A. J. Ayer's review of A Theory of Determinism: The Mind, Neuroscience, and Life-Hopes by Ted Honderich

    The problem first of clarifying and then of answering the questions how far human thoughts and actions are subject to causality and whether this is consistent with their being free is one to which many different approaches have been made throughout the history of philosophy. I doubt if any of them has been the product of such intense research as Professor Honderich has devoted to the construction, the defence and the evaluation of his theory of determinism. Agreement among philosophers, especially on fundamental questions, is difficult to reach, and I shall be arguing against Honderich's theory at many crucial points. Nevertheless, I think that his readiness to accept even the most startling implications of his views, the patience he displays in examining alternatives to them, his assiduity in setting out and trying to meet a wide range of objections, are all highly creditable to him.

    The book, which runs to well over six hundred pages, is divided into three parts, ten chapters and 74 sections. It begins by elaborating a view of causality as a form of necessary connection, and proceeds to develop a hypothesis concerning the 'nomic correlation' between a person's states of consciousness and states of his central nervous system, from which Honderich extracts the concept of a mental item and its neural associate as 'a psychoneural pair'. A chapter on the causation of these psychoneural pairs, leading to a shorter chapter on the causation of actions, brings the first and most philosophical part of the book to a close.

    In the second part Honderich turns to science. Its title is 'The Truth of the Theory', which prompts me to say both that I agree with him in taking the theory of determinism to be empirical, and that I regard his marrymg philosophy with science as an example of a current trend which is much to be welcomed. This part of the book consists of a very long chapter on neuroscience and quantum theory, followed by a shorter discussion of the topics of prediction and knowledge. Honderich's acquaintance with the natural sciences, like my own, is only second-hand, but he has been conscientious in seeking out the best sources for his account of contemporary physics and neurology. I shall nevertheless be arguing that he underrates the threat which Quantum Theory poses to the very strong version of determinism which he upholds, and that his belief in the total predictability, not only of all human experience and action, but even of purely physical states and processes, goes far beyond anything that is warranted by the empirical evidence at our disposal.

    By now we have reached page 379 and there are still 236 pages to come; followed by 22 page of references and an index of 25 pages. There are also diagrams which do not enter into the pagination. A lot is being asked of those whom Virginia Woolf addressed collectively as 'the common reader', but there will be a reward for their persistence. The third part of the book, which explores the bearing of Honderich's theory of determinism on our claims to knowledge, our moral judgments, the rationale of reward and punishment and the concept of free will, is the most interesting and the most original.

    My philosophical disagreement with Honderich is rooted in our different conceptions of causality. He construes particular causal statements, symbolised in the form of the conditional 'If c then e,' as expressing a necessary connection between the individual properties of things in space and time. In this formula the cause necessitates the effect in the sense of being required for it in the existing circumstances. What fully necessitates the effect is a temporally precedent set of 'causal circumstances' cc of which c is a member. At one point Honderich writes of the effect as having to happen 'given things as they were', thereby making the relation between cause and effect symmetrical, but some twenty pages later he destroys the symmetry by retaining the rule that a set of causal circumstances necessitates a particular effect, while allowing an effect to necessitate only one or other of a number of sets of causal circumstances. Causal conditionals are said to be transitive, in the sense that when the propositions 'If p then q' and 'if q then r' express causal truths, it is supposed to follow that 'If p then r' is also a causal truth. This may not seem counter-intuitive if one mentions only three propositions, but the extension of the transitivity over even a century, let alone the billions of years that have elapsed since the universe came into existence, if it had a beginning, has strange results. It is at least not obvious that the proposition 'If Queen Victoria thought Prince Albert handsome, King George V collected postage stamps' expresses a causal truth entirely on its own.

    I have to say that I disagree with Honderich not only on this point but over almost every opinion that I have so far attributed to him. To begin with, I see no need for the concept of individual properties. I do not see why the colour red should not be directly instantiated by several of the books on my shelves; to require it to be instantiated by the redness of each book appears a gratuitous complication. More importantly, the insistence that a cause and its effect can always be represented as properties, whether individual or universal, of physical objects seems to me wantonly Procrustean. Can it even be adapted to events tike a flash of lightning or the darkening of the sky, let alone omissions like the absence of rainfall, or social phenomena such as a rise in the rate of interest or the demise of a political party? My own proposal that 'c causes e' can be replaced by the formula 'Its being the case that p has a lawlike connection with its being the case that q' has the merit that it allows an unrestricted range of sentences, including negative sentences, to express the existence of particular causes and effects. To be sure, my near-Humean interpretation of 'having a law-like connection' is far too lax for Honderich's taste, and he would also object to my introducing any element of generality into particular causal statements, maintaining, as he does, that the causal conditional 'If p then q' does not refer to anything beyond what is contained in the assertion of p and q. He nullifies this, however, by his unobtrusive adoption of the principle 'same cause, same effect'.

    If, in common, I am bound to say, with most other philosophers nowadays, Honderich believes that I ask too little of the causal tie, I believe that he makes it far too strong. For he requires, in order that a set of causal circumstances should necessitate a particular effect, that the relation should continue to hold whatever changes were to occur in the universe, provided only that these changes are logically compatible with cc and with e. To take a simple example, if there is even any time interval between the striking of a match and the lighting of a cigarette, a gust of wind may blow the match out: but the force of the wind is not logically but only causally incompatible, if that, either with the striking of the match or with its chemical composition or with the ignition of the cigarette. Honderich can, indeed, protect himself against this counter-example by including the negative condition of such a wind's absence in his set of causal circumstances: but short of making his proposition analytic by defining at least one of the causal circumstances in terms of its production of the effect, he cannot rely for very long upon this avenue of escape. For instance, I doubt if he would wish to make it a defining property of the anopheles mosquito that it has a propensity to cause malaria, or of the drug Atabrin that it serves as a prophylactic against the disease.

    This objection is very serious for Honderich because it can be used to attack the central feature of his book, his construction of what he calls psychoneural pairs. He assumes, on scientific grounds, that every mental event is correlated with one token of a member of a disjunction of types of neural events. To avoid triviality, the range of the disjunction needs to be limited and preferably specified, but I am not pressing this point. It follows that a mental event is necessary to the occurrence of the simultaneous neural event: in other words, the mental event would not happen in the absence of its neural partner. The converse also holds. Neither event would occur in the absence of the other. Their combination, however, is not logically necessary. Consequently, on Honderich's view of causal necessitation, there is nothing of which either the neural or the mental event can be an effect on its own. They have to function causally as a pair. This fits in nicely with Honderich's overall scheme, since he can then represent mental events as properties of the central nervous system, without having to accept any of the current identity theories, which he shrewdly criticises. Unfortunately, we have seen that the merely logical compatibility of cause and effect with any conceivable change in nature fails to knot his causal tie, and this removes a vital step from an ingenious argument.

    Concerning th quantum theory, I shall remark only that it undeniably defeats La-place's formulation of the principle of determinism. Laplace claimed that if one knew the exact position and momentum of every particle at any given time, one could. calculate their exact positions, masses and velocities at any other time. The quantum theory implies at least that neither the antecedent nor the consequent of Laplace's conditional could be satisfied. There can be no knowing the exact position and the exact momentum of subatomic particles at any given time. This follows from the weakest construal of the principle. If, like the large majority of contemporary physicists, we accept Niels Bohr's 'Copenhagen' interpretation of the theory, no meaning is to be attached to the statement that a particle has at any given time a definite position and momentum, since the exact measurement of either precludes an exact measurement of the other. As an old friend of the verification principle, I am fairly happy with this conclusion. Honderich is not at all happy with it, and tries to shrug it off. There are many who will sympathise with his scientific realism but apart from that, his struggle with the quantum theory does not yield much more than an expectation that it will be superseded.

    One reason why I am not dwelling on the quantum theory is that Honderich's determinism, in my view, goes further than even Newtonian physics would legitimately carry him. For instance, he commits himself to the proposition that any action that anyone performs 'is the effect of a causal sequence whose very earliest initiating elements occur at roughly the time of birth of the person'. One may be puzzled here by the expression 'earliest initiating elements', since Honderich must surely hold that this particular causal sequence is only a small series of links in a much larger causal chain. Indeed, if I understand him rightly, he accepts as a consequence of universal determinism 'that each future event is conditionally predictable with any desired degree of precision.'

    It is here, I believe, that he would still come to grief, even if he were rid of the menace of contemporary physics. Admittedly, in classical mechanics, universal determinism was an accepted feature of an abstract theory. My point is that it did not hold good at the level of observation. It was rare that the measured result of an experiment was exactly what the theory predicted. The experimenters were satisfied with fairly close approximations, ascribing their discrepancies to 'observational errors'. But, as Charles Sanders Peirce pointed out, this was simply a term of art: there was seldom any good reason to believe that there was anything wrong with the experiments. In his view, and in mine, the deviations were a matter of chance, and I follow him in taking chance to be an objective factor in nature. It does not worry me at all that most physicists hold its fundamental Jaws to be statistical.

    Let me say at once that if I am right about the objectivity of chance, I do not consider that I am thereby giving any encouragement to the champions of free will. For it seems to me that they commonly rely upon a concept of desert, which itself depends upon a wholly mysterious concept of self-determination, and I do not see how there is any place for such concepts if, to the extent to which an action is not completely determined by physical or mental causes, it has to be viewed as occurring by chance. On this point I agree with Honderich, who also rejects such libertarian concepts. The question is how one can then find room for rational belief, for moral responsibility, with its appendages of reward and punishment, and for what Sir Peter Strawson, in a justly celebrated essay, called our reactive attitudes, such as gratitude and resentment, pride and remorse.

    Honderich copes very well with the seductive accusation that any strong theory of determinism rules out the possibility of rational belief. He shows that the fact that me possession of a belief is causally explicable in no way precludes its being rationally assessed. He might with advantage have used the examples of a mechanical calculator, or a chess-playing automaton, which plainly validate this distinction.

    In tackling the concept of free will, Honderich agrees with Hobbes and Locke that a person can properly be said to be acting freely when he is not prevented by external or internal constraints from doing what he chooses. Like them, he allows no meaning to the question whether the agent could have chosen otherwise. Here we again part company, though not very seriously. I find it reasonable to say that an agent could have chosen otherwise in cases where our stock of accredited beliefs would not have enabled us to predict that he chose as he did. This falls short of what any staunch libertarian would require.

    When it comes to moral responsibility and the ensuing problems, Honderich finds an original way of undercutting both those who affirm and those who deny the compatibility of free will and determinism by representing the issues as matters of policy tather than matters of belief. The rationale of holding people responsible is that it may bring about an improvement in their behaviour. Acceptance of even so strong a theory of determinism as Honderich's does not commit us rationally to succumb to an attitude of dismay. If we believe that we are not always prevented from doing what we choose, we caa envisage ourselves as acting 'in enabling circumstances rather than frustrating ones'. Honderich goes on to explain 'that these sircumstances have to do with at least the way of my wold, the absence of self-frustration, independence of others, and absence of bodily constraint.' Our 'life-hopes' may then be subsumed under what Honderich calls 'intransigence'.

    I think that we need to know more in this context about what is being counted as 'independence of others' and 'bodily constraint'. Otherwise one might suspect that Honderich is opening the door to one version of compatibilism. A fact, also, to which he should have paid more attention is that whatever view one takes of determinism, it is only in a very general and limited way that we are actually able to predict what is to come. If we knew in detail what lay in store for us there might be occasion for dismay or even intransigence, but none at all for hope. The best we can do, on Honderich's principles, is hope that our attitudes will be optimistic. Honderich would not pretend that his desire that his readers should entertain this hope is the conclusion of an argument, but it is thoroughly characteristic of the man.

A letter to the editor about Ayer's review

    Sir: None ever wished it longer, said Johnson of Paradise Lost, thereby making us forget what else he said of it. Professor Sir Alfred Ayer maybe wished shorter my A Theory of Determinism: The Mind, Neuroscience and Life-Hopes (LRB 19 May). Still, he approves it for a singularity or two, and it is lovely to have the approval of someone who has never been in much danger of the fulsome. I reply with gratitude, then, but reply I do, in something like his own spirit of independence, since he has not convinced me of anything.

    His principal objection is to what he certainly over-describes as the central feature of my book, part of my solution to the mind-brain problem, the problem of the relation of a mental to a simultaneous neural event. What I say is taken to depend on my view of causation. That view, in sum, is that an effect is an event preceded by a set of things -- a causal circumstance -- such that the event would still have followed whatever else had been happening in addition to the causal circumstance and the event.

    Professor Ayer's objection has the premise that we typically do not or cannot specify a whole causal circumstance for an effect. Rather, we may think of something or other prior to the effect such that if certain things had been happening in addition to it, the effect would not have occurred. When the match lights, we may think of the striking and a bit more, and leave out the absence of a gust of wind. I agree with all that, and talk a lot about it in the book. I do not at all agree that it supports the conclusion he tries to draw from it, which is that we do not understand effects to be events related in the given way to antecedents, that we do not have that conception of them. It plainly doesn 't follow. Further, anyone who supposes we do not take effects to be related in the given way to antecedents is on the way to supposing we take them to be a kind of mystery, which exactly we do not.

    Professor Ayer has therefore found no vital flaw in my mind-brain solution. That solution, by the way, as I am reassured to report, does not contain the second part of a thought he assigns to it. That is the thought 'that a mental event is necessary to the occurrence of the simultaneous neural event: in other words, the mental event would not happen in the absence of its neural partner.' Other words needed, dear Freddie, and a correction, too, to something similar in your fifth paragraph.

    In dealing with the third and culminating part of my book, which he likes best and to which he gives least attention, Professor Aycr does not shed a clear and consistent light. He gives no leg up to that common reader he mentions earlier. He says that I agree 'with Hobbes and Locke that a person can properly be said to be acting freely when he is not prevented by external or internal constraints from doing what he chooses.'  That report, since absence of constraint is of course compatible with determinism, makes me sound like yet one more compatibilist, one more member of what must surely be a wearied philosophical tradition, whose burden is that determinism and freedom are logically . consistent.

    In fact, it is exactly as true or false that I agree with Kant and with, say, Bradley and Sartre, all of them of the opposed Incompatibilist tradition, that a person can properly be said to be acting freely when he is not constrained, and also has originated the action. Further, I do not, in the sense Professor Ayer says, reject all concepts of origination or Free Will, allow them no meaning. I don't come close to thinking them insignificant in our lives either -- absolutely the contrary -- nor suppose that determinism poses no threat to rationality. In a way it does.

    Professor Ayer does indeed go on to convey something of this, but tardily, and without making at all clear what is in fact a new view of things, whether right or wrong. It has to do essentially with what are called life-hopes, and, more important, is a view that escapes darkening preconceptions given us by Hobbes, Kant et al -- by both of the traditions of thought about the human consequences of determinism. It is a view, too, that is relevant not only to determinism but to something more popular, near-determinism, which is the idea that there is irrelevant micro-indeterminism down below, as suggested by an interpretation of quantum theory, but macro-determinism up above, certainly including neural determinism.

    Finally, to look at but one secondary matter about which Professor Ayer has not convinced me, he may sound a bit persuasive when he remarks that 'it is at least not obvious that the proposition "If Queen Victoria thought Prince Albert handsome, King George V collected postage stamps" expresses a causal truth entirely on if own.' However,   what he is supposed to be showing, at bottom, if I have him right, is that if some causal circumstance including Victoria's inclination really had as an effect some later fact about the existence of George, and that later fact really caused him to collect stamps, it nevertheless isn't right to say that the initial circumstance caused the collecting.

    Who, as Johnson might have said, unsound as he was on Free Will, could cudgel the mind to believe it? If A caused B, and B caused C, then A caused C. This fact is perfectly preserved when causal statements are analysed into what they are, which is certain conditional statements. Nor would the conceptual truth be complicated, as Professor Ayer suggests, if the period of courtship and gestation for the Royal Family were longer, and George appeared on the scene only some millions of months after Albert caught Victoria's eye.

    There is more to be said. common reader, much more --- about those bits on Quantum Theory and classical mechanics, and whether causes are individual properties, and that fast argument about margins of error for the reality of chance -- but necessarily I stop.

Ted Honderich
University College London

Ayer's review was in The London Review of Books, 12 May 1988. Ted Honderich's letter to the editor appeared on 2 June. For elaboration of one point, go to Freedom and Responsibility Neither Compatible Nor Incompatible With Determinism -- a Note on Why.

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