordinary things—cups, rooms, and landscapes. How they are bound up is the philosophical problem of perception, into which we shall not enter, (ii) Contents, in the case of sensation, lack such a connection, but are in a special way qualitative and are somehow assigned a bodily location. These contents of sensation are, despite their difference from the contents of perception, no less distinguishable from that element in an experience which is the subject, (iii) So too are distinguishable the contents of moods, such as feeling good about nothing specific, despite their difference from the contents of both perception and sensation. Feeling good about nothing specific is not, so to speak, an unattached or freefloating phenomenon, unpossessed or unhad. (iv) Finally, and clearly, all of the contents so far mentioned are different in character from those which get definition by way of language—or rather, get definition only or more explicitly by way of language. Here, as in the example of thinking a question, we have what were earlier called representative contents, or the most important ones. They are of great importance in any full characterization of consciousness, and are rightly the subject of inquiry of several kinds, sometimes in the endeavour to give a limited characterization of the mind. (Searle, 1983, Harman, 1973, Fodor, 1979) None of this diversity in contents, to repeat, should distract us from perceiving the single duality within consciousness. That what is on one side is in fact a category of diverse contents does not take away from the fundamental duality.

Consciousness, then, to speak somewhat differently, consists in general in the relation of subject and content. This relation, further, is such that its terms cannot exist independently of one another. They can occur only in the given relation. Each term depends for its existence, that sort of existence it has, on being in relation with the other. There can be no subject without a content, and no contents not in relation to a subject. It is perhaps audacious, given the undeniable uncertainty in which we find ourselves in thinking of consciousness, but it may be possible to regard this mutual dependency as a matter of nomic correlation of one of the kinds specified earlier (1.5), although nomic connection of a uniquely primitive character.

Given the terms 'subject' and 'content', which lack close definitions, it would perhaps be yet more audacious to speak of logical or conceptual dependency. Still, it is tempting to think of a kind of logical connection. It seems that any attempt to conceive of a content of a mental event, separately from its inherent subject, is in a way futile. It appears to be futile in that inevitably one finds oneself conceiving something else. If I now consider an event of a moment ago, my idle contemplation of the cup on my table, and attempt to subtract from my present conception only a part of it—the subject within the event of a moment ago—and to hang on to the remainder, I am in fact left with something other than the content of the event. I am left with the cup—the idea of that ordinary physical object—or a proposition about the cup. These, however different, are independent existences. To succeed in thinking instead of the content, it seems, is necessarily also to think of something else, to think of that which exists for something else.

We shall leave the matter as we have it. Consciousness, to come to a kind of summation, consists in general in this interdependent existence of subject and content. If we have the content primarily in mind, and are willing to strain language somewhat, consciousness can be described as the existencefor of contents, or their existenceto. Contents exist /or or to subjects. To go further, consciousness is for-ness or to-ness. We can as tolerably say, however, having the subject primarily in mind, that consciousness consists in the existence-through of subjects. They exist only through contents. Consciousness is existencet-hrough. Perhaps the straining of language is not worth the effort.

More might be attempted (Ayer, 1954a; Alston, 1976; Hannay, 1979; Wilkes, 1978) but we have, I submit, gone some way in analysing consciousness with the general conception of it as the interdependent existence of subject and content. Still, admissions cry out to be made. The conception is not literal and precise, in the sense in which, say, the logico-linguistic criterion is such. Nor does it satisfy all the demands mentioned above (p. 73) in connection with behaviourism, causalism, and functionalism. Not all of those, of course, as already noted, are reasonable—above all, any wholesale dismissal of what we get by direct awareness is entirely futile. As will be evident, it is my view that we have a choice between attempting to satisfy certain demands and failing to get hold of our subjectmatter, or relaxing the demands and getting some hold on it. The latter course is not merely preferable but necessary.

2.2.1 Definition of Mental Events

Three important matters remain, of which the first has to do with the things we discriminate within consciousness—the parts, segments, or passages of consciousness, as distinct from subject and contents. I have spoken so far in an ordinary way of episodes of consciousness and of experiences, occasionally of modes and states of consciousness, and also of sensations, moods, perceptions, desires, thoughts, intending, deciding, and so on. Each of these things, whether taken as type or token, is itself a matter of subject and content. It is fairly common to speak of these various things as events, states, and processes of consciousness, but we do not have a single wholly general term for them. I shall continue to speak in these and the more ordinary ways of the things in question, but it will be at least economical to have a single wholly general term for all of them. Or rather, it will be economical to use one of these terms in a yet more general way, to cover all of these things and also things that do not fall naturally into any of the mentioned ordinary or semi-technical categories—or ones like them. This will be the term 'mental event'.

It will cover anything whatever that occurs within consciousness. If it is supposed that there is a want of definiteness about the idea of something's 'occurring within consciousness', we can as well speak of anything whatever of which there is a logical possibility that an individual can discriminate it in his or her or its consciousness. The discriminating of something within consciousness is to be understood as a case of remembering. It is fixing attention upon something afterwards, thereby assigning it, by way of its content, to a type. I cannot, it seems, both be aware of my hand and at the same time aware of that awareness. To attempt to perform the double act is instead to oscillate between the two. It is of course necessary that the alternative definition of a mental event not be of something which in fact can be recalled afterwards, but of something of which there is a logical possibility of recall. It is clear that we do not in fact have the capability of sharply recalling more than a very few of our mental events as defined, even immediately afterwards. This is so, to speak metaphorically, since almost all mental events, although not the ones in which we are principally interested, are elsewhere than at the focus of consciousness.

It will be as well to make explicit the third-person criteria of individuation for a mental event, criteria which are implicit in what has just been said. A mental event is individuated, first, as being within the consciousness of a given individual. Secondly, its individuation is a matter of content and/or temporal location. Two tokens of the same type of mental event, all of whose tokens are indistinguishable to the owner in terms of content, are distinguished by temporal location. In the case of many mental events, such as the elements of one's visual field, their discrimination is one with, or bound up with, the discrimination of ordinary things, their features and so on. The distinguishing of mental events by content is of course open to error, including error owed to memory and the sorts of error established by psychological research into introspection or direct awareness. It is perhaps not too hazardous to say, despite this, that our capability is such that there is not much else that we can do better. The perceptual distinctions we make are of the very stuff of our existence, and in some way the foundation of all else.

Given what has been said, a mental event can be simple or complex and of any duration. It may then be a momentary thought, a fleeting hope, the colouring of a feeling, the forming of an intention, a decision, a choice, a continuing sensation, an inactive or an active sensation (pp. 221 f.), a persistent mood. Given what has been said, further, a mental event is not necessarily what is sometimes called a whole mental state, which is to say all of a person's consciousness at or for a time, but typically is a part of a whole mental state, a part which itself has parts. Mental events, like most events, are composite—they contain or are constituted of other mental events. Certainly there is no need, as is sometimes supposed (Thorp, 1980, pp. 434), to consider whole mental states in connection with the relation between the mental and the neural.

Also, given what has been said, that a mental event is wholly within consciousness, it is also true that in speaking of the mental life of people, in a large sense of the term, we are speaking of more than mental events. In this sense of the term, my mental life contains my knowing things, and remembering things, and seeing things. To claim that I know that my postal code is NW3 2RT is evidently to claim or be committed to more than that I believe it. It is to claim or be committed to the fact external to me that my code is NW3 2RT. To say that I see the squirrel with the black nose is to be committed not only to the existence of a visual experience, but also to the existence of something else. Mental events are contained in rather than identical with what can be called personal epistemic facts. It is mental events that raise the questions with which we are concerned.

As in the case of the physical events considered in the last chapter in connection with causation and other nomic connection, mental events strictly speaking are to be regarded as individual properties or sets of such properties. That is not to say that necessarily they are in any sense physical, that a theory which identifies the mental with the physical, or some of the physical, is forced upon us, whatever our attitude to such a theory. It is to say, rather, that mental events have a kind of dependent existence. (1.1) We have in fact already come close to the view that mental events are at bottom individual properties, or rather sets of them, in allowing that the interdependent existence of subject and content may be a matter of nomic correlation.

Mental events, however language enters into their specification, as of course it does (McGinn, 1982, Ch. 4), are not individuated only by our descriptions, or at any rate our ordinary descriptions. We have in consciousness a finer mesh than that. Our awareness is finergrained than our language. I can say truly on two occasions 'I'm getting alarmed' and yet the mental events not be of the same type. That is, there might have been some discernible difference in content between them. Any will do. They will, of course, despite this, have been 'of the same type' in some looser sense. It is likely to be, even, that two mental events are not of the same type on two occasions when I give a more explicit description of my experience, perhaps that I was thinking that my daughter is a quick thinker. Whether two or more mental events count as of the same type in our sense, to repeat, is a matter of whether there is a discriminable difference in content, any such difference, whether or not the events are included under the same more or less general description.

It is not to be supposed, either, that the same type of mental event cannot occur in the lives of two or more individuals. That is, it is a logical and conceptual possibility that two mental events, one in my conscious life and one in yours, be of the same type. Certainly there would be overwhelming or very great practical difficulty in the way of establishing that such a thing had happened. What we would need to do is establish about Brown's mental event and Green's mental event that if either man also had the other's experience, he could not distinguish in content between the two events.

2.2.2 Mental Dispositions

What has been said in this section concerns consciousness itself, and hence, as remarked above, does not concern all of our mental life. It does not concern all, either, of what can be called the mental. This is so, since the mental in a wide sense, as often conceived, includes not only the facts of consciousness but also dispositional facts. There is the dispositional fact that I believe my name is what it is, which is a fact about me when I am not thinking of my name. So too there are dispositional facts of desire, fear, intention and so on. For there to be such a fact about me is for it also to be true, roughly, that in certain circumstances I would consciously believe, desire, fear, intend, or whatever.

Determinism has to do in part with consciousness itself, most notably with the mental events of choosing, deciding, and the like, and with mental events which precede them. It also has to do with what else gives rise to all these events, and here the given dispositional facts are of large importance. They will get attention in connection with the second hypothesis of the theory of determinism of this book. In answer to the question of what sort of facts they are, the answer will be given that they are neural facts, facts about the Central Nervous Systems of ourselves and other species. The determinism to be considered here, then, like other determinisms, has to do with all of the realm of the mental as widely conceived. While it is true that we have until now been concerned with consciousness, and used the term 'mental events' in the ordinary way so as not to include all that may be included in the mental, the rest of the subjectmatter of that domain is not thereby excluded from our coming reflections. That is not to say that we shall take dispositional mental facts or what is called the unconscious or the subconscious to be 'another realm of consciousness'—that we shall suppose there is consciousness and then inaccessible consciousness and then the brain, the latter involving dispositional mental facts. If there is a similarity between mental events and the given dispositional facts—that both enter into the explanation of behaviour—that is no sufficient reason for blurring the distinction between them, for etherealizing the dispositional facts.

2.2.3 The Mental as Physical

All three hypotheses of our determinism concern the physical in several ways, including that part of the physical which is the neural. Neural facts, preeminently facts about brains, will be the subjectmatter of Chapter Five. How, in general, are physical events, facts and so on to be understood?

According to a long and dominant tradition, the physical is bound up with the spatial. To follow a recent and acute account in this tradition the physical is what is spatial and also has certain further properties. (Quinton, 1973, pp. 4653) More precisely, with respect to the spatiahty, since we wish to exclude points, lines, and two-dimensional surfaces, we shall take it that the physical occupies a continuous three-dimensional region of space—that it is voluminous or geometrically solid. Hence what is physical has shape, however irregular, and also size and location. Further, in its geometrical solidity, what is physical is the unique occupant of its space. These primary qualities are insufficient by themselves for a physical thing, however, since they are possessed by a volume of empty space. Are we then to add only that what is physical must also be observable, that it must possess some secondary quality—colour, hardness or solidity, texture, smell, taste? To do so would exclude many very small particles whose existence is well established in physics, and also physical forces, such as magnetism. The obvious solution is to take into account that these stand in causal or other nomic connections with somehow observable space-occupants.

We thus come to the definition of the physical as (i) that which is an observable space-occupant or (ii) a space-occupant which is in causal or other nomic connection with an observable space-occupant. There are other views of the physical related in several ways to this. The physical is defined, for example, as that which is recognized by science, that which is 'an essential part of the coherent and adequate descriptive and explanatory account of the spatio-temporalcausal world'. (Feigl, 1958 p. 377) It may be defined, similarly, in terms of falling under scientific law, which is to say in terms of entering in nomic connections (E. Nagel, 1979, pp. 146 ff.)

We shall take it, as anticipated earlier (p. 16), and in some agreement with these views, that all causes and effects are physical events, but we need not look further into the views. Given the conception we have, are mental events as we have conceived them excluded from being physical? They are not. That an event is characterized in terms of the interdependent existence of subject and content does not entail that it is not physical according to such a conception of the physical. We are not forced to the conclusion that mental events are somehow outside of the physical world. Nor, however, are we prevented from embracing that conclusion. It does not follow from the conception of an event in terms of interdependent subject and content that it is physical in the sense defined.

In fact, however, there are independent good reasons to suppose that mental events are physical in the given sense. For a start, there is a clear answer to the question of where thoughts and feelings take place, if not so fine an answer as we might like. They take place in and indeed are sometimes identified in part by a location, which is to say the location of a person or other organism. I may indeed remind my friend of a thought I had by referring to it as the one I had on first wakening in Venice. I can indeed mean the thought that occurred there, rather than mean the thought that occurred nowhere on an occasion when my location was Venice. Further, there seems no reason to suppose that mental events do not also occupy space—as do other events of which we cannot specify the minute space or the minute and myriad spaces which they occupy. (Thorp, 1980, pp. 60 f.) Finally, as we shall see more fully, and as generally is supposed, there are causal relations between mental events and observable space-occupants.

Certainly we may feel a resistance to the idea that mental events are in space. The resistance, to my mind, does not amount to what can properly be called an argument. It may amount to the irrelevant truth, that prepositional or abstract objects, which enter into the contents of many mental events, are not in space. Whatever is claimed of these traditional philosophical mysteries, including the claim that they enter into mental events, they surely cannot be regarded as parts of them. Only if they were parts could they generate an argument for the nonspatiality of mental events. The resistance to taking mental events as spatial seems also to have to do with something less specific, pertaining to those of them which are perceptual and also those which are representative. We are tempted, perhaps, to associate them with what they are of, or what they represent, and hence to remove them from where they otherwise would seem to be, and hence to be lost for a location to which to assign them. However, there can be no ground for the initial association: my perception of the pen in my hand is not that pen, however much I 'enter into' the latter.

The idea that mental events are physical, however, is not the idea, whether true or false, that mental events are in fact identical with the particular kind of physical events with which they have often been and are still identified, which is to say neural events. These, as remarked, are events of the Central Nervous Systems of ourselves and other species. They are, by way of the summary description on which we shall depend until Chapter 5, electrochemical events. We now turn to the question of the relation of mental and neural events, and first to propositions to the effect that they are indeed identical.


What we have so far enables us to rule out certain purported answers to the question of how mind is related to brain, or, more precisely, how mental events are related to neural events. Any purported answer which does not begin from an adequate conception of consciousness and hence of mental events is in fact no apposite answer at all—good, bad, or indifferent. It does not get into touch with half of its subject matter. Thus there is no apposite answer to be had which begins from the conception of mental events as merely behavioural events. Quite as certainly, any purported answer which begins from the logico-linguistic criterion strictly used, or causalism or functionalism, is no apposite answer, since it too in fact does not deal with half of the subject-matter. (Cf. T. Nagel, 1986, pp. 1314.)

In particular, we have no answer to the question of the nature of the psychoneural relation in what can be called Causalist Identity Theories, often taken as the preeminent Identity Theories. In them it is said that certain events causally conceived are in fact identical, which is to say numerically identical, with neural events. (Smart, 1959; Armstrong, 1968, 1980; Lewis, 1966, 1972) The latter proposition of identity might be true—all that is required for its truth is that neural events cause certain things and are effects of others. But that proposition of identity can be true without its also being true that mental events adequately conceived, realistically conceived, are or could be identical with neural events.

Again, we have no account of the psychoneural relation when it is said, roughly, that certain formal sequences are instantiated or realized in sequences of neural events, as they may be in other sequences. (Putnam, 1975; Fodor, 1968; Dennett, 1978) The Functionalist proposition of instantiation or realization could be true without its being the case that mental events adequately conceived stand or stand only in the given relation to sequences of neural events.

What criteria, other than the criterion of sticking to the subject, must be satisfied by a satisfactory account of the psychoneural relation? A second is suggested by, although it does not depend on, certain ordinary beliefs. Have in mind, for a moment, all the neural sequences which took place in a man while he was wondering if the chair he was sitting on was level. Readers hesitant about the nature of neural facts can have in mind, more generally, all the nonmental events which took place during the time he was having the momentary experience which he would then have described in the mentioned way. Might the neural or all the nonmental sequences have been occurring, just as they were, and the man not have been wondering as he was? Might he have been thinking about something wholly different, say the age of his brother? Might he even have been wondering about what he was wondering, but in some slightly different way?

If the question is taken just for what it is, and no more, perhaps few of us will answer yes. What most of us are inclined to accept is that it was not a matter of coincidence or accident that some of the neural or nonmental processes were occurring as they were while the man was wondering as he was. Whatever the nature of the connection between the two things, there was a connection, rather than two things accidentally or coincidentally occurring together. The proposition is supported by a vast amount of entirely ordinary experience of connections between conscious life, one's own and that of others, and nonmental facts.

What can be called the conviction of psychoneural intimacy or perhaps the axiom of psychoneural intimacy is suggested by these ordinary beliefs, but is a good deal closer to many neuroscientific assumptions and propositions, as we shall see. (5.2) This second criterion of a satisfactory account of the psychoneural relation, again to express it quickly, is that mental and neural events are intimately related, that they are bound up together—but something more definite needs to be said.

Each type of mental event stands in a certain relation to types of neural events in a larger or smaller locale or collection of locales in the brain. That is, each specific type of mental event occurs simultaneously with a specific type or specific types of neural events or neural sequences in the locale or locales—call this co-occurrence. Further, this co-occurrence is in some way guaranteed, which is here to say that the events of the two kinds are somehow or other necessarily connected, in a direct way. The conviction or axiom is not that they are nomically connected, in one of the causal or other ways defined earlier. (1.5) It is that they are nomically connected or else in some other way necessarily connected. One alternative possibility might be said to be that they are logically or conceptually connected, as they would be if there were entailments between mental and neural statements—statements asserting the occurrence of mental or of neural events. Another alternative is that mental and neural events are somehow identical. There is in fact no hope for the first alternative possibility. In fact, then, the axiom of intimacy limits us to (i) psychoneural nomic connection or (ii) psychoneural identity or something like it. We need to choose between them.

The third criterion for a satisfactory account of the psychoneural relation is that of mental indispensability. It derives from a fact on which there is agreement by almost all philosophical parties. Philosophical disagreement is rare, and reluctant. The fact, I think, must be regarded as a pre-theoretical or prephilosophical, something against which theory and philosophy are to be tested rather than a proposition derived from theory or philosophy. It is true, for example, in the •s,     situation that existed, that the mental event of my having seen the olives on the table was somehow indispensable to my wanting one and my reaching for one. The conviction or axiom of mental indispensability is not in itself the proposition that earlier mental events cause events and actions, although such causation will indeed make for indispensability. That the conviction is not identical with the causal proposition follows from the fact that one can have the conviction without having a belief as to causal relations between mental events, or between mental events and actions. Typically, indeed, people have the conviction without also having an articulated view of the mind. Also, the conviction is shared by philosophers who espouse Free Will, and take at least some mental events to be in no causal or nomic connection with anything prior to or simultaneous with them.

The conviction or axiom of mental indispensability comes to this: earlier mental events are essential to any full explanations of certain later mental events and also actions. Earlier mental events are ineliminable parts of any full explanations of many mental events and also actions. Any full answer to the question of why each of many mental events occurred, or of why an action occurred, must mention earlier mental events. There is no full explanation of my reaching for an olive which does not mention my wanting one, no full explanation of my wanting one which does not mention my seeing them on the table. The conviction of mental indispensability defeats any true epiphenomenalism, which is to say any picture of the mind which makes mental events into wholly inefficacious sideeffects or accompaniments of neural events. This nineteenth century picture (Huxley, 1893) is improved upon in certain ways by the several contemporary philosophers who feel themselves driven to it, but is not made tolerable. At the very least, as one allows, it is 'rather paradoxical'. (K. Campbell, 1970, p. Ill)

There is more to be said of the convictions of psychoneural intimacy and mental indispensability—in part, about some neuroscientific doubts of the latter (Ch. 5)—and also of what will be called personal indispensability (3.1), and of pictures of the mind which go against them. Our present imperative business, in which we will take the two convictions as true, which requires no boldness, is that of one set of accounts of the psychoneural relation, and of how they stand to the convictions. Those accounts are Identity Theories other than those put aside a few pages back. Most of these implicitly do involve an adequate conception of consciousness and mental events. They assert that mental events, conceived with a proper realism, are numerically identical with neural events.

The clarifying and discriminating of these Identity Theories requires an understanding of what is to be meant by saying that one thing is identical with another—that a mental event m was identical with a neural event n. To put the question in a form that does not carry the paradoxical suggestion that two things might be one thing, what can be meant by the claim that two descriptions or terms, say 'm' and 'n' pick out or designate one thing? Here and hereafter, by the way, term 'neural event' and like terms are used without any suggestion of simplicity or unity. A neural event may be of many parts and so on.

It is a habit among some philosophers to engage in what is perhaps to be called theorism (P. S. Churchland, 1986, p. 258; E. Nagel, 1979), and hence to avoid giving an actual analysis of their principal claim, that a mental event is identical with a neural event. They do not attend directly to the relation of the phenomena. What is supplied, instead of a clarification of their principal claim, is a certain amount of analogy, sometimes a great deal (P. S. Churchland, 1986, Ch. 7) having to do with scientific theories and the reduction of one to another. The psychoneural identity-claim is to be understood on the analogy of the intertheoretic reduction of heat to total molecular kinetic energy, lightning to electrical discharge, and so on. The analogies are in fact distant from our subjectmatter, and in some respects hopeless. (Cf. Brandt & Kirn, 1967, p. 515, Mackie, 1979, p. 21.) The principal thing to be said, however, is that whatever good or bad analogies exist, the given theories of the mind, if they are to achieve their ends, must identify neural and mental events. What is it to do exactly that?

One familiar and long-running understanding of any identity-claim is that the item which falls under one description is in just the same place, at just the same time, as the other. No two things, it may be thought, can be in precisely the same spatio-temporal position. This criterion of identity-claims seems unacceptable. There is no doubt it gives a logically necessary condition of identity, and no doubt that we make use of it much of the time. But it is at least uncertain that identity-claims can be understood in this way—that spatio-temporal coincidence gives logically necessary and sufficient conditions of identity. The spatio-temporal coincidence of what falls under two descriptions seems to be consistent with there being two things in question rather than one. As has been objected, two physical processes which are in the same place and time may indeed be two processes rather than one. (Davidson, 1980c) Consider the rotation and the warmingup of an iron ball.

The mentioned objector has in the past offered another account of identity for events: that what seemingly are two events are one if the two have the same causes and effects. (Cf. Steiner, 1986, p. 250.) Again it is to be allowed we make use of the criterion, but it must seem at least doubtful as the whole truth about identity. One reason is that it is •",   at least a conceptual possibility that seemingly two things have the same causes and effects but are not spatio-temporally coincident. Thus they cannot be identical. Secondly, it is a conceptual possibility that c be an event without causes and also without effects. What would seem to be two such events could not, in terms of the given account of identity, really be two. But surely that must be left as conceivable.

The acceptable and most established understanding of identity-claims, quite as longrunning as spatio-temporal coincidence, lies behind both our inclination towards that criterion, and our opposed willingness to say that what falls under one description may be spatio-temporally coincident with what falls under another description and yet there may be two things in question. In this acceptable understanding, to say that what falls under one description is identical with what falls under another is to say that what falls under the first has true of it all and only the propositions that are true of what falls under the second. What may seem to be two individual properties, or two collections of individual properties, is in fact one property or propertycollection if any truth about the first is a truth about the second, and vice versa. What may seem to be two things is in fact one thing if the seemingly two things share all truths. This Leibnizian understanding of identity-claims, in terms of what is called the Indiscernibility of Identicals, although it certainly raises problems, is surely the correct one. (Ayer, 1954b, 1984b; Wiggins, 1980, Ch. 1; Munitz, 1971)

2.3.1 Dualistic Identity Theory With Neural Causation

The first realistic Identity Theory we shall consider combines the assertion that mental events and neural events are identical with the claim, as it is sometimes made, that it is neural events that have certain causal roles. Let us have the mental event m, my seeing the olives, and the simultaneous neural event n, and also the later mental event m', my wanting an olive, and the simultaneous neural event n'. The theory asserts that m was identical with n, and m' was identical iwith n'. Further, with respect to any causal relations, they hold between n and n'. The earlier event was cause or causal circumstance of the second.

That is to say of the earlier items, given our understanding of identity-claims, that that which falls under 'm' shares all truths with that which falls under 'n'. A fundamental and critical question arises, one which philosophers have not given the attention it requires, and whose answer has large consequences. The question, in our current terms, is this: What are the truths about that which falls under 'm'? There are two possible answers to this question. To give one of them is to give further and essential definition to the Identity Theory in question. (Giving the other answer takes us to another theory, or rather several of them, to be considered below.) Let us say here that the truths about that which falls under 'm' are of two kinds. There are truths, first, which pertain to consciousness or mentality. They have to do, that is, with the interdependent existence of a subject and contem There are truths, secondly, of a neural kind. That is, to use the summary description mentioned earlier, they are truths about the electrochemical. The truths which are shared by that which falls under the first term and that which falls under the second term are of these two kinds.

What this must come to is that m has both a mental part or property or character or whatever, and also a neural part, property, character, or whatever. The assertion of identity, thus, is just to the effect that there is a single entity, which entity has both a mental and a neural part or whatever. To claim m is identical with n is to say there is one thing with the two sorts of part or whatever. We shall say the same, of course, of m' and n', having in consistency given the same answer to the question about the nature of the truths involved. What is to be added to this, as remarked above, is that any causality involved is a matter of the neural. What is to be added, as it is often expressed, is as follows: it is the first event as neural or under its description as neural that has a causal role. On the assumption that the first event caused the second, it is m (or as we can as well say, n) as neural or under its description as neural which caused m' (or as we can as well say, n'). What this comes to, as it must, is that it is the neural part of m and not the mental part of m which is a cause of the later event.

I suspect this Identity Theory has as many supporters as any, partly for the reason that it involves mental realism, which is in fact the most common view of consciousness. It is partly misleading to speak of it as an Identity Theory or a Monism, of course, although it has to do with a single entity, since it involves two categories of parts or properties. To label it, taking into account its causal part, it is the Dualistic Identity Theory With Neural Causation. Figure 2 models it. What we can refer to either as m or n has the mental part or property indicated by the Greek letter psi and the neural part or property below indicated by the Greek letter phi. Here and hereafter, it is to be noted, 'phi' and the like will designate the neural. They will not be used more generally, as is common, for the physical. As for causal connection in the model, it is of course represented by — C —>.

The Dualistic Identity Theory With Neural Causation may be taken, too quickly, to satisfy the conviction of psychoneural intimacy. This is so since it does speak of the mental and the neural in terms of one thing. Without further reflection here on whether the theory satisfies that conviction, let us consider whether the theory satisfies the conviction of mental indispensability. It attempts to do so by way of its identity-claim. To repeat, the theory gives a certain definition to the claim that m was identical with n. It is that there is one thing which has both mental and neural properties. Certainly we can say this as readily as we can take what I now hold in my hand to be one thing, a pen with its new nib, rather than two things, pen and nib. Again, what is on the table can be regarded as one thing, rather than two things (a book and its jacket), or 403 things (jacket, binding, and 401 pages). In each case, it is a matter of choosing a classification of things, and thus a principle of counting. We may not come to discover that the world is different from what we thought by choosing a different classification, but we can indeed classify as we want, with lesser or greater profit.

It is essential to see that the given proposition of identity about m and n does not give us the conclusion that all the parts or properties of the single entity have the same standing. The fact that psi and phi are parts of a single entity does not give us the conclusion that psi's relation to other things is the same as phi's relation to those things, however we may speak. (Cf. 1.1.) By analogy, the fact that pen and nib are parts of one thing does not give us the conclusion that the pen stands to the depositing of the ink on paper in just the way that the nib stands to it. In particular, to come to the fundamental point, we do not have the conclusion that since psi and phi are parts of one thing, and since phi is causal with respect to the later item, psi is also causal. The given theory attempts to secure the indispensability of the mental through causal connection, which is certainly sensible, but in fact it fails to do so. We cannot get psi as causal and therefore indispensable by speaking of one thing of which another part, phi, is causal.

By way of another analogy, consider a yellow pear, a half-pound in weight, which is put on a scale, whose pointer then goes to the half-pound mark. The explanation of the latter event does not necessarily include the colour of the pear. Its yellowness does not become part of the explanation of the pointerevent in virtue of being a property of a thing which also has the property of being a half-pound in weight.

We are driven to conclude that this Identity Theory, if it does have the recommendation of sticking to the subject of the psychoneural relation, and might be supposed to satisfy the requirement having to do with psychoneural intimacy, does not satisfy the requirement of mental indispensability. It is therefore untenable. Further familiar objections have been attempted against this and other Identity Theories, objections derived from a theory of identity statements as a species of necessary truths. (Kripke, 1971, 1980; cf. Ayer, 1984b) We can rest content with the objection from mental indispensability, which has what it is reasonable to call theoryindependent strength.

2.3.2 A Variant of Anomalous Monism

There is a related view of the mind which merits special attention, the doctrine of Anomalous Monism. (Davidson, 1980e) Part of its interest is that it incorporates a particular argument for identifying mental and neural events. In setting out a variant of Anomalous Monism, I shall keep to the conception of the mental which we have, take the mental to be part of the physical, in general use the terminology with which we are familiar, and also pass by certain matters without discussion.

The view emerges from reflection on what seems to be a contradiction produced by three propositions. The first, called the principle of causal interaction between mental and otherwise physical events, is that there are causal connections between mental and otherwise physical events. It is the proposition that mental events cause events in brains and nervous systems, and also ordinary physical events external to individuals, such as the movement of olives, and that neural and other external physical events also cause mental events. The second proposition is the principle of the nomological character of causality: wherever there are causal connections between events, there are also nomic connections between them. This, certainly, can be taken as in accord with our own findings in the previous chapter. It is more plainly expressed as the truth that causal connection is a species of nomic connection. The third proposition, more adventurous, is in part that there are no psychoneural nomic connections. Indeed, there are no nomic connections between mental events and any other physical events. Arguments for this principle of the anomalism of the mental will be considered later. (2.7)

We escape the seeming contradiction in the three propositions and arrive at a certain view of the mind by way of a certain understanding of the second proposition, that causal connection brings in nomic connection. Given this understanding, we can hold all three propositions together. The understanding depends on the idea that a thing can be identified by one of its properties, and be said to cause a second thing, but it need not be in nomic connection with the second thing in virtue of the mentioned property. More particularly, a thing picked out by certain of its individual properties can truly be said to cause another, and be in nomic connection with it in virtue of other individual properties. When I put the yellow pear on the scale, it is not false to say that something yellow caused the scale's pointer to move to the half-pound mark. However, there is no nomic connection between yellowness and the pointer's so moving. There is no nomic connection between the thing in virtue of its being yellow and the pointer's moving to the half-pound mark.

It is not logically necessary, then, if a mental event causes another physical event, that it is in this nomic connection in virtue of its being a mental event. It is not logically necessary that the second thing is caused by the first in virtue of the mental or conscious part or property of the first. Furthermore, the third of our principles denies that there are any such nomic connections.

The upshot is clear enough. To state it in one way, the mental event must be identical with some other physical event, a neural event. There is one thing which is both a mental and a neural event. We can at this point ask the critical question posed earlier about the Dualistic Identity Theory With Neural Causation. What are the truths shared by the mental event and the neural event? The answer, as before, is that they are both mental and neural truths. We also have another result, essentially of the third principle. It is that there is no nomic connection between the mental part and the neural part of our single thing. We thus arrive at a particular Identity Theory. It differs from what we have lately considered, the Dualistic Identity Theory With Neural Causation, in one way. As shown in Figure 3, it does not contain the idea that it is the neural part of an entity, rather than its mental part, which causes subsequent items, notably later entities of the same sort. It also asserts (by '—— Nom ——' in Figure 3) what was left unassorted in the previous view, the absence of nomic connection between the mental and the neural part of our single thing.

A question longs to be asked. Certainly we may speak of the whole and mean the part in talking of things in causal and nomic connection. Or, we may speak of something by way of certain of its properties, speak of it as causing something else, and it may not be in causal and nomic connection with the second thing in virtue of the mentioned properties. Suppose, however, that we speak more carefully, as we must in order to be clear, and specify just the individual properties of the first thing that are relevant to its causing the second thing. Does that causal connection bring in nomic connection between the very same items? Evidently it does. The only causal connections that there are, strictly speaking, are nomic connections. To express the point differently, if it is specifically the weight of the pear that made the pointer move to the half-pound mark, then the weight of the pear was in nomic connection with the pointer's movement.

Is there a difficulty in this idea that it is in virtue of certain of its properties rather than others that an event is the cause it is? Well, it may be said that the event of the pear's being put on the scale would not have been the event it was if the pear was not yellow. Thus there would be a barrier to saying that that event would have caused the pointer's movement to the half-pound mark if the pear had not been yellow. That event would not have occurred. Does that make the pear's being yellow causally relevant to the given effect?

It is clear that it does not. Certain conditional connections hold between the weight of the pear and the pointer's movement. They do not hold between the yellowness of the pear and the pointer's movement. The yellowness was logically necessary to the event's being the event it was, but not logically necessary to the event's being in a certain sense the cause it was. Certainly it may be said that the cause there was, in a loose sense of that phrase, would not have existed if the pear had not been yellow. That is consistent with the yellowness being causally irrelevant to the effect. That we can say, as we do, that the cause that there was would not have existed if the pear had not been yellow is owed to the fact of language mentioned above—roughly the fact that we take the whole for the part—and not to any fact of causal necessity about all properties of the pear. There is no such fact.

We can then rightly identify the causally relevant property or properties of a thing, and the resulting causal connection is in fact a nomic connection. To return to our variant of Anomalous Monism, we can and must ask what are the causally relevant properties of an entity of the kind we have, one that is both a mental event and a neural event. To avoid the question would be to fail to have a determinate picture of the mind. There are three possible answers: it is the mental part of the entity that is causally relevant, or the neural part, or both parts. Not much needs to be said of any of these three possibilities.          

(i) If it is the mental part that causes another physical event, then we have nomic connection between the mental and the otherwise physical. That contradicts the third of the three principles with which we began. Causal connection of the specified kind gives us nomic connection, which connection is denied by the third principle.

(ii) If both parts of the entity are said to cause another physical event, then it remains true that the mental part of the entity causes the later physical event. We have just the same upshot of contradiction.

(iii) If we take the view that the neural part of the entity causes another physical event, there is a different embarrassment. It is of course exactly the one we have encountered with the preceding theory—the Dualistic Identity Theory With Neural Causation. We do not get the indispensability of the mental. That the neural part of the entity has a certain causal role does not give us the upshot that the mental part is an ineliminable part of any explanation of a later thing. In Figure 3, we do not have it that is indispensable with respect to the later event.

We cannot but conclude that this account of the psychoneural relation fails. The initial three propositions, together with the fact of causally relevant properties and the fact that causal connection is nomic connection, issue either in contradiction or in a denial of .         mental indispensability. Essentially the same argument can be directed against the original doctrine of Anomalous Monism. (Honderich, 1982, 1983, 1984; Smith, 1982, 1984; cf. Stoutland, 1980; Macdonald & Macdonald, 1986; Melchert, 1986; Ayer, 1984c, pp. 3-4)

2.3.3 Two-Level Identity Theory

Another view of the mind presented as an Identity Theory also merits special attention, not only because of its emphasis on the reality of mental events. (Searle, 1983, Ch. 10, 1984, Lect. 1) Again, partly since we shall persist with exactly our own conception of mental events, what follows here is properly regarded as a variant of the original doctrine. One fundamental element of the Two-Level Identity Theory, to put it in one way, is that the events in which we are interested are open to two levels of description. At one level they are to be described as neural events, and at another level as mental events. Or, to speak differently, what we have are phenomena at different levels in the very same underlying stuff. A second element of the view is that the event at the neural level or microlevel causes the event at the mental level. What we may call the neural phenomenon causes the mental phenomenon. By way of analogy to these two elements of the view, consider the individual molecules and molecular behaviour in a bucket of water, and the liquid properties and behaviour of the water. The same stuff, at one level of description, has molecular properties and involves molecular behaviour, and at a higher level of description has liquid properties and involves liquid behaviour—it is wet, it flows, and so on, none of which is true of an individual molecule. Further, the molecular facts cause the liquid facts. A third element of the view is that an event as neural causes later such events, and also actions. My arm's going up is caused by an earlier series of neuronfirings. Indeed, perhaps, it can be said to be caused entirely by the firings. (Searle, 1983, p. 268) It will be thought, at this point, that the view can be diagrammed as in Figure 4.

If that were so, it would evidently offend against the axiom of mental indispensability. However, the first and third elements of the view can be argued to have consequences to which we have not attended. In virtue of the first element, according to which 4» and c)) are the same event at two levels of description, or are phenomena at different levels in the very same underlying stuff, we can add that an event as mental causes a later such event as neural. Further, the first as mental also causes the later as mental, and the first as neural causes the later as mental. Thus, as the proper diagram in Figure 5 indicates, there is the possibility of arguing that we do well and truly cater for the axiom of the indispensability of the mental.

This view evidently gives the same answer as its two predecessors to the critical question about what the shared truths are when a mental event is said to be identical with a neural event. The truths about that which falls under 'mental event' include both mental and neural truths. However, the present view is not well described, or at any rate not fully described, as conveying the idea of an entity of two parts or properties or characters. The essential idea conveyed by the claim of identity, that a mental event is a neural event, is that the mental event is of the same stuff as the neural event. Figures 4 and 5, then, must be read differently than their predecessors. For example, the relation of phi to the whole m/n rectangle is not of the order, or not simply of the order, of part to whole.         

Exactly how, if it is, is mental indispensability achieved? It is not achieved by the second element of the view, that the event as neural causes that event as mental, or that the neural phenomenon causes the mental phenomenon. That is, in terms of Figure 5, mental indispensability is not achieved by way of the proposition that phi causes psi. The causation by psi of the later psi' and phi', if considered only in terms of the causation of psi by phi, would not give us the indispensability of psi. This is so for the reason that phi by itself directly causes psi, and, further, phi is at least part of a causal circumstance for phi' which does not include psi.

Mental indispensability, if it is achieved, is achieved by the first element of the view, which, again to put it as plainly as it can be put, is that psi is of that stuff of which it is true that phi is also of it. (It would make little difference, perhaps, to speak of phi as being the stuff of psi.) The essential question is therefore this: If stuff is causal, does it follow that what is realized in it is in the same way causal? Consider the pear again. Is the stuff of the pear causal with respect to the movement of the scale's pointer? It is hard to resist the answer yes. Is the shape of the pear, which is realized in the stuff, causal in the same way? The answer is certainly no. A reply to this argument would depend on a good deal of notably difficult metaphysical or ontological clarification, having to do with the essential ideas of stuff, levels, phenomena, and so on. This clarification is not provided. If, in clarification of the two levels, we make use of the idea of two sets of individual properties, it is clear we shall not get shared causality and hence mental indispensability.

One's scepticism as to the causal indispensability of psi, despite its being realized in the same stuff as phi, is not reduced by the analogy with the bucket of water or by like analogies, which are certainly weak. One's scepticism as to the causal indispensability of psi, despite its being realized in the same stuff as phi, is also not reduced by the persistent assertion that in psi and phi we do not have 'two ontological categories', 'two mutually exclusive classes of things', 'mental things and physical things'. (Searle, 1983, p. 271) Against these unspecific assertions, which call out for further analysis, there is the ruinous fact of the second element of the view: that phi causes psi. For any causal relation or connection, as is undeniable, two of something are required. Thus the monistic character of this view of the mind, to speak generally, is sharply reduced, or indeed reduced to a mere appearance, by the dualism brought into sharp focus by the given causal connection.

The Two-Level Identity Theory is a strong attempt to clarify the psychoneural relation. It is among several views closest to the one to which we shall come. None the less, it is difficult or impossible to escape the conclusion that like its predecessors it fails to secure mental indispensability and so is unacceptable.

2.3.4 Local Idealism

There are two Identity Theories which unquestionably deserve the name. They attend to our conviction of the indispensability of the mental in different ways. The first derives from giving a different answer to a question of which we know. (p. 94) We are told here as elsewhere that a mental event is identical with a neural event. This must mean that what falls under the term 'mental event' shares all truths with that which falls under 'neural event'. In answer to the question of what the truths are about that which falls under 'mental event', the answer given until now has been that they are both mental and neural truths. Suppose the answer given is instead that all of them are mental truths. They have to do exclusively with interdependent subject and content.

There is a disastrous upshot. What follows from this answer, together with the assertion of Leibnizian identity, is that the truths about that which falls under the term 'neural event' are wholly mental. What we took to be neural, to be a matter of electrochemical events, is transformed into wholly a matter of interdependent subject and content. The brain is, to speak quite accurately, mentalized. Neuronal events are assigned only the character of thoughts, desires, and the like, as we ordinarily understand them.

What we have is what can be called Local Idealism. In the tradition of philosophical Idealism, associated with Hegel and Berkeley, all that exists, including all of the physical, is taken to be of some mental character. In the place of this global Idealism, what we have in this Identity Theory is the proposition that a selected part of the neural world is turned mental. Neural events are turned mental. (Figure 6) It is a bizarre view, considerably more bizarre than Idealism itself, and cannot detain us. That it secures the indispensability of the mental, as indeed it does, cannot begin to give it a recommendation.

2.3.5 Eliminative Materialism

Mental events are identical with neural events, we are again told. If a question can be asked about the truths with respect to mental events, so a question can be asked about the truths with respect to neural events. One answer, that they are both neural and mental, would give us, if we added a proposition about neural causality, just the first view considered above, the Dualistic Identity Theory with Neural Causation. Suppose, however, as is more natural, that we answer that the truths about that which falls under the term 'neural event' are all of them electrochemical truths. (Figure 7)

Again there is a disastrous upshot. Given Leibnizian identity, we have the conclusion that mental events are wholly neural in character. What we have, plainly, is a denial of the existence of consciousness and of mental events as ordinarily and realistically understood. Thoughts and feelings, as ordinarily conceived, are neuralized. They become a matter of neural activity. They are, as ordinarily understood, eliminated.

The proposition is, so to speak, psychologically less impossible to affirm, if no particular conception of the mental is employed in one's reflections. That, indeed, seems sometimes to be exactly the case. For example, in considering whether the mental and the neural are identical, which question may already be obscured by reflection on intertheoretic reduction rather than the actual relation of identity, we may in effect be invited to operate with the conception of the mental in 'folk psychology', which conception is in fact not supplied. We have no conception in the relevant sense, no account of the nature of the mental, when we are in effect directed to the conception of the mental in 'that roughhewn set of concepts, generalizations and rules of thumb we all standardly use in explaining and predicting human behaviour'— preeminently the concepts of belief and desire. (P. S. Churchland, 1986, p. 229 f.) Every conception of the mental—Brentano to Functionalism—is a conception of that which includes beliefs and desires in a standard sense, and none is given relevant content by that fact.

Hardly any philosophers have actually embraced what we can call Eliminative Materialism, although some have been inclined to it and others have contemplated it. (P. S. Churchland, 1986; P. M. Churchland, 1981, 1984; Feyerabend, 1963; Rorty, 1970) Philosophers described as materialist have generally not gone so far as to exclude the existence of consciousness from the universe. Any such exclusion, it must surely seem, however much it accords with certain programmes and credos in the philosophy of science, is no less bizarre than the denial of neural reality in Local Idealism.

2.3.6 Semi-Eliminative Materialism

Let us finish by looking at what can be regarded as departures from Eliminative Materialism. They are in fact departures owed to the difficulty or impossibility of denying the existence of consciousness realistically conceived. It is said that neural events are identical with mental events, and we are given to understand that the truths in question are wholly neural. However, various additions are made at this point, all of them following the same pattern. (Cf. Wilson, 1980.)

There exists only something neural with respect to m/n in Figure 8, but, it may be said, there is the further point that the owner of m/n, the person in question, has it from the inside. There is the neural m/n, but there is also the special presentation of it, which is in some way private to the person in question. There is this special manifestation of the event. Or, it may be said, we must take into account a certain general distinction, between what is called the evidence and the evidenced. The event m/n, which is neural, is that which is evidenced by certain evidence. (Cf. P. M. Churchland, 1984, pp. 334.) Or, it may be said, finally, the assertion that m is neural is to have added to it a familiar doctrine owed to Frege. It is that certain fundamental terms in language have both reference and sense. That is, such a term picks out a given referent, and the term also has a sense. In the standard example, the terms 'the Morning Star' and 'the Evening Star' have the same referent, a certain heavenly body, but there is the difference between them that they have different sense. The term 'm' has a referent which is wholly neural, but, we are to understand, there is also the matter of the term's sense.

The third addition is perhaps least arguable, since it simply is not true, as a fact of language, that such mental terms as 'feeling sad' or 'thinking about Oxford' or 'm' are understood to have neural referents. However, putting aside any claim as to standard usage, we can recall that Frege defines the sense of a term as the mode of presentation of the referent. (1960 (1892), p. 57) This third addition could be taken as simply the claim that for certain neural events there is a special mode of presentation. The third addition is thus tantamount to one or both ) of the previous two.

There seems an inescapable contradiction about Identity Theories of this kind if they do add anything significant to Eliminative Materialism. An attempt is made to avoid it, certainly, but it is exceedingly difficult to escape the idea that the existence of consciousness is both denied and asserted. With the idea of evidence, as with those of a special manifestation or presentation [e in Figure 8), we appear to have consciousness reintroduced after its existence has been denied. The fact of mentality, not to be got rid of, is allowed to persist in a certain way. However, what can be said briefly is that these theories fail for the reason that they fail to secure the indispensability of the mental. To have a chance of securing it, of course, they must indeed    . reintroduce the mental by way of the idea of a special manifestation or presentation. More is needed, however. The relation between a neural event n and the private manifestation or presentation of that event is obscure. So with the relation spoken of in terms of the evidenced and the evidence. What seems clear, however, is that it does not give us indispensability. The neural event, as in Figure 8, may be taken to be a cause of a later neural event. There is no reason to think, however, that the evidence or manifestation of the first event stands to the second event in the same relation as the first event itself stands to the second event. We have no reason to take it as part of any explanation of the second event and its manifestation. This is, in fact, a theory of the psychoneural relation which amounts to traditional epiphenomenalism.

This survey of attempts to identify the mental and the neural strongly suggests a general and I think hitherto unnoticed truth about all Identity Theories, and not only those surveyed. All Identity Theories must answer the critical question. Inevitably it must be answered in such a way as to produce either a kind of dualism or a true monism. In the former case, mental indispensability cannot be achieved by what is on hand. In the latter case, the upshot is Local Idealism or Eliminative Materialism, neither near to being tolerable. Identity Theories, in sum, face a defeating dilemma.


Our progress so far has taken us through the subjects of nomic connection and in particular nomic correlation, the nature of consciousness or mentality, some of the criteria to be satisfied by a defensible account of the psychoneural relation, and the failure of the account of that relation given by Identity Theories. Thus we can now move quickly—to a conclusion as to part of what is to be put in place of Identity Theories. The conclusion is tentative, since it depends also on what is to come, a consideration of other pictures of the mind which include accounts of the psychoneural relation. Some (3.2) are akin to Identity Theories and some—the Indeterminist Theories (3.53.3)—wholly different. These are better considered in connection with the second hypothesis of the theory of determinism of this book.

The Hypothesis of Psychoneural Nomic Correlation, which as a result of our progress so far will need a good deal less clarifying than defending, is to the effect that each of our thoughts and feelings, indeed every item in our ongoing conscious experience, is in nomic connection of a certain kind with a simultaneous neural event. The hypothesis applies not only to wonderings, believings, inclinations, desires and moods, but also to guessings, judgings, resolvings, intendings, choosings and decidings. The first group of examples is of mental events that are in some sense passive and the second of mental events in some sense active. Whatever this distinction comes to, the Correlation Hypothesis makes no improbable distinction between them in terms of nomic connection, as used to be and sometimes still is attempted. (Johnson, 1940, Pt. 3, Ch. 7, s. 6; cf.Broad, 1925, p. 121 f.; Popper1982b, pp. 155-6)

Something popular in one school of psychology (Kohler, 1966; Wertheimer, 1966) and sometimes still proposed, 'psycho-physiological isomorphism', suggests this hypothesis: every type of neural event simultaneously necessitates and is necessitated by a single type of mental event. Hence, of course, each is necessary to the other. An hypothesis of this kind would make mental and neural events into nomic correlates of the third of the three kinds specified earlier: simultaneous mutual necessitators. (1.5) Such an hypothesis, if it is superior to Identity Theories, has a fairly familiar disadvantage. To state it quickly, there is at the very least a possibility that different types of neural events go with one and the same type of mental event. This was concluded early in the modern history of neurophysiology partly because of the surprising results of certain injuries to the human brain.

As we shall understand it, then, the Hypothesis of Psychoneural Nomic Correlation has to do with sets of neural events. The hypothesis is this:

For each mental event of a given type there exists some simultaneous neural event of one of a certain set of types. The existence of the neural event necessitates the existence of the mental event, the mental event thus being necessary to the neural event. Any other neural event of any of the mentioned set of types will stand in the same relations to another mental event of the given type.

Another formulation is somewhat less abbreviated.

Each mental event of a given type is such that there occurs simultaneously some neural event of one of a certain set of types. Since the neural event occurred so did the mental event, and would still even if any other events logically consistent with both events had also occurred. If the mental event had not occurred, neither would the neural event, even if any other events logically consistent with the absence of both had also occurred. Any other neural event of the mentioned set of types will stand in the same relations to another mental event of the given type.

This makes neural event and mental event into lawlike correlates of the first of the three types distinguished. (1.5) What we have, then, more fully, are neural and mental events that are simultaneous; the neural necessitating the mental, the mental thus being necessary to the neural; the mental dependently necessitating the neural, the neural thus being dependently necessary to the mental, (p. 75) We here have a clear and explicit account of the psychoneural relation, owed to our inquiry into nomic connection (Ch. 1) as distinct from accounts in terms of unexplained 'mapping' and the like. (Thorp, 1980, Ch. 3)

Consider, by way of example, my wondering a moment ago if the chair is level—consider, that is, the mental event which I would so describe. The hypothesis has the consequence that there also occurred a certain neural process which necessitated the mental event, and that the mental event was necessary to the process. Also, according to the hypothesis, should there ever occur another neural process of the same type, in me or anyone else, there would also occur a mental event of the given type. Should there ever occur another mental event of the given type—which is to say a mental event indistinguishable in content from the first—it would be accompanied by some neural event of a certain set of types but not necessarily the type of the first neural event. If in the future there never occurs a neural event of one of the set of types, neither will there ever occur another mental event of the given type. The psychoneural relation, then, to use a familiar terminology, is not a oneone or a manymany but rather a onemany relation.

The Correlation Hypothesis thus asserts neural and mental events to be in one of a certain family of relations well established in science. As remarked earlier (pp. 14, 47, 53), these relations are spoken of in terms of functional interdependence, concomitant variation, uniformities of coexistence, and so on, and given mathematical expression. (Cf. Mackie, 1974, Ch. 6; Bunge, 1959.) There would be little to be gained from attempting to find close analogies within physics, chemistry, or whatever. There is little point of thinking, exactly, of the gas laws or electromagnetism or the orbits of a double star. This is so partly since the problem of the nature or explanation of psychoneural nomic correlation—of which something more will be said below—remains insoluble.

It is essential to the argument to come not to confuse the Correlation Hypothesis with anything else. It is unlikely to be confused with one particular thesis of mental supervenience. Here it is asserted that identical neural events, if accompanied by mental events, are accompanied by identical mental events, and that there is no mental change without a neural change—but that these relations are not a matter of nomic connection but rather, to speak plainly, accidental connection. (Cf. Davidson, 1980e; Honderich, 1981a, p. 302, note 19.) More is to be said of this. (p. 141 ff.) The Correlation Hypothesis is more likely to be confused with one of several other theses of mental supervenience. These describe mental events as supervenient on neural events in the sense that mental events are determined by or dependent on neural events. One thesis given this general characterization is that identical neural events, as a matter of nomic connection somehow conceived, are accompanied by identical mental events. Another such thesis, somewhat stronger, is that for each mental event, as a matter of nomic connection somehow conceived, there exists a neural event such that any event of that type is accompanied by a mental event of the type of the given mental event. (Kim, 1979, 1982, 1984; P. F. Strawson, 1985, pp. 5568) „         

The most important of several differences between the Correlation Hypothesis and such theses of supervenience, which are of course related to it, has to do with the general characterization of the latter, however well that characterization may be deserved. The Correlation Hypothesis cannot in any ordinary way be said to represent the mental as determined by or dependent on the neural. What it in itself asserts is no more than precisely the specified nomic connections. It will be as well to recall them specifically, in a different order from before. What it asserts of a mental event m and a neural event n is as follows.

The events m and n were simultaneous; since n occurred, so did m, and still would have despite certain changes; if or since no neural events of certain other types occurred, if n had not occurred, neither would m, despite certain changes; since m occurred, and certain neural events other than n did not, n occurred, and still would have despite certain changes; if m had not occurred, neither would n, despite certain changes.

Clearly enough, these claims would be misleadingly summarized by saying that the mental is determined by or depends on the neural. For one thing, there is involved no causal connection, in the fairly standard sense defined earlier. (1.5) In particular, m is a simultaneous necessary condition of n. Secondly, any proper talk of n's determining m will have to do most importantly on something else: causation by prior things of n, or m, or both. For example, if we were to add the proposition that m was to be explained as the effect of a prior circumstance, and that n was not an effect, that would be crucial. It would make it proper to say that the neural was determined by or depended on the mental.

The matter of such causal connections is of course one to which we shall come. (Ch. 3) It is only when we have considered them, in connection with the second hypothesis, that we shall have a full account of mind and brain. More precisely, what will be called the Theory of Psychoneural Union, or the Union Theory (pp. 165 ff.), will provide a full characterization of what we have been concerned with so far, the fundamental psychoneural relation itself. According to the Union Theory, the principal fact about the psychoneural relation is nomic connection, but it is not the only fact—a mental and a neural event also constitute what will be called a causal pair. Our coming reflections will also involve views that may have come to mind already—for example that the relation between mental and neural events consists in no more than the causation by neural events of mental events, the latter necessarily being later in time, and the causation by some mental events of neural events, the latter being later in time. (pp. 151 ff.)

Taking into account only what we now have, the Correlation Hypothesis, it is indeed clear that what is to be said in a word or two of it is that it makes mental and neural events into nomic correlates rather than items in a determinationrelation or a dependencyrelation. Similarly, the hypothesis cannot with reason be called reductionist. The term is used variously and loosely. Reductionist hypotheses, by one definition, deny the existence of mental events realistically conceived. (P. S Churchland, 1986, pp. 278 f.) A reductionist hypothesis, in another sense, is one which explains the mental in terms of the neural, or supplants a mental by a neural theory, or takes the neural to be basic or more fundamental. (Cf. Wilkes, 1978, p. 30.) What the Correlation Hypothesis itself asserts, clearly, cannot unmisleadingly be said to be that the neural is basic or more fundamental, or that one theory is to give way to another. Here, as often enough in philosophy, there is more than fruitlessness in generic labelling.

More importantly, it is unenlightening or worse to describe the Correlation Hypothesis as
dualism, for at least five reasons. It has nothing much to do, first, with what is generally called Cartesian Dualism, which asserts the existence of a mental substance, res cogitans, out of space. It is to be distinguished, too, from what is sometimes called bundledualism, which, as it seems, takes thoughts, desires, and so on to be independent entities, in fact substances, whether or not in space. On the present view of mentality, which informs the Correlation Hypothesis, talk of mental events is strictly speaking talk of individual properties. They are, if it is useful to say so, dependent rather than independent. (1.1) Secondly, as has not been remarked explicitly until now, they are to be taken as individual properties of such structures as the Central Nervous Systems of our species. More plainly, in our case, they are properties of our brains. They are properties, then, of exactly that which also has neural properties. Perhaps it is such a view, incidentally, that has sometimes been described as a dualaspect view (T. Nagel, 1986, pp. 30, 32) but that description has traditionally carried metaphysical implications that are no part of the Correlation Hypothesis. (Schaffer, 1967)

Thirdly, as remarked before, mental events are not excluded from the category of physical events. Nothing said earlier of their fundamental nature, having to do with the interdependent existence of subject and content, excludes them from the physical as defined. They are in space, and causally related to space-occupants with further properties. If dualism is understood as by definition taking mental events to be nonphysical, as generally it is, then the Correlation Hypothesis on this ground alone is not dualistic.

Fourthly, it is commonly said or implied that dualism makes the mental in a certain sense irreducible: it puts the mental outside of the reach of law, which is to say not in nomic connection, or not in nomic connection with the neural. This, it is sometimes said or implied, is really what makes dualism into dualism. (P. M. Churchland, 1984, p. 12; cf. P. S. Churchland, 1986, pp. 323 f.) It is the fundamental thing about dualism. Evidently this irreducibilityclaim is no part of the Correlation Hyothesis. Very definitely, it does not make mental events into somehow novel properties beyond prediction or explanation by science. On the contrary, the Correlation Hypothesis is in the given sense a reducibilityc\siim. 'Dualism, the idea that minds (unlike brains) are composed of stuff that is exempt from the laws of physical nature, is a desperate vision which richly deserves its current disfavour' (Dennett, 1984, p. 28)—that firm sentence, whatever else is to be said of it, has nothing to do with the Correlation Hypothesis.

Fifthly and finally, as became clear earlier, typical Identity Theories or monisms are as dualistic as the Correlation Hypothesis. In fact all arguable Identity Theories, those involving a realist conception of the mental, propound the existence of two categories of properties. They are only misleadingly given the name they have. Given this consideration and the four previous ones, there is little or no point in referring to the Correlation Hypothesis as asserting or presupposing a dualism— as I must own up to having spoken of a predecessor of it in the past. (Honderich, 1981a, b) It is best labelled as asserting just what it does, which is psychoneural nomic correlation.

Something related is to be said of doctrines of what is called mind-brain parallelism. Traditionally conceived, and as the name suggests, these are views of sequences of mental events, and neural and other events, say the two sequences associated with me over the past half-hour, which deny any connection between them. That several touches on my shoulder came together with two identical or similar mental events is not to be explained by any connection between the former pair and the latter pair. Causal connections may or may not hold within each sequence, but none holds between sequences. The views are associated with Leibniz's doctrine of preestablished harmony, and with Malebranche's occasionalism: all events are caused not by other events but directly by God.

It is true that parallelism has traditionally been conceived as opposed to psychoneural causal interraction. (pp. 151, 161) In that, the Correlation Hypothesis is like it. Parallelism is, in its fundamental nature, as much opposed to all psychoneural nomic connection. All such connection raises the problem which parallelism seeks to avoid by its extravagant speculation of no connection. There is thus little to be said for the idea (Addis, 1984; Wilson, 1981, p. 306) that the Correlation Hypothesis is to be regarded as a doctrine of mind-brain parallelism. It is, rather, antiparallelist. (Mackie, 1981, p. 347)

To say a word now about the problem which parallelism seeks to avoid, and which is raised by the Correlation Hypothesis and also the other two hypotheses to come, it is the problem of explaining the nature of the connection between what are obviously two different sorts of events and the like—at bottom two sorts of individual properties, the neural and the mental. In one familiar form, the problem has been stated in terms of the physical laws having to do with the conservation of energy and momentum. On the assumption that mental events are nonphysical, and the assumption that any nomic connections with the neural into which they might be thought to enter would be subject to the laws in question, it appears that they in fact could not enter into any such connections. The issue, which has been considerably disputed (Broad, 1925, pp. 1039; Spicker and Engelhardt, 1976, p. 144 f.; Schaffer, 1968, pp. 667), may be thought not to arise at all for the Correlation Hypothesis, which allows for the physicality as defined of mental events.

Still, whatever is to be said of that, it is very plain we have nothing like an account of the ultimate nature of psychoneural connection, and more particularly of psychoneural nomic connection. What we lack, if the connection is not in a sense primitive, would be supplied by an adequate idea as to a sequence or mechanism joining the mental and the neural. This problem, as fundamental as any, arises for all accounts of the mind which are starters at all, by which I again mean all accounts which conceive adequately of the mental. It arises for the principal Identity Theories we have considered, as it does for epiphenomenalism, for causal pictures and other freer suppositions we shall be considering in connection with the second hypothesis, and also theories of mental supervenience. It is difficult to suppose that it can be escaped even by that kind of supervenience noticed earlier in this section (p. 108) which supposes that there is no nomic connection between the mental and the neural. Also, the problem arises in scientific contexts, where indeed, contrary to the views of some philosophers, an adequate conception of the mental is commonly assumed. Neuroscience itself, in its assumption of ordinary psychoneural connection, must recognize the problem. It is faced directly by any scientific theory or speculation as to the emergence of consciousness in the evolution of species and in the development of an individual from a fertilized cell.

As will be anticipated, it is no argument against the Correlation Hypothesis that it offers no solution. The principal philosophical attempt at a solution which respects the reality of the mental consists in panpsychism, the doctrine that all that exists somehow already includes the mental. If it continues to have resourceful proponents (Sprigge, 1983) and also reluctant friends (T. Nagel, 1979, Ch. 13, 1986, pp. 49 ff.), it is difficult not to side with its critics. (McGinn, 1982) No doubt, if this were an ideal book, more would be said of it.


The Correlation Hypothesis calls up a good deal of questioning, resistance, and objection. We shall consider some of it, mainly since to do so will clarify the hypothesis. One question that will be left unconsidered is that of the relation of the hypothesis to another matter: the explanation of mental and neural events by earlier facts. In looking at Identity Theories we did take into account their entailments or implications for the explanatory question. The like entailments or implications of the Correlation Hypothesis will not be considered here, but in the next chapter.

It is natural enough, as already remarked, to be inclined to accept some proposition about nomic connection between brain and mind. The Correlation Hypothesis, however, may strike some as extreme. Is there not something moderate which satisfies our natural inclination? The idea of such a thing may derive from a selection of small ideas. One is that people generally, because they are physically similar, behave in similar ways—they walk and talk similarly—but, because they are not physically identical, they do not behave identically. A second idea is that a particular individual, given his or her physical constitution, does or would behave in the same ways in identical circumstances. It is not a great step to the General Correlation Hypothesis and the Individual Correlation Hypothesis, both distinct from our Correlation Hypothesis. The Individual Correlation Hypothesis, to give it first, is as follows.

For each mental event of a given type in the life of a given individual, there exists a simultaneous neural event of one of a set of types. The existence of the neural event necessitates the existence of the mental event, the mental event thus being necessary to the neural event. Another neural event of one of the given types will stand in the same way to another mental event of the given type.

My having had what I describe as a mental image of Charlotte Street. was necessitated by one or another of certain neural events, and necessary to it, but, given the Individual Correlation Hypothesis, an identical neural event in you might not be accompanied by that image, or indeed any mental image. The General Correlation Hypothesis, which is not at all tied to individuals, is as follows.

For each mental event of similar types, there exists some simultaneous neural event of one of certain similar types. The existence of the neural event necessitates the existence of the mental event, the mental event being necessary to the existence of the neural. Another neural event of one of the types will stand in the same way to another mental event of the given types.

When the two of us standing under Admiralty Arch have what we report as a view of the Mall, there exist certain neural events. Mine is of one of a certain set s of types, which set is similar to a set s', including the type of which your neural event is an instance.

Anyone who takes up the Individual Correlation Hypothesis, and has the natural inclination to accept some nonaccidental relation between minds and brains, is likely to want the General Correlation Hypothesis as well. How well the inclination is satisfied by the latter hypothesis is wholly uncertain until more is said, however, since the hypothesis as it stands is entirely indeterminate. How similar are the mental events in question, and how similar the two sets of types of neural events?

What might be said to make the Individual Correlation Hypothesis persuasive, or more persuasive? An attempt can be made to explain in a general way how it comes to be true, as perhaps it is taken to be. (Thorp, 1980, pp. 53 f.) At a certain time, young Albert is aware of a bun, and, as it happens, there is occurring the neural event or process n. What happens, we are told, is that his awareness comes into nomic correlation with ntype neural events. Thereafter there is nomic correlation in his case between ntype neural events and a type of mental event pertaining to buns. We can say, metaphorically, that neural events of the given type were blank until a certain time, and thereafter were bunimprinted. Young Bertie, on the other hand, since an ntype neural event occurred in his case at a time when he was aware of his thumb, came to have another type of mental event nomically correlated with the given neural events. In him, n-type neural events are thumb-imprinted.

The Individual Correlation Hypothesis is surely hopeless, and not helped by the imprinting story. Can we suppose that ntype neural events in Albert's case do necessitate buntype mental events? No, since if this were so, the same type of neural events would be accompanied by such mental events in the case of Bertie. It is equally true that no mental event of the given type is necessary to such a neural event, since such neural events turn up in the case of Bertie without the given mental events. These conclusions follow directly, of course, from the analysis of nomic connections to which we have come. (Ch. 1) In fact, there is no remotely arguable analysis of nomic connection which will save the Individual Correlation Hypothesis, or anything relevantly like it.

What may now come to a hopeful mind is roughly this: that ntype events plus something else, x, constitute the nomic correlates of mental events concerning buns, and ntype events plus something different, y, constitute the nomic correlates of mental events concerning thumbs. This is not the Individual Correlation Hypothesis, but something suggested by it and the imprinting speculation. There is an immediate problem. What are x and y? It would be pointless to take them as types of neural events themselves. The result would be tantamount to accepting what we are considering replacing, which is the Correlation Hypothesis. Can x be taken to be just (a) Albert's history, or a bit of his history, or the past fact of imprinting? Can it be (b) the non-neural 'storing' in him of the initial bunawareness? All of this is vague, and the first alternative is surely incomprehensible. It cannot be that the older Albert's mental events of the given type are in fact owed in part to the direct or unmediated influence across time of an event in the past. If nonneural 'storing' is opted for, we have already abandoned any recognizable hypothesis of psychoneural correlation. What we have is a mysterious bundle: the later neural events of the type of n, the nonneural x, and the mental events. To succumb to yet another temptation, to take x as another type of mental event, and to provide a neural correlate for it, would merely complicate matters and be of no help.

It is not only the Individual Correlation Hypothesis, certainly, that pays attention to the plainly true idea that learning and an individual's past environment somehow play a fundamental role in the occurrence of mental and neural events. We shall, as already remarked, be much concerned with the matter in due course. The Correlation Thesis in itself neither entails nor excludes any arguable account of this background. As for the imprinting story, it in fact does no more than raise terrible questions. Was Albert's initial awareness of the bun not or not at first correlated with any neural process? Was it 'freefloating' before the link with ntype neural processes was established? If that is so, to ask but one question, why should all mental events not be freefloating in the same way?

Thus we do not have a more moderate alternative to the Correlation Hypothesis. The failure of the Individual Correlation Hypothesis affects its partner, the General Correlation Thesis. The latter, like the former, needs a partner. Given our natural inclination to require some considerable psychoneural uniformity in the life of a given individual, the general hypothesis needs the companionship of the individual hypothesis. It must do without it. On its own, however, the general hypothesis will not suffice. If we have the common inclination to accept that psychoneural relations are not accidental or coincidental, we cannot regard the General Correlation Hypothesis as sufficient. Or rather, to speak more sensibly, we cannot so regard the various hypotheses of correlation which would fall under it, since it is indeed indeterminate, no more than a schema.

So much for one kind of resistance to the Correlation Hypothesis. Other kinds have to do with its neural side. One possible doubt can be dealt with by the reassurance that the hypothesis is not tied in any way to a discredited or indeed any particular theory of brain localization: a theory which assigns mental events to locales in the brain. It is of course independent of the 'narrow localization' of some 19th Century neurophysiologists, notably Gall and Spurzheim. (1810) Evidently, and consistently with what was said earlier of psychoneural intimacy (2.3), the hypothesis is consistent with different neuroanatomical and other accounts of the brain and also with speculations about it, including various contemporary ones. It is consistent, I think, with very nearly all such theories and speculations, which is to say that vast majority of them which are not specifically devised to preserve indeterminism or the freedom of the will.

The hypothesis does not assert that your brain is very like mine. Or, to say something with more content, it is wholly consistent with facts that have sometimes been produced against psychoneural correlation—for example, that Byron's brain weighed 2,200 grams but Anatole France's only 1,100. (Wilkes, 1980, p. 115) That the hypothesis is consistent with such facts is owed in part to its making the brainmind relation manyone rather than oneone. Also, the hypothesis does not assert that a type of neural event ever recurs exactly, although it properly allows this to be a possibility. (Cf. Wilkes, 1980, p. 114.)

The hypothesis does not consist in a proposal to neuroscientists to investigate the brain or psychoneural correlation in a particular way: guided by the ordinary categories we use in speaking of consciousness—the various emotions and so on. It is sometimes said that there is no reason to think that these ordinary categories have neural counterparts, since the the categories are of a 'practical' rather than a 'theoretical' kind, the categories of 'folk psychology'. Such sceptical opinions are held by very different sorts of philosopher, some committed to what can be called the autonomy of mental life and some committed to neuroscience. (Hampshire, 1972a, pp. 167; P. M. Churchland, 1984, pp. 43 f.) As it seems to me, it is an unlikely idea that our ordinary categorization of consciousness is not in some good relation with neural categories. All that is to be said here, however, is that the Correlation Hypothesis does no more than assert that what is in fact discriminable in consciousness, with the aid of folkpsychological categories or any others, has a neural counterpart. This is no proposal to neuroscience, or a constraint being imposed on it, but, rather, close to being an assumption of it.

To begin now on a larger matter, effectively the subject of the remainder of this chapter, it consists in doubts and objections which have to do with consciousness and with mental events.

2.5.1  Atomizing Consciousness

Does the conception of a mental event carry the idea that consciousness is somehow 'atomistic', that it consists only in 'a sequence of elementary ideas', that one's stream of consciousness consists just in an ordered series of events? It has been supposed that psychophysiological isomorphism and the oneone theory of psychoneural correlation mentioned earlier (p. 107) involves such an atomism, and is therefore mistaken. (Popper and Eccles, 1977, pp. 88 f.) The same objection, whatever its worth, might be made to the Correlation Hypothesis. No doubt any view of the mind which, so to speak, denied relations between mental events would indeed be absurd. Any view would be absurd, that is, which denied the unity or continuity or connectedness of consciousness, at a time or over time. It is plain enough, however, that distinguishing mental events within consciousness in no way commits one to denying its unity. To discriminate is not to deny continuity or connectedness, in which, it may be argued, its unity consists. (Parfit, 1984, Pt. 3) By way of the obvious analogy, a  . stream of water is not made any less a stream for us by the fact that we pick out stages, parts, variations in flow, eddies, and ripples. Nor, of course, are these discriminations in any way illusory or not based in the given reality.

2.5.2 Resemblance of Nomic Correlates

Can it be that the mental is simple, and that the neural is extremely complex, and that this is a barrier to nomic connection, or a barrier to nomic connection on the assumption that such connection is not mediated by a mechanism or sequence? (Cf. K. Campbell, 1970, p. 51.) It is hard to suppose that there is any serviceable generalization on which the objection can depend, to the effect that all nomic correlates are of similar complexity, or of similar complexity when they are unmediated. In any case, the initial assumption that the mental is simple and the neural complex is itself open to objection. Clearly, despite the great complexity of the brain, there are neural events which are in relative terms extremely simple. All of them, further, are of a common character or of common characters. Also, despite the unitary nature of many mental events, and the common character of all consciousness, there is the richness and great informationcontent of, say, ordinary visual experience.

2.5.3 Counting Mental Events

If one thinks of any stretch of one's experience, the question can be asked of how many mental events were in it, or of the number of mental events in which it consisted. Consider just two seconds of my visual experience. A great many answers seem possible. One is that my perception of my hand was one mental event, and my perception of the pen in it a second, and my lesser visual awareness of such things as the Post Office Tower out the window a third. However, it is possible to say, differently, that my whole visual experience during the two seconds was a single twosecond event, or that it was two onesecond events. It is also possible to say a good deal else—for example that the number of mental events was very large indeed—the number of items that logically could have been discriminated by me, the number of items of visual information. An objection that may arise from this is that the idea of a mental event is somehow indeterminate. Do we not in fact have no decent idea of a set of things if we have no settled rule as to counting them, whether or not we are able to act effectively on the rule? We would have no decent idea of a human being, surely, if we had no rule for counting them. Do we, despite all that has been said, have no decent idea of a mental event?

The simple reply to this, albeit one that skirts several problems, is that we do have a rule for counting mental events. It is that anything discriminated in consciousness is to count as one mental event. The rule does not tell us how to discriminate—there are indeed many ways—but that is nothing to the point. There is a large category of adequate counting rules which are in this way indeterminate. There is thus no objection here either to the idea of a mental event or to the Correlation Hypothesis. Consider an analogy of the idea of a mental event and an analogy of the Correlation Hypothesis—the idea of a solid thing and the generalization that all the solid things on my table are inflammable. What of the book, say Wittgenstein's Zettel. Does it with its paper jacket count as one solid thing or do they count as two? It would be unusual but certainly not something that counts as a mistake to take each page as one solid thing. Evidently there are more possibilities. That this is a matter of decision, of a kind, does not make indeterminate either the idea of a solid thing or the generalization that all the solid things on my table are inflammable. The generalization's truth, if it is true, is not affected by how we count the things in question, and neither is its falsehood if it is false. So with mental events and the Correlation Hypothesis. What it comes to is that each thing logically discriminable in consciousness, no matter how the discrimination goes, has a neural correlate.

2.5.4 Linguistic Individuation

Is the claim that we can discriminate mental events put in doubt by something said of them originally, that a single piece of language may describe occurrents of two or more different types? It was said that an ordinary sentence, such as T was thinking that my daughter is a quick thinker', can be used truly of different mental events. Does it follow that an ordinary piece of language cannot be used by me to discriminate or single out a mental event? The matter touches on many others, but the short answer appears to be no. The sentence in and by itself does not discriminate a mental event. It can be used by me, however, in my act of singling out the mental event. (Cf. Mackie, 1981, p. 350.) It is a marker or pointer, a marker of that of which we have fuller knowledge than the marker itself expresses. It is of use to me despite the fact that it is not so specific as to prohibit its use of a related but different mental event. Consciousness is in no way unusual in this respect. I can today use the description 'the velvet tie he is wearing' in my own entirely successful enterprise of picking out a certain material object. I can use the same description tomorrow, of another tie, in picking out that somewhat different thing.

2.5.5 Conceptualization

If I can use the same description for different things that I pick out as having occurred in my consciousness, is it none the less necessary that in order to do so I must in some sense conceptualize them? Must I in some sense put a discriminated thing under an idea, or, more likely, ideas? The question is not clear, but perhaps clear enough. We may feel tempted to reply that we cannot know what we cannot in some sense put under a concept. We may feel tempted to reply that if thereafter an ordinary piece of language may serve as a marker for something we experience, we require in the first instance that it be put under fuller private description, in a superior 'language of thought'. Having been tempted to this, we may then succumb to a certain vague uncertainty. Are our capabilities of private description not very meagre? To take only simple examples, I may be in doubt as to whether my feeling of a moment ago was nostalgia, in the ordinary sense, and have no confidence as to any closer description. I am hardpressed, to say the least, to provide for myself some conceptualization, even an image, of the sensation in my ankle as I sit in this chair. Is there the general upshot that mental events escape us, that we cannot single them out?

Just one of several replies has to do with the first part of this sequence of reflection, the temptation. It is not always necessary that in order to discriminate something within my experience I must conceptualize it at all. It is not necessary, in order to have a hold on it, that I put it under a description of any kind, however my own. Instances of two shades of colour, for which I use no names whatever, are clearly different to me. I can indeed tell the one from the other. I can say of a third colourpatch that it is the same shade as one of the other two. There is a longish history of psychological research concerned with such discrimination—notably research into Just Noticeable Differences in sensation and perception, and also allied phenomena. Very rightly, Fechner's psychophysical law about sensationmagnitudes and stimuli has not been doubted on the grounds that sensations are not conceptualized. (Marks, 1974)

2.5.6 Reidentification

Will it be supposed that there is a particular difficulty about the reidentification of mental events, the assigning of two tokens to one type? It may be thought that this is necessary in connection with the Correlation Hypothesis. Consider a thought of mine a moment ago, that reflection on the nature of the mind is frustrating. Can I be confident, in a few minutes, that a second mental event of the same type, which of course is to say an identical mental event, has occurred? The answer is that it is most unlikely. Memory is not up to the task. On the other hand, it is plain that we can contrive to feel or think precisely the same thing twice. If I touch the top of the table twice in the same way in a very short time, or if I repeat a sentence to myself in a very short time, it may well be that I can confidently judge that the two mental events were indiscriminable.

None the less, it needs to be admitted that our capabilities in this respect are small: they are effective with very little of consciousness. For the most part, indeed almost always save for contrived situations, we cannot confidently suppose that the same type of mental event has obtained again. Nothing much follows from this. In particular, it does not follow that the Correlation Hypothesis cannot be confirmed, as we shall see.

2.5.7 Richness and Transience

Nor is there a threat in what has already been in view, what can be called the richness and transience of consciousness. If I ask myself, when walking in Hampstead High Street, what I have been aware of during even the past few moments, I shall be wholly defeated if I try to produce anything that could be called a complete answer. There is no possibility whatever of my doing so. Such facts, however, do nothing to make the Correlation Hypothesis unclear. That a generalization covers a multitude of items, such that it is impossible to nab each one, in no way makes the generalization unclear. The theory of evolution is not made unclear by the fact that we cannot enumerate the multitude of properties of species which fall under it. So with the law of gravity and indeed innumerable other examples.


Wittgenstein declares in Zettel, entirely in opposition to what has been suggested here, that 'no supposition seems . . . more natural than that there is no process in the brain correlated with associating or thinking . . .' (1967, 608) None of us, of course, is without the convictions that make one view or another seem natural. The question is whether they issue in effective argument. Let us consider four objections offered by or derived from Wittgenstein, and after that two of a somewhat similar kind.

2.6.1 Mechanism                                                 ,

In one place in Zettel he is concerned with the matter of how it comes about that we think and act as we do as a result of a certain visual experience. He mentions the idea of 'a physiological explanation' and writes:

But must there be a physiological explanation here? Why don't we just leave explaining alone?—But you would never talk like that, if you were examining the behaviour of a machine!—Well, who says that a living creature, an animal body, is a machine in this sense? (1967, s. 614)

Wittgenstein is to be taken as speaking the first and second sentences in this dialogue, and the last. It would be unreasonable to suppose that there is not an objection in what he says. One is intimated, at any rate, and it applies as readily to the Correlation Hypothesis as to the idea of physiological explanation. So applied, it is that the hypothesis gives us a falsifying picture of men, a picture of men as being or being like machines. It gives us no more than the doctrine of mechanism.

We need to make certain distinctions in reply, since, as has often been noted, there is more than one doctrine of mechanism, or mechanism socalled. (Gregory, 1981, Ch. 3; Glover, 1970, pp. 28 f.) We need to distinguish at least three kinds of pictures of our existence. If the Correlation Hypothesis were a picture of the first kind, or necessarily a part of one, it would be hopeless. We are all certain, to speak very briefly, that men are not relevantly like solar systems, the reproductive mechanisms of plants, clocks, Turing machines, neurons, or perhaps computers. All of these are at least arguably mechanistic in a certain primary sense: they have characteristic functions or activities, to which they are limited. We are all certain that men by contrast have a different kind or different kinds of control over their existence, that they direct their own activities in a different way or ways, and hence that they do not in the same way have a characteristic function or activity, or even a discrete set of these. This conviction, despite its imprecision and the several kinds of controversy in which it is involved, rightly defeats any picture of men as mechanistic in the given primary sense: being limited in a certain way to a characteristic function. (Cf. Dennett, 1984.)

We must judge that most talk of men as being machines, talk of them as being in some exact and complete analogy with clocks, solar systems, and so on, which are limited to characteristic functions, is fundamentally mistaken. The conviction thus defeats a certain amount of declaration by some neuroscientists. The conviction, or so it seems to me, is unlike others in being not only an attitude but a guide to truth. However, to come to the principal proposition here, the Correlation Hypothesis is not at all reasonably regarded as such a mechanist picture. It says only what it says, that there is psychoneural nomic connection, and in itself it does not fight with the given conviction. Nor, as it will turn out, is it necessarily part of such a picture.

A second mechanist picture is of men as machinelike in that all that they do can be described and explained simply by the principles of mechanics, that part of physical science which deals with the behaviour of matter under the action of force. A different but related proposition is that every biological event, including every human event, is in fact also a nonbiological event. That is, each biological event is subject to nomic connections which are also exemplified by nonbiological systems. There is perhaps slightly more room for uncertainty about the rejection of some mechanist pictures of this kind. What is to be said of them here is just that their falsity, if they are false, is no danger to the Correlation Hypothesis. It does not assert them. Indeed, it may be thought to conflict with them. (Cf. Trusted, 1984,p.129.)

It must be allowed that the Correlation Hypothesis does give us a part of a third and different picture. Again briefly identified, it is a picture of human consciousness and indeed human life as having about it an order or regularity related to that of machines and the natural world. It is an order and regularity of overwhelming variation and complexity, and in this quite different from that of all machines, but an order or regularity none the less. It is owed to no more than nomic connection. What is to be said of this picture, as distinct from its predecessors, is that the fact that the Correlation Hypothesis is part of it is no problem—since this picture is not in conflict with the mentioned truthguiding conviction about our control and direction of our lives, our not having a limited function or activity.

That is, we are rightly certain we are selfdirecting and not subject to certain kinds of limits, in some sense or senses yet to be made clear, but we are not at all convinced in the same proper way that our lives do not have the order and regularity asserted by the Correlation Hypothesis. No doubt there are persons, including philosophers, who have different convictions, convictions which are affronted by no more than the order of the Correlation Hypothesis. These are convictions not in the same way fundamental. In the case of conflict between these convictions and the hypothesis, there can be no prediction of the outcome. It is far from certain that the hypothesis must bow to the attitudes. If it is defeated, its defeat is unlikely to be owed to them.

2.6.2 No Conceptual Necessity

The second objection from Wittgenstein, which is open to briefer reply, consists in certain rhetorical questions. They follow on his proposition quoted above, to the effect that it is a preeminently natural supposition that no process in the brain is correlated with associating or thinking. Why, he asks, should there be a process in the brain correlated with my thinking something or other? Why, so to speak, should there not be 'chaos' instead? Is it not perfectly possible that nothing physiological corresponds to certain psychological phenomena? When I remember something, why must there be something or other in the nervous system which corresponds to my remembering? (1967, ss. 60810)

The intent of the rhetorical questions is uncertain. There are two possibilities as to their interpretation. They may be taken, first, as conveying the idea that the Correlation Hypothesis is not a logical or conceptual truth, that our ordinary notions of thinking, remembering, and the like have in them nothing about neural processes. That the hypothesis is not such a truth is undeniable, but not to the point. That our ordinary ideas of thinking and the like are as claimed is probably also undeniable, but not to the point. It has not been supposed that it would be selfcontradictory or in any related way incoherent to deny psychoneural nomic connection. Nomic connection, as we know from our inquiry into it, (Ch. 1), is certainly not logical or conceptual connection. What we are considering is the question of whether psychoneural nomic connection is an empirical truth, whether as a matter of contingent fact there is such connection.

Do the questions rather convey the absolutely relevant proposition that there is not in fact nomic connection of the given kind? Or, perhaps better, that really there is nothing in life to suggest that there is? These are propositions whose truth or falsehood needs establishing. Certainly falsehood or improbability is not to be discerned on their faces. There can be no possibility whatever of settling by fiat or declaration that the Correlation Hypothesis is true, or well supported. There can be no more possibility of settling by fiat that it is false, or ill supported.

It may be that the first interpretation of the rhetorical questions, that there is not a conceptual connection between the mental and the neural, is the gist of a remarkable remark in another collection.

Thinking in terms of physiological processes is extremely dangerous in connection with the clarification of conceptual problems in psychology.Thinking in physiological hypotheses deludes us sometimes with false difficulties, sometimes with false solutions. The best prophylactic against this is the thought that I don't know at all whether the humans I am acquainted with actually have a nervous system. (1980, Vol. 1, s. 1063)

One might try to make sense of this, that is, by understanding the mentioned 'clarification of conceptual problems in psychology' as being the conceptual analysis of ordinary notions of thinking, remembering and the like. It is charitable to take the last sentence as referring ' to the truism that my ordinary layman's experience of others does not logically entail that they have nervous systems.

2.6.3 Explaining a Prejudice

The third objection begins by offering an explanation of what is called the 'prejudice' in favour of such views as the Correlation Hypothesis. (Wittgenstein, 1967, s. 611, cf. 1959, ss. 3048, s. 412) We are said to be inclined to suppose that thinking and feeling is 'a process in the head', some 'occult' stream in which earlier parts somehow give rise to later. If we try to think of this as a plain matter of cause and effect, we become puzzled, since we are taking ourselves to be dealing with 'gaseous' stuff. At this point we are inclined to try another path in order to account for the seeming process or stream. We accept the existence of correlations between mental and neural events, and we add \*hat has not been mentioned so far, the idea that there are ordinary causal relations between neural events. These ordinary connections underlie and help explain the seeming connection between mental events.

What, according to Wittgenstein, is to be said against this course of reflection issuing in a 'prejudice', is that it begins with a 'primitive' idea. It is 'primitive' to take thinking and feeling as 'a process in the head'. They have to do, rather, with behaviour. We thus need not ever start on the given course of reflection which issues in the Correlation Hypothesis and like things.

If we abstract from the rhetoric, what we have in these remarks first of all is a behaviourist account of consciousness. It is an account which remains in ways obscure in Wittgenstein's work, but certainly goes against any ordinary account of consciousness and also the account in terms of interdependent subject and content. As will be anticipated, given what has been said of behaviourism (p. 73), there seems to me no compelling reason to reopen that subject. Secondly, there is in the given remarks the idea that if we do suppose that consciousness is what can be called ongoing experience, we do not find it easy to accept that there are causal connections holding it together. This is more persuasive than the behaviourism. It has to do, in part, with the admitted difficulty we have in getting consciousness into focus, and also something hitherto unmentioned, the evident conceptual connections within it. However, to come to a third crux, there is no reason to think that this is what leads us, in the way outlined or any other way, to the Correlation Hypothesis. We do not come to it by way of a frustrated search within consciousness for what holds consciousness together. That story cannot count as serious. Furthermore, it would not much matter if the story were true, or if it had somehow influenced us without our explicit awareness. What does matter is whether the hypothesis is conceptually adequate and true. No light is shed on either question, the first of which is our question at the moment, by the speculation as to the origin of the hypothesis. Wellformed truths can have disreputable pasts.

2.6.4 The African Lad

The fourth and last Wittgensteinian objection, given shape by another philosopher, will appear to be more telling. (Anscombe, 1972; cf. Wittgenstein, 1967, s. 608 ff.) It needs more attention, as do several related contentions. It depends on the fact that if the Correlation Hypothesis is true, then, in virtue of one part of it, if a certain neural process occurs, someone has a certain belief—let us say he believes at a certain moment that the art galleries in Paris close at 7 p.m. He has a belief which he might so express. It is not near to being true, of course, that it is within the capabilities of neurophysiologists or others to bring about artificially a neural process of the given type, perhaps by electrostimulation of the cortex. It is perhaps as good as certain that no such thing will ever be accomplished in the future. It will remain a factual impossibility, none the less, there is no logical or conceptual barrier to our engaging in speculation about such a thing: the speculation is about something logically possible, something not selfcontradictory or incoherent. We can conceive, as the objector invites us to do, of neurophysiologists producing such a neural process, and also draw a conclusion which follows from the Correlation Hypothesis. It is that if the process were brought about artificially, someone would then be believing that the art galleries in Paris close at 7 p.m. It follows from the hypothesis, and this is the essential premiss of the objection in hand, that if the process were brought about in the given way, or indeed in any way whatever, someone would be having that belief.

Let us now add some more supposing. We add that the person is, say, an African lad who has never heard of art galleries, Paris, or clocktime. His life-experience has been such that he has had no conception of these things. This, we are told, brings us to absurdity. It is quite absurd to suppose that if the neural process in question were made to occur in his brain, he would then have the belief in question. That this would happen, however, as already allowed, is a consequence of the Correlation Hypothesis. The hypothesis is not partly to the effect that if a certain type of neural process obtains, and a person has a certain history, a certain mental event will occur. It is partly to the effect that if a type of neural process obtains, a certain type of mental event will then obtain. What follows from the hypothesis is absurd, and so the hypothesis must be mistaken.

To state the objection in a general way, it is not enough, for a person to believe certain things—or to think of them, want them, or decide or intend to get them—that a certain neural description is true of him. He must be a person with a certain history, a certain learninghistory. The Correlation Hypothesis may seem arguable or even true, and certainly not lacking in sense, until one actually fixes one's gaze upon one of its consequences. It is then seen for what it is, naive or worse, since it leaves out of account something essential to belief, which is the experience that precedes it.

Things are not so simple. By way of a preliminary, it is important to keep in mind that the hypothesis does not commit us to anything whatever about the origin or development of neural or of any physical processes. It says nothing about what neuroscience will or may ever do. We are not forced or even inclined by the hypothesis to think that it will ever be a practical possibility to do what has been imagined. We need not give up the thought noticed already (p. 91) that in the world as it is, and as it is likely to be, a man believes something only if he has a certain history. There is the fact of mental indispensability. That is quite consistent with the hypothesis. We are committed to thinking, as rightly said, that if a certain neural process were made to occur in the given way, a certain belief would be had—if an African lad, with no previous experience of art galleries, Paris, or clocktime, were to be the subject of a certain neural process, he would then be believing that the art galleries in Paris close at 7 p.m. The neural process would necessitate the believing.

To come to the point, are we then committed to thinking a thing which in fact is absurd? Given what has been said, there is only one kind of absurdity that is a possibility. Although the objection does not make it clear, and perhaps trades on the fact, it is the absurdity of thinking something that is selfcontradictory or incoherent. But is it selfcontradictory or the like, whatever the practical possibilities, to suppose a person might have a certain belief without having had the experience that always, in the world as it is, goes with it? Is there a connection of a logically necessary or conceptual kind between beliefs and past experience? Until the objection is made specific, by the spellingout of the kind or kinds of supposed logical connection between beliefs and past experience, and hence the kind or kinds of supposed selfcontradiction or the like, it is entirely incomplete. Still, it may be thought to be persuasive none the less.

What is it that moves us, as I allow that something does, in the direction of supposing that the speculation is somehow selfcontradictory or incoherent? Most importantly, I think, it is the truth that for the African lad to believe the given thing when he does, he would also have to have many other beliefs. To have the belief, to make use of a favoured phrase, he would have to have an awareness of certain forms of life. It is somehow incoherent to suppose that he could have the belief about 7 p.m., as we understand it, without also having beliefs about the number of hours in the day and so on. It is incoherent to suppose that he could have a belief about art galleries without also having a grasp of the existence of painters and dealers.

What importantly moves us towards supposing that our speculation is incoherent, then, is a particular implicit aspect of it: the suggestion that a belief might be had, so to speak, in isolation. What is objectionable is this, and not anything to do with the African lad's lack of ordinary experience. This can be brought out by doing some different imagining. Let us, in the unrestrained way suggested by the original objection, imagine that neurophysiologists, in a long campaign of activity, produce in an African lad very many neural structures which go with an appreciation of all the relevant forms of life. That is, they manufacture a great number of dispositional beliefs on his part. They make possible a great number of occurrent beliefs, mental events in our standard sense. We may imagine they begin with fundamentals and go forward from there. In the end, they produce a neural process which is the correlate of a certain mental event, the lad's believing that the art galleries in Paris close at 7 p.m.

This speculation, although there is more to be said of it, does not raise in us the idea that it is somehow selfcontradictory or incoherent. The absence of experience on the part of the lad, the absence of anything that could be called a learninghistory, does not make it so. Of course, to be tedious, it is possible to confuse this truth with a certain falsehood: that the whole business is in another sense imaginable, which is to say something like a foreseeable practical possibility. The Correlation Hypothesis, to repeat again, has nothing to do with the falsehood. It requires only that something is a logical possibility, and indeed it is. If the hypothesis did have the consequence that the belief could be produced 'in isolation', it would be untenable. It has no such consequence.

It may occur to readers that our latter piece of imagining is open to a certain question. We have imagined the artificial production of a corpus of dispositional beliefs. Is it possible to think realistically of this sequence of production without also bringing in mental events along the way on the part of the African lad? Let us suppose the answer is no. (Cf. pp. 146 f.) To be brought to have a dispositional belief about artists and dealers, he must as a matter of fact, let us say, consciously ask certain questions, contemplate this and that. If this is so, it is no harm to the Correlation Hypothesis. What is true, rather, is that the speculation as to die artificial production of the neural correlate of a belief must be complicated, in a way that undercuts the objection. If it is true that any speculation about the artificial production of neural correlates will be a speculation about something somehow approximate to a learninghistory, that in no way affects the hypothesis. It is about such neural events, not how they might or might not come about.

Will someone still object, in an emphatic tone of voice, that the simple fact is that the African lad 'would not have a belief? What this insistence may reveal is a determination to conceive of beliefs and other things in a certain way—as what we can call learned mental episodes. Beliefs, in this conception, necessarily are learned, learned in the course of taking part in forms or ways of life. Anything that is a belief does by definition have this kind of origin. The reply to this must be that if one so chooses to define beliefs, it does indeed follow with iron certainty that the African lad does not have the belief that the art galleries in Paris close at 7 p.m. But it hardly needs saying that there is no necessity whatever of defining beliefs in this way. Certainly it cannot be said to be their ordinary definition, to the extent that such a thing exists.

But further, this is not greatly important. We have in our inquiry arrived at a conception of mental events, including beliefs. It does not have the feature in question. Mental events were not defined as the products of a certain kind of experience. What matters is that mental events, understood as we have understood them, are indubitably the stuff of our fundamental subjectmatter, which is human experience, our conscious lives. It would not seriously matter if there were some divergence between our conception of mental events and an ordinary conception of beliefs.

2.6.5 Standardly Based Mental Episodes

There is a related argument by a philosopher who is in fact sympathetic to something close to the Correlation Hypothesis, but who would, in a sense, restrict its application. He takes the view that such a person as the African lad, or, to be yet freer in our science fiction, a newly manufactured physical replica of me, could not count as thinking of Vienna, as I can, since ex hypothesi he lacks 'a certain historical and cognitive relationship' with it. (Kim, 1982, p. 57) That he cannot count as thinking of Vienna, we are told, can be seen more clearly by way of another speculation. When I think of Vienna at some moment, I have a visual image of a church and thoughts of hot summer weather. Someone else we can imagine, or try to imagine, the lowan, has never been to Vienna, but at some moment has just the same image. His image can in fact be traced to a church in Iowa which he saw in his boyhood. We would hardly say of this person, it is maintained, that he was thinking of Vienna.

Indeed we would not, if he had only the image, which is all that is mentioned in the speculation. What, however, if his mental event is in all its parts and aspects identical with that one of mine which I describe as thinking of Vienna? He speaks of 'Vienna' in connection with his thought and does not speak of 'Iowa'; the right word occurs in his thought; his momentary experience has a recollective aspect; it involves precisely as much internal confusion, if any, as my own; and so on. Differently from the case of the African lad, however, his mental event is owed in part to his past ordinary experience, rather than neurophysiological manufacturing, although not to any experience of Vienna. Does the fact that the lowan's mental event has this bizarre history—it is owed in part to a church in Iowa—make it other than a case of thinking of Vienna?

It is possible to maintain that he is properly said to be thinking of Vienna. What else could he be said to be doing? But, as must be admitted, we also have an inclination to say that he is not doing so. That is, we have the inclination to require, of anyone who counts as thinking of Vienna, that his thought has some kind of standard history. It would be difficult indeed to specify such a history— certainly the thinker does not have to have been to Vienna—but we are inclined to such a requirement.

This does not affect the Correlation Hypothesis. Let us give the name standardly based mental episodes to such things as my thinking of Vienna conceived partly in terms of the history of the thinking. Such an episode by definition requires a standard history. Evidently no neural process by itself can guarantee such an episode. It would have to  reach into the past to do so. But the Correlation Hypothesis is in no way weakened, or made insufficient to any end of the present inquiry, by being limited to mental events. The fact that it does not cover standardly based mental episodes is perfectly consistent with the fact that it covers all of consciousness or experience—choices, decisions, formings of intention, and all of what is relevant to our inquiry.

2.6.6 Twin Worlds

A sixth objection was made by a critic of an earlier version of the Correlation Hypothesis but applies as well to it. (Stich, 1981; cf. Putnam, 1985a; Stich, 1983) It includes two contentions, of which the first, with which we will be concerned for a while, is that there is a categorization of consciousness which has to do with one or another context, and this categorization gives us the commonsense or ordinary conceptions of thoughts, beliefs, and the like. We can see this, it is claimed, by considering several examples.

The first one involves a thought experiment about twin worlds. Harry on Earth, and another Harry on the very similar planet Yon, which has on it some doppelgdngers of Earth folk, have wholly indistinguishable experiences. These consist in mental events of one type, which each Harry might report by saying that he was thinking that President Reagan had been shot. If either Harry had also had the other's experience, he could not distinguish the two in any way whatever. However, since there are also two Reagans in question, it follows in the contextual conception of consciousness that Earth Harry and Yon Harry were having different thoughts.

Is this true of the ordinary conception of thoughts? On the assumption that there is something clear which deserves the name, I doubt it. Certainly it does not follow from the falsehood of what Yon Harry thought—on the assumption that Yon Reagan was not shot— that his thought in an ordinary way of speaking was not the same as Earth Harry's. We shall not change our mind about what thought I had a moment ago, when I reported that my mother's maiden name was Armstrong, if it turns out that I was mistaken. As for the different fact of there being two referents in question, Earth Reagan and Yon Reagan, that will make common sense hesitate about whether there are two thoughts or one, since, so to speak, common sense has not thought much about twin worlds and the questions they raise.

Suppose common sense concentrates on the supposition that the two items of consciousness really are wholly indistinguishable. The two mental events, in their representative contents, are identical. Evidently these representative contents, as they are, are not fixed by anything whatever outside of them, or outside of Earth Harry and Yon Harry. Their aboutness is of their own nature—as is the aboutness of my present thought of the wine in the refrigerator, which is independent of whether the bottle contains vinegar and whether there is any bottle in the refrigerator at all. Common sense, if it fixes on the indistinguishability of the mental events of Earth Harry and Yon Harry, will surely be inclined to conclude that there is the same thought in question, and allow that on different occasions it is made true or false by things that are numerically different. Or, perhaps, common sense will not merely hesitate but be stymied by the question in hand. The contrary assumption, that common sense will take wholly indistinguishable mental events to be different thoughts, strikes me as remarkable.

A second example is different, although it has the same aim: to show that identical mental events may be different thoughts. A young girl has an experience which she reports by saying that she has been thinking that President McKinley was assassinated, and she gives evidence of understanding many relevant things, including the fact that assassination involves death. Grown old, and suffering from senility, she has an experience which issues in her saying again that she has been thinking that President McKinley was assassinated. But she gives evidence that she does not connect assassination and death, does not know the difference between a president and a senator, and so on.

Here, evidently, we might well speak of two thoughts in an ordinary way. But it is surely far, far too brave to suppose that her two experiences would be indistinguishable, that they could be described as identical mental events. It might be that we would accept that there was an aspect of both experiences that was indistinguishable, an aspect having to do with the use of certain words. The experiences at the two times would then involve one type of mental event which was instantiated both times. But this is far from granting that the two experiences were alike in terms of all the involved mental events.

I conclude, despite various ancillary problems, that the contextual conception of consciousness, in terms of what can be called contextual mental episodes, is no closer to the ordinary one, if such a thing can properly be spoken of, than the conception in terms of mental events as we have them. The relative closeness of talk of mental events to ordinary talk of consciousness is of some importance, but my principal rejoinder to our critic has to do with the second contention in his objection, which we have not yet considered.

Certainly the mental can be characterized in the contextual way, and we shall then say that Earth Harry and Yon Harry have had different contextual thoughts athough their experiences are indistinguishable. On the assumption that Earth Harry and Yon Harry are physically identical, we shall nevertheless not have any evidence that the contextual thoughts and the neural processes are in a nomic connection of the kind specified by the Correlation Hypothesis. Certainly—this is the second contention of the objection—it will not follow from nomic correlation between mental events and the neural that there is such a correlation between contextual thoughts and the neural.

It is true, since a contextual mental event is partly a function of other things than the thinker, but it is hard to see why on earth (or Yon) it matters. The claim that there are nomic correlations between contextual thoughts and the neural is not part of the Correlation Hypothesis, or of course a premiss of it. Nor is the falsehood of the given claim about contextual thoughts and nomic connection inconsistent with the hypothesis. It is not the case, either, that any relevant subjectmatter is left out when we characterize consciousness in terms of mental events. There are not two subjectmatters, but two classifications of one subjectmatter, which is all of consciousness. That the subject-matter is open to a classification such that all of it falls under the Correlation Hypothesis is entirely sufficient to raise the questions to which we are coming.

So much for six objections. Let us finish here by glancing back and bringing together what are best described as clarifications of the Correlation Hypothesis, not qualifications of it or concessions with respect to it, and by making a principal point more explicit. The hypothesis is concerned with mental events, the discriminable constituents of consciousness, and hence, as was remarked in the beginning, not with all of what constitutes a personal epistemic fact. (p. 85) It is not concerned, that is, with all of what is the case when I know, believe truly, remember, or see something. Secondly, it is not concerned with all of what constitutes beliefs, thoughts, wants, decisions, or intentions where these are taken as learned mvQtal episodes, explicitly defined as somehow owed to participation in certain forms of life and not to neurophysiological construction. The hypothesis, further, is not concerned with all of what is needed for standardly based mental episodes as defined or with all of contextual mental episodes as defined.

What the Correlation Hypothesis is concerned with is all of consciousness or experience. None is left out. Further, to conceive of consciousness in terms of mental events is to conceive of it in an ordinary way, arguably the most ordinary way. It is not essential to its role in a theory of determinism that consciousness is so conceived, but certainly it simplifies matters. A final and related consideration—the principal point to be made more explicit—begins from the fact that the inquiry of this book can properly be described as an inquiry into the explanation of our choices, decisions, and actions. Speaking very generally, we wish to know which of two sorts of explanation is the true one, a determinist explanation or another, and what follows from that. It is clear that whichever of the two sorts of explanation is true, it will have to do only with mental events as defined, and not with personal epistemic facts or any of the mentioned sorts of episode.

A man may be said to have shot his mistress either because (i) he believed her to be unfaithful to him, or because (ii) he knew her to be unfaithful to him. The explanation given in each case is precisely the same. If I produce (i) as an explanation today, and tonight become convinced that it is a fact that she was unfaithful and he had good reason to believe it, and so produce (ii) tomorrow, I will none the less then be giving no further explanation. The personal epistemic fact about him gives me no further explanation of his action. I may have a further answer as to why he believed what he did, but no further answer as to why he did what he did. (Kim, 1982; Stich, 1978) So with what we labelled as learned mental episodes—beliefs and so on which by definition were acquired by participation in certain forms of life. To cite something beyond the contained mental events is not to give a further explanation of ensuing mental events or actions. The same is to be said of the other defined episodes.


2.7.1 Definitional Precedents

Several bodies of work which are technical, partly as a result of their connection with linguistics, but also schematic and in a rather free way speculative, contain or suggest four objections against the Correlation Hypothesis. (Quine, 1960, 1970; Davidson, 1980).

The first begins from the falsehood of behaviourism. When we speak of mental events we do not mean only something about bodily movements. There is no such synonymy. Behaviourism, sometimes rightly called 'definitional behaviourism', may be thought to be one of a number of futile definitional enterprises in philosophy. It is not possible either, it may be said, to define 'right action' and other terms of the language of morality in a naturalistic way—that is, by way of such nonmoral terms as 'distressminimizing action'. Again, it is not possible, as is widely agreed, to carry out the phenomenalist programme of translating our physicalobject language into a language of sensations or sensedata.

This set of definitional failures, it is said, sharpens our understanding of the issue of psychoneural nomic connection. The set of failures are precedents. They give us hints, some reason to expect that in fact there is no psychoneural nomic connection. (Davidson, 1980e, pp. 21518) The objection, applied to the Correlation Hypothesis, is this: (i) the various definitional enterprises are hopeless; (ii) the Correlation Hypothesis is such a definitional enterprise; (iii) the hypothesis can be expected to be as hopeless, or at any rate must be regarded with scepticism.

It is of course the second premiss that calls out for support by the objector. Many remarks made about nomic connection generally are germane to it. First, a distinction is made between statements of nomic connection, which is to say laws and lawlike statements, and statements of causal connection. The former, it is suggested, are in a particular way languagedependent. Certainly this is not the distinction we have, which makes causal statements a species of nomic or lawlike statements, but a distinction having to do with facts noticed earlier in connection with Anomalous Monism. (2.3.2) It is said, secondly, that the lawlikeness or nomologicality of a statement is much like the analyticity of a statement, since both are linked to meaning. Lawlike statements, thirdly, are such that the predicates in them, for the things stated to be in nomic connection, were 'made for each other', as we can know a priori. Fourthly, and much the same, judging a statement to be a law or to be lawlike is in some sense an a priori matter. More plainly, nomic connection is not established by empirical inquiry but by a priori means. Finally, it is to be expected that a given statement will be lawlike within one theory and nonlawlike in another.

The audacious remarks are not intended as, and, more important, are not near to being, arguments for the startling proposition that nomic connections, say between a hammer blow and the shattering of a goblet, or between the temperature and pressure of a gas, are connections which depend on language, on the synonymy of terms. It is a proposition not much heard of since Hume dispatched it. We have in the remarks no reason to think that the fact of two things being in nomic connection is a fact in the same category with the fact that all bachelors are unmarried, or any sufficiently like category. The remarks may owe something to the known fact that laws of nature may give rise to related analytic truths. (Ayer, 1963a, p. 215 f.; E. Nagel, 1979, pp. 52 f.) and also to a general theory about language and conceptual schemes. (Quine, 1953, 1960; cf. Grice and Strawson, 1956) We shall not pursue these matters. In fact it is difficult to believe that the startling proposition, which cannot seem other than a plain falsehood, is what is really intended. What is really intended remains uncertain.

We earlier arrived at an account of nomic connection—an account having affinities with other widely accepted accounts—which is fundamentally unlike what might be christened the linguistic or conceptual speculation. Further, our own account and those like it are within the broad tradition of philosophical and other belief to the effect that causal and like connections are not, despite certain caveats, a matter of definitional connection, synonymy, analyticity, languagedependence, the a priori, or like things. It is difficult to entertain the idea that the broad tradition, together with the common assumption of science and ordinary belief and practice, is seriously misguided. The linguistic or conceptual speculation is the more distinctive the closer it approaches to the proposition that there would be no nomic connection in a world without language. As it becomes more distinctive, then, it comes closer to what no one can believe.

2.7.2 Holism

We have it that to ascribe a belief to someone, as in the case of the African lad, requires that one also ascribe many dispositional beliefs to him. There is the same requirement with a great many mental events. It is true of thinking, memory, emotion, desire, choice, decision and intention, or at any rate standard cases of these. It is true, it seems, of all mental events whose contents are representative of other things, or those where the representation brings in language, (p. 81) It is not true, then, or at any rate not at all widely true, of sensation and perhaps perception. This general if not entirely clear fact, although use was made of it in dealing with the Africanlad objection, may well be taken as itself more enemy than friend to the Correlation Hypothesis.

Consider a man's feeling at some moment that more than a halfbottle of wine at dinner is excessive. It follows, we can say, that he cannot be ignorant about certain things—he has dispositional beliefs about grapes and about whole bottles of wine. We can add that he has attitudes other than the one explicitly given in his thought. We could not ascribe the particular attitude to him if we believed, for example, that he was in the grips of some general Dionysian conviction that all standards and selfrestraints are nonsense, or that he would in no conceivable circumstances feel selfcritical or embarrassed. That is a small beginning. Moreover, each of the dispositional beliefs and attitudes we mention is itself like the initial mental event. It also somehow entails that he has other beliefs and attitudes, and so on. In short, to ascribe the mentioned feeling to the diner is to be committed to attributing to him a large and indefinite corpus of beliefs and attitudes. This is so because there are logical or conceptual connections or relations between concepts, and nomic relations between most things conceived. It is sometimes claimed, more strongly, that a concept has its very existence in these connections or relations. This is unclear. If what is meant is the idea that a concept cannot in some good sense stand on its own, be itself individuated, it is certainly mistaken. If the latter were true, logic as we conceive it would presumably be impossible, since it would be impossible to specify relations between single thoughts. The diner's feeling, it is as true to say, was not other or more than the feeling that more than a halfbottle of wine at dinner is excessive. That is consistent with the fact that concepts do stand in relation to others, and hence that it would be mistaken or even senseless to speak of someone having but a single belief. To have one belief a person must in some way have more.

All of which is one thing that can be meant by the dictum that mentality is holistic, that it is to be thought of in terms of wholes rather than either items or collections of independent parts. The same facts can be put into other lights. They can be regarded, as has been popular, in several epistemological ways, as fundamental to the answering of certain questions or the development of certain theories. A first question posed is that of how we understand another person, how we come to judge or know that he or she is thinking, meaning, feeling, or intending something. Or, what comes to much the same thing, there may be considered the philosophical problem of devising or constructing a theory of how we do go about ascribing mental events to others. This will in part be a theory of the translation or interpretation of utterances. Secondly, there may be posed the different question of how we explain the nonverbal actions or behaviour of another. Or, there may be considered the problem of constructing a theory of how we do it.

In each case it is rightly said, for example, that we cannot hope to proceed by taking an utterance or action singly and reading off its mental significance: that it expresses or derives from a single desire or whatever. Rather, we make sense of a particular utterance or action by making further judgements or assumptions about the person, to the effect that he has more beliefs, desires, and so on than the one in question. We assign a complex system of beliefs and the like. As we must, we consider a whole rather than any single part in isolation. Our theoryconstruction cannot be piecemeal.

Mentality is holistic, then, which is essentially to say that if we are to ascribe a mental event of common sorts to a person, we must also ascribe a good deal else to him—a whole, a related body of things, of indefinite size and shape. More follows from this, it may now be supposed, than that a neurophysiologist could not conceivably manufacture an 'isolated' mental event in someone, without doing a lot else first. There is a more fundamental consequence. The fact of mental holism undercuts our second speculation, having to do with the supposedly successful neurophysiological procedure, and it undercuts more than that. This is so, it may be supposed, because the objection from holism ruins the very ideas of a mental event and of the neural correlate of a mental event, whether the latter is artificially produced in whatever way, or entirely natural. The fact that consciousness is holistic thus ruins the Correlation Hypothesis. The hypothesis asserts specific connections between determinate items, but there are no determinate mental items to be had.

Furthermore, it may be said, any neural correlate of the mental event when the latter is taken seriously, which is to say the event and what necessarily goes with it, must be a correlate for that whole indefinite body or network. The neural correlate in question must be something obscure, a large and indeterminate neural somethingorother. There would be no point in merely labelling it, for example, as the general type of neural correlate related to thoughts about wine and excess, or anything of the sort. We would not thereby get a grip on anything.

That is the objection from holism, or rather, one explicit objection that can claim the name. It cannot in fact be assigned to any philosopher associated with the doctrine of mental holism, but is derived from their reflections. (Quine, 1960; Davidson, 1980e, f, g; cf. Peacocke, 1979b) It or something like it has been an unofficial helping hand with nearby objections to the Correlation Hypothesis to which we are coming.

Should the objection from holism in fact persuade us to abandon hope of such views as the Correlation Hypothesis? It seems, on reflection, that anyone persuaded will have been persuaded by a kind of illusion, which owes something to the distraction of large theories of language and meaning. The illusion is dependent at bottom on the failure to respect an essential distinction, one we already have on hand, but which it is possible to forget. Let us reflect on the crux of the objection from holism. Why is it that if the diner cannot be assigned his particular feeling, m, without also being assigned a collection of other items, then we cannot suppose that m is determinate, and further, that it had a determinate neural correlate n? To consider the question, I think, is to be at least uncertain of the answer.

To consider the question, what certainly needs to be remembered is that the further items are not themselves mental events but dispositions. As we have already assumed, such dispositions are neural structures, (pp. 86, 128) They are neural structures such that persons who have them are disposed to certain mental events. In certain circumstances, they will have certain thoughts, feelings, or whatever. These dispositions, incidentally, are therefore exactly like dispositions generally. That is, if we put aside a certain amount of mystery about 'powers' and the like, as we shall, they are sets of standing or persisting conditions. Any disposition is a set of standing conditions such that when it is conjoined with a further condition or conditions, the resulting whole is in fact a causal circumstance for something. To say that a cube of sugar will dissolve in water, that it has that disposition, is to say that the sugar has certain properties such that when they are conjoined with certain properties of water, the resulting causal circumstance has the effect of the sugar's dissolving.

Thus, if we begin with the diner's mental event, and then ascribe to him a great deal else, we do not thereby ascribe a corpus of mental events to him. It is not as if his mind were flooded with ideas, or that his feeling occurs in a setting or surround of other mental events. It must be such an image that takes the place of an actual argument in the reflections we are considering. Certainly nothing else, and in particular no argument, suggests itself. As for the neural correlate of the mental event, there is no cause to try to think of anything indeterminate, a correlate for some indefinite corpus of conscious items. The additional items we ascribe to the diner are in fact neural things which are in some way necessary to the mental event. The question of their exact relation is a matter of the second hypothesis of the theory being set out, the subject of the next chapter. What is to be proposed now is that the objection from holism collapses or disappears as soon as dispositions are rightly located—at the neural end of the Correlation Hypothesis. They are no complication of the mental end. That end has to do with a single mental event, which mental event is perfectly determinate and can have a perfectly determinate neural correlate.

2.7.3 Indeterminacy of Translation

The third possible objection, or sort of objection, derives from something left out of the sketch of mental holism just given. This is best known as it is presented in the consideration of the problem of translation of a wholly unknown language—hence called the problem of radical translation. There is the alleged fact of the indeterminacy of translation. Consider two linguists, native speakers of language l, who set out to translate language m. They succeed, and each comes to be able to speak m as it is spoken by native speakers. However, they do so by devising different translation manuals, where such a manual includes a dictionary and a good deal else. As a result, consistently with the fact that each linguist has mastered m, there is the upshot that their translations of certain theoretical rather than observational sentences of m—the sentences they give as translations in l -- diverge and indeed conflict. They may differ in truth value or be logically incompatible.

It is maintained, secondly, that no sense can be attached to the idea of one of these translations being correct and the other incorrect. It is said, famously to some and notoriously to others, that there is no 'fact of the matter' as to whether one sentence or the other of 1 is the translation of the given sentence in m. There is not an objective matter to be right or wrong about. (Quine, 1960, Ch. 2, 1970) The description 'the indeterminacy of translation' is sometimes used for the alleged fact of the two translation manuals, sometimes for the alleged fact that neither can be said to be correct.

Despite what has been said so far, the aim of these reflections is not an understanding of the nature of translation but of something larger, language and meaning. Essentially the same two views are taken of the interpretation by one speaker of a given language of certain sentences uttered by another speaker of that language. Extremely little flesh is put on the bones of the core doctrine of indeterminacy when it is expounded in terms of translation, and yet less in connection with interpretation. The same is true when the core doctrine is stated in the nearby terms of meaning. We are to understand, however, that the situation is as with translation. With respect to certain sentences of our language, as it seems, you and I may understand them differently, and there is no fact of the matter as to which of us is right. There does not exist the true understanding or the meaning.

As in the case of the doctrine of holism, the doctrine of the indeterminacy of translation, interpretation, and meaning is not made to issue directly in objections to the Correlation Hypothesis and like views. Again, however, it is likely that the doctrine in itself has predisposed some philosophers against such views.

(i) One of two possible objections is that indeterminacy puts insuperable difficulties in the way of confirming the Correlation Hypothesis, since there must be indeterminacy in the ascription of beliefs and the like. There are several brief replies, (a) The doctrine, as it must be in order to get consideration, is limited to the translations or interpretations of certain sentences, the ascribing of certain mental events. It is limited to theoretical as against observational sentences and the like. Thus, if we were to accept all of the doctrine, it remains true that we would be in no doubt about many, indeed countless, ascriptions of belief and the like. To take an entirely rudimentary example, we can have a fully warranted belief, about a man reporting his experience at two moments as he sits facing a red screen, that his perceptual experience was indeed identical at the two moments. It is also possible to have an odd view of what is needed to confirm the Correlation Hypothesis. It is a matter of inference from several kinds of evidence, of which more in due course. In short, although the matter is a large one, the doctrine of indeterminacy is not inconsistent with the confirmation of the Correlation Hypothesis, (b) If we need a certain grip on a man's experience in order to answer a question about certain mental and neural events, and we have not got that grip, then we are unable to answer the question. To be in trouble is to be unable to see which of two answers is correct, one for and one against nomic correlation. It is not to be able to see that one answer is mistaken. If exact and final judgements about some mental events are impossible, then this contributes precisely as much to uncertainty about the nonexistence of psychoneural nomic correlation.

(ii) The other possible objection, in brief, is that a part of a man's experience simply is not determinate, but is itself in some way indeterminate, and therefore cannot be conceived in terms of the Correlation Hypothesis. The notorious conclusion of objective indeterminacy, to say the least, has not been shown to follow from its premiss. That is, to speak again in terms of translation, it has not been anything like shown that the presumed fact of two translations entails the impossibility of a correct translation. The conclusion, of course, is in analogy to what is perhaps better known: for example, the idea that something in certain ways not open to measurement lacks certain dimensions. The notorious conclusion, if taken seriously and as having significantly wide application to mental states, would undercut virtually all reflection and theory about the mind. It would dissolve a part of the subjectmatter, to the benefit of no view in particular, including the Identity Theories to which the proponents of the view are inclined. (Quine, 1960, p. 264; Davidson, 1980e)

2.7.4 Domains

We come finally to the objection from domains, the different domains of the mental and the neural. It is specifically advanced against psychoneural nomic connection. (Davidson, 1980e) It has the doctrine of indeterminacy in its background, but more so the doctrine of holism. That, we have supposed, is that it cannot be that any of many mental events can be ascribed to a person without also ascribing to him, most importantly, a body, system, or network of dispositional beliefs. To say this is in fact to assume something touched upon above, that the given dispositional beliefs are in accordance with at least a certain rationality. It consists in facts of consistency, coherence, truth, evidence, cogency, plausibility, and so on. The diner, as we know, has the disposition to have the thought, in certain situations, of a halfbottle of wine. He must then also have a disposition pertaining to a whole bottle. Moreover, to come to a rudimentary instance of the main idea which we are now considering, this pair of dispositions is such that he will not suppose that two wholes make a half, or that he can take away half of something and have the whole left, or anything else along such inconsistent lines. He rightly takes certain things to be inconsistent, and hence others to be logically necessary.

It is essential, if we are to assign beliefs and the like to him, that we credit him with at least a certain rationality. If a man's utterances showed no consistency from moment to moment, to speak only of that, and we took his utterances to reflect his actual ongoing consciousness, it would follow that he had nothing worth the name of a belief. Certainly we could not assign an occurrent or a dispositional belief to man who said, without hint of a secondary meaning, that he was in favour of drinking only a halfbottle, since that left him with the whole one to start on next time.

Consciousness and the dispositions to it, then, are a matter of at least some rationality. The domain of the mental is a matter of standards or criteria of rationality. In order to ascribe many or most mental events to others, we are under the constraint of taking them to be rational. Consider now our descriptions and theories of the nonmental, including the neural. This subjectmatter will include, incidentally, mental dispositions taken as neural, or mental dispositions in their neural properties. We do not ascribe or impute rationality to what is in this domain. There are instead very different ruling ideas or principles in our description of the nonmental. To describe it we depend, for example, in the measurement of length, temperature, time, and a great deal else, on the postulate of transitivity. (Davidson, 1980e) If x is greater than y, and y greater than z, then x is greater than z. Thus we have two domains or systems, the mental and the nonmental, depending upon disparate assumptions or principles. They do not make up a single closed domain which is necessary for nomic connection.

That is all of the premiss. In brief, beliefs and the like are a matter of rationality, and the nonmental is not. Criteria of rationality do not have counterparts in the nonmental, the description of which is governed by quite different things. Therefore—to come to the conclusion—there is reason to think that there are no nomic connections between the mental and the neural. The argument depends in the end on a quick transition, and is open to a direct reply.

Certainly there is no gainsaying that there are differences between the mental and the nonmental, and the account given of the mental does indeed rest on a truth. It is apparent, too, that the requirement of ascribing rationality to someone in order to understand him would place a certain wholly acceptable constraint on certain conceivable attempts to confirm the Correlation Hypothesis in a direct way, by ascribing certain mental events to him. Still, there remains a certain fact: we have not actually been given a reason for thinking that the given conclusion follows from the disparateness of the two domains. (Cf. Mackie, 1981, p. 350; Lycan, 1982; Kirn, 1979)

Certainly there is no clear general truth to the effect that there cannot be nomic connection between items in different domains, items whose investigation or description depends on different assumptions. There appears to be no relevant sense in which the two domains of the mental and the nonmental can be said to be other than a single 'closed' domain without begging the question at issue. What we need is a true premiss, something independent of the conclusion that there cannot be the given nomic connection. No reason is actually supplied for precluding nomic connection between the two domains.

Two further comments need making. The argument from domains, as we know, is restricted to mental events that have what can be called representative or perhaps prepositional content. It does then pertain to a great deal of consciousness, but not to all. It is not advanced as relevant to sensations, for example. Thus it is not denied that there is nomic correlation between the neural and some of consciousness. This discontinuity is unreassuring, to say the least. If anything can count as a single domain, it is consciousness or experience as a whole. It alone, all of it, has the character of interdependent subject and content. It is most curious to suppose that it falls into two truly fundamentally different parts, and, it might be added, two parts distinguished in the given way. If nomic connection is allowed in one part, that must necessarily cast doubt on any argument against its possibility in the other part.

The other comment, which can be made in connection with more objections than the present one, can be brief. We are presented with what are properly called, somewhat sceptically, theoretical considerations. They are to persuade us of the falsehood of the Correlation Hypothesis. There is evidence relevant to this empirical hypothesis, to the empirical question of whether the connection between the mental and the neural is nomic, which does not consist in theoretical considerations in anything like the same sense. It is the evidence provided by direct research into the subjectmatter, which is to say neuroscience. It is no blinkered empiricism to regard such evidence as superior to the given theoretical considerations.

The objection from domains, then, appears to be like its predecessors in failing to prevent further consideration of the Correlation Hypothesis. The hypothesis may not be true, but it does not fall victim to the philosophical and theoretical objections we have considered. It is not the whole story of the psychoneural relation, as we shall see, but we have found no reason to doubt that it is the main part.

Editor's Postscript
This inquiry into the traditional and still orthodox accounts of the connection between brain and consciousness is different indeed from Ted Honderich's subsequent thinking. The traditional and orthodox accounts consist in what can be called cranialism. Compare the theory of Radical Externalism or Consciousness as Existence. Go to Radical Externalism: Honderich's Theory of Consciousness Discussed. Also the entries on consciousness in the index of this website and in particular the paper Consciousness as Existence, Devout Physicalism, Spiritualism. If you wish to quote from the inquiry above, it would be safest to get a book from which it comes -- books are less likely to have mistakes introduced by scanning machines.

HOME to Det & Free Website front page
HOME to T.H. Website front page