A THIRD CONCEPT OF FREEDOM OF THE WILL
determinism and Freedom Philosophy Website --
useful for philosophers within one language to make their way
occasionally into the contemporary philosophy of another language. They
may find that their own thinking is congealed. They may find that what
is alive and what is dead in
Oxford, London, Boston and New York is not exactly what is alive and
dead in, say, Hamburg. Nor is it possible to sustain the illusion
that there is more philosophical intellect in Oxford than in
Hamburg. The nature and in particular the inevitability of philosophy
is against such nonsense. Prof. Dr. Ulrich Steinvorth of the
philosophy department of the University of Hamburg is in the high
tradition of German philosophical argument. He is also aware of and not
content with the English-language thinking that there are two ideas of
free action -- roughly speaking, action fully in accord with the
agent's desires, and action in accord with free will in a traditional
and obscure sense. There is, he believes, another good idea. It should
not have disappeared in the time when Kant was confuting Hume.
Abstract: Contemporary neuroscience has stimulated interest in the idea of free will, since a wider a wider public had the impression that neuroscience might prove that there is no free will. In the discussions between neuroscientists and philosophers there is a tendency to oppose two conceptions of free will, a Kantian or libertarian one, which is considered incompatible with science, and a compatibilist one which is presented as being in accord with science and determinism and defending a weak notion of free will and responsibility. Such an opposition does not take account of a third conception of free will that up to Kant’s time was generally accepted both by the adherents and the critics of free will. I present this conception (which I call scholastic), distinguish it from the Kantian conception which still underlies most contemporary thinking about free will, and point out its virtues.
Here are some of the definitions and descriptions of free will taken from Kant’s Critique of Pure Reason:
“a special kind of causality in virtue of which events in the world happen”,
“a causality of starting a series of events completely by itself”, 
“a faculty of really and truly starting a state, hence a series of consequences of it”,
“a spontaneity which by itself can start acting and excludes the presupposition of another cause that would determine its action according to the law of causal connection”.
There are two elements in Kant’s conception: causality and spontaneity. Free will is a faculty of causing events, yet its agency is spontaneous. What does that mean? First, that it is not caused by other causes. Second, that it works “completely by itself”. You’ll see shortly that we must distinguish these two points, though Kant probably did not. I’ll call a spontaneity that is not caused conditioned spontaneity, and a spontaneity that works completely by itself, absolute spontaneity.
Both the idea that free will is a causality and that it is a spontaneity are not new in the history of philosophy. What is new is the way Kant combined them. Hume taught modern philosophers that to understand free will we must understand it as a causality. If we ascribe free will to men without admitting that men cause their actions, actions would just happen to them and they could not be their authors nor be made responsible for them. But if they cause their actions, they are, as Hume presupposes in agreement with Newtonian physics, elements in the chain of causes and effects that binds nature together. Hence what is called free will is in fact being caused, determined by causal laws and not free. Hence, it is one of those “confus’d ideas and undefin’d terms, which we so commonly make use of in our reasonings”.
Kant hoped to escape this consequence by arguing that there is a kind of causality beside the natural one, a causality by freedom, which is spontaneity. The idea of an uncaused causality is already used by Aristotle. Aristotle explains motion by the agency of movers that are themselves unmoved. Yet Aristotle’s unmoved movers act in space and time. In Newtonian physics, uncaused movements in space and time are impossible. Hence Kant, to save the idea of unmoved motion, must declare that the causality of free will is of a completely different kind than the causality of the objects in space and time. It is no causality that we can use to describe and explain observable events, but “a pure transcendental idea” never to meet in experience. We must postulate it as a condition of the possibility of a faculty in virtue of which we are capable of absolutely spontaneous actions.
Many contemporaries agree with Kant that free will in a strong sense must be an absolute spontaneity, something that produces an action out of nothing, since if there is something from which it will produce an action, it is caused or determined by it. Yet for good reasons most contemporaries think as well that an absolute spontaneity is impossible. From nothing arises nothing, not even an absolute spontaneity nor its acts.
This is a first defect in Kant’s conception of free will. There is another defect. By his opposing causality by freedom and natural causality, Kant commits himself to rejecting any empirical criterion for distinguishing actions of responsible persons and actions of lunatics. If free will is a transcendental idea, it is a necessary condition of all action. If we can never meet the properties of a free action in experience, we can no longer say of a certain action that it lacks or shows such properties. Ascribing someone free will and, hence, responsibility, becomes a matter of belief or arbitrary conventions.
I conclude that if free will is conceived as an absolute spontaneity the conception is incoherent indeed and we must reject it, as most contemporary neuroscientists do. I presuppose that this conception is not only followed by Kant but also by libertarians. Nevertheless, it does not follow that we must prefer a compatibilist conception, as is implied by the popular opposition of libertarian and compatibilist ideas of free will. For there is a third conception that up to the time of Kant has even prevailed in philosophy. I call it the scholastic conception. It does not conceive of free will as an absolute spontaneity and is yet incompatible with the understanding of determination of events that prevails up to our days.
The scholastic conception has been formulated by several authors. I follow the definition given by the late Spanish scholastic Luis de Molina which seems to have been the best known in the 17th and 18th centuries. Molina defines freedom of the will as
“that which, if the conditions of doing so are met, is capable of acting and not acting, or of doing something in such a way that it might as well do its contrary”.
This definition presupposes that that which is capable of acting and not acting thinks of, or deliberates about, that which it might do or not do. The object of deliberation was called proposition. Propositions are thoughts, but thoughts that incline us in a specific direction. They are conscious or reflected impulses. That is why the capacity of acting and not acting is a capacity of saying both yes and no to the proposition or the thought of an action we incline to.
Let us formulate the scholastic conception of free will this way:
x has free will as regards action a iff x is capable of saying both yes and no to the proposition that proposes to do a.
The condition that is used in this definition to define free will is today often called the principle of alternative possibilities.
Let us note the most important differences to the Kantian conception.
First, free will in the scholastic conception is no faculty of an absolute spontaneity. It does not produce an action out of nothing. Rather, it responds to a thought that can be caused by natural causes or may happen by chance to a person. Therefore, it is a power of responding to thoughts. It is not a creative power but a power of intellectual criticism.
Second, in spite of being a responding faculty, its responses are not determined since they can be both affirmative and negative. Its liberty consists in its capacity of both following an impulse and blocking it. It is a power of denying any impulse we can make the object of deliberation, hence of opting between at least two possibilities. That is why the scholastics called it a power of indifference, i.e. a power of making oneself indifferent to the attraction of an impulse.
What is most intriguing in the scholastic conception is its idea of a conditioned spontaneity. Free will presupposes stimuli to act on, yet it responds to them by choosing among alternative possibilities. Though its choice is unpredictable, it does not render the agent a lunatic. Rather, authorship and autonomy spring from free choice among possibilities. Before considering this point, let us test the claims of the scholastic conception and ask, first, does it, unlike the Kantian conception, define free will by an empirical criterion? Second, does it, like the Kantian conception, define free will? Third, is it, like the Kantian conception, incompatible with determinism?
2. Does the scholastic conception use an empirical criterion?
I objected to the Kantian conception that it excludes using an empirical criterion to distinguish actions that spring from free will and actions that do not. Doesn’t the scholastic conception fall victim to the same objection? If we want to find out whether someone is able to say both yes and no to the execution of a proposition, what we observe is that he either says yes or says no. But we never directly observe his capability of saying both yes and no. So there seem to be no empirical criteria for deciding whether he can do the one as well as the other. This, however, would be a wrong conclusion.
Take the case of solubility of a solid in water. What you (“directly”) observe is that the solid either dissolves or does not, never its solubility. Yet this does not imply that there are no empirical criteria to decide whether the solid is soluble. If it dissolves when placed in water it is soluble. We use the fact of its dissolving as an empirical criterion for its solubility. In the same way, capacities, although unobservable, have empirical criteria for their identification. A man can swim if when placed in water he is not drowned. In the same way, a man can say both yes and no to a proposition if when exposed to certain observable stimuli he responds in a certain observable way. The way he responds to them is the criterion we use to decide whether he has free will.
There is an objection against this
argument. The logical
relation of the capacity of swimming to its empirical criterion is
different from the relation of the capacity of saying both yes
and no to its empirical criterion. Even before looking
at what this criterion is, we can see the difference. We may formalise
the definition of the capacity of swimming by reducing the modal
operator can to a when-sentence:
x can swim =df x is not drowned when thrown in deep water
Yet we cannot define the capacity of saying both yes and no to a proposition by reducing can to when. For we obviously cannot define it this way:
x can say both yes and no to p =df x says yes to p when in s1 and says no to p when in s2
What the scholastic definition claims is that if or when x has free will, he can, in the same situation s, say both yes and no to p. Thus, the can implied by the capacity of free will as conceived by the scholastics cannot be reduced or dissolved.
This difference distinguishes the capacity of free will from the capacity of swimming and other capacities. Yet does it show that there are no empirical criteria for the use of the term free will? Certainly not. This becomes evident when we remember how we decide whether someone is either capable or incapable of saying both yes and no to the same proposition. There are cases when such a decision is difficult, but there are paradigm cases when it is not difficult. A neurotic who does not leave his home without checking several times that he did lock the door is a paradigm case of the lack of the capacity of saying both yes and no to a proposition. For he cannot say no to the proposition that he check the locking of the door.
By contrast, the cured neurotic is a paradigm case of this capacity. He can say no to the proposition that he check his locking the door. It would be silly to argue that he is not also able to say yes to the same proposition. If someone doubted this ability of his, the former neurotic could prove his ability by checking his locking. If he proved unable to do so, we would attribute him another neurosis or a new form of the old one.
So there are empirical criteria for the application of the scholastic conception of free will. The fact that this capacity’s can cannot be reduced to a when-sentence whereas the capacity of swimming can does not show that the scholastic conception is not empirical. Rather, it shows that even capacity terms whose can is irreducible may be applied in accordance with empirical criteria.
3. Does the conception define free will?
In a way, this question is curious. If someone is able to say no to a proposition although he might as well say yes to it, his option is open and hence, it seems, his will is free. But let us be cautious since we want to know what has been understood by free will in the history of philosophy. Perhaps only the scholastics have understood by free will the capacity of denying. In fact, Ernst Tugendhat and Jürgen Habermas have explicitly rejected the identitification of free will with the capacity of denying. According to Tugendhat, this capacity is not freedom of the will but freedom of action; according to Habermas, the idea of free will belongs to the “pre-Kantian philosophy of consciousness” and cannot be defined as the capacity of denying.
So let us ask what the freedom of action is under which Tugendhat wants to subsume the capacity of denying. The common definition is that it is the capacity to act according to one’s nature or will. It has never been controversial that men have this capacity nor that animals have it. Hobbes ascribes it even to water that acts freely if its flow is unimpeded. Nor was there controversy that this freedom looks different in men than in animals and children. The controversy was over whether, to take account of men’s responsibility, we must, in addition to freedom of action, ascribe to them freedom of the will. The determinists argued we need not. Hobbes argued against Bishop Bramhall that men are justly punished because their actions are noxious, not because they do not happen necessarily. The advocates of free will argued that determinists can understand both punishment and reward only as a means to form men’s will the way society or its ruling class wants it formed, and that this understanding is insufficient.
When we look at the history of philosophy for the origins of the distinction between freedom of action and of the will we detect that the distinction depends on harder facts than on how to understand responsibility. It is to the third book of Aristotle’s Nicomachean Ethics that the definition of freedom of action is usually traced back. Yet we can detect there even the origin of the scholastic definition of free will.
Aristotle distinguishes voluntary actions and actions that are chosen after deliberation. He distinguishes them because, as he says, we call voluntary not only the actions of people who can be virtuous and vicious but also the actions of animals and little children who cannot be virtuous or vicious.  As the essential property of a voluntary action he points out that it must have its origin “in the agent”; as the essential property of the deliberately chosen action, that it is done when we deliberate “about things that are in our control and are attainable by action”, not about things that we cannot change or do not want to change.
This distinction appears quite convincing, since we may well say that children and animals can act voluntarily and involuntarily yet cannot choose after deliberation nor say both yes and no to the execution of the thought of an action. Therefore, Aristotle provides us with a solid argument first that we should distinguish between freedom of action and freedom of the will, second that philosophers after Aristotle have understood freedom of action as the faculty of acting voluntarily and freedom of the will as the faculty of choice after deliberation about things that we might as well do as not do.
Aristotle provided even the scholastics with the principle of alternative possibilities. For to describe a specific property of deliberately chosen actions he uses the description “when the origin of an action is in oneself, it is in one’s own power to do it and not do it”. Moreover, he defines the voluntary not only by the concept of an action’s having its origin in the agent, but also negatively by defining the involuntary. An action is involuntary, he says, “when done (a) under compulsion or (b) through ignorance”, an example of the latter being Oedipus’ killing his father. Hence we can define, with Aristotle, voluntary actions as actions that are done neither under compulsion nor through ignorance, and such actions do not presuppose choice after deliberation and can be attributed to children and animals. Therefore, the Aristotelian distinction between freedom of action and freedom of the will would be this:
freedom of action =df the faculty of acting without compulsion and ignorance;
freedom of the will =df the faculty of choice after deliberation whether to do or not do an action.
We may conclude that Tugendhat’s and Habermas’s theses on the meaning of the terms capacity of denying, free will and freedom of action do not stand the test of investigating its historical use. For in the Aristotelian tradition, which dominated the ideas of free will and free action, freedom of the will, in contradiction to Tugendhat’s claim, is identical with the faculty of denying a proposition or suspending an impulse and, in contradiction to Habermas’s claim, does not presuppose any philosophy of consciousness.
But we must also conclude that belief in free will is not that exciting. For it may seem pretty clear that we better distinguish between acting without compulsion and acting after deliberation than reduce the latter to the former. Why then did so many philosophers want to reduce freedom of the will to freedom of action? The answer is that they rejected the idea that choosing after deliberation is a choice that might have been otherwise. Locke is a classical representative of this rejection. He agreed with the scholastics that free will can only be a power of denying or blocking an impulse by reflecting and deliberating about it:
“… we have a power to suspend the prosecution of this or that desire, as every one daily may Experiment in himself. This seems to me the source of all liberty; in this seems to consist that, which is (I think improperly) call’d Free will.”
Nevertheless, he refused to call the power of negation or suspension free will because he rejected the principle of alternative possibilities. If freedom consists “in a power of acting, or not acting”, then a man who exerts such a power “cannot be free. The reason whereof is very manifest: … he cannot avoid willing the existence, or not existence, of that Action; it is absolutely necessary that he will the one, or the other, i.e. prefer the one or the other: since one of them must necessarily follow; and that which does follow, follows by the choice and determination of his Mind, that is, by his willing it: for if he did not will it, it would not be.”
Locke is attacking the very essence of the Aristotelian distinction between the voluntary and the deliberately chosen and their corresponding definitions of freedom of action and freedom of the will. He rejects the idea that deliberation gives actions a quality of freedom that voluntary actions lack. His reason is the idea that whether we choose to say yes or no to the proposition, we are necessitated by causes to do whatever we do. This idea is taken from the principles of contemporary non-Aristotelian physics which postulate that whatever happens, happens in accordance with physical laws that necessitate whatever happens. Curiously enough, this idea did not prevent Locke from sticking to the idea that men have a power of suspension, which is a consequence of the Aristotelian distinction of choosing after deliberation and voluntary action and incompatible with modern physic’s idea of laws that necessitate whatever happens.
4. Is the conception incompatible with determinism?
Like Locke, Leibniz recognised that philosophy has to take account of Aristotle’s distinction between voluntary action and action chosen after deliberation, but unlike Locke was well aware that conceding a power of suspension to the Aristotelian tradition is incompatible with determinism. So he admitted that after deliberation we are not determined by causes. But he added that we are determined by reasons. Reasons, he said, do not “necessitate” the way causes and motives necessitate. Nevertheless, they “incline” us to choose what we choose, and they do so deterministically. Hence, freedom of the will is either compatible with determinism or an illusion. Leibniz allows us to distinguish consistently between the voluntary and the deliberately chosen but forestalls hopes of finding in the deliberately chosen the place of free will. 
Yet even this criticism has found an answer in the history of philosophy. The answer was given already by Descartes. In fact, in his defence of determinism Leibniz argues against Descartes’ defence of free will. When asked to explain his wavering remarks on free will in his Fourth Meditation, Descartes explicitly defended the conception of free will as a faculty of indifference, that is as a “positive faculty to determine oneself to either of two opposites, that is to the following or avoiding, asserting or denying”. What is more, Descartes makes explicit what is implied by the use of such a faculty. He says that “morally speaking (moraliter loquendo) we are nearly incapable of moving ourselves to the opposite, but absolutely (absolute) we are. For it is always possible for us to renounce following a clearly recognised good or affirming an unmistakable truth, in case we think it good that thereby the freedom of our choice (arbitrium) is demonstrated.”
What is implied by deliberation, according to Descartes, is this. First, it turns causes and motives into reasons. Second, reasons are reasons only if they leave us the liberty to reject them. Only because we can choose after deliberation do we act on reasons. What we then act on are reasons only if we have deliberated about what the reasons are reasons for and are capable of both following and rejecting it. So, although we normally follow the reasons that convince us and are determined by them in a way similar to the way causes determine us, considering things “absolutely” we must recognise that we may reject even the most convincing reasons. As a proof of this, Descartes refers to our capacity of acting on reasons that make our actions unpredictable, in particular to the reason of proving one’s freedom.
Before commenting on Descartes’ argument, let us consider Leibniz’s counter argument:
“… they say you still have, even after having recognised and considered everything, the power not only to will what pleases you most but also to will the contrary, just in order to demonstrate your liberty. But mind that also this whim or spite or at least this reason that hinders you from following the other reasons enters the deliberation and makes you be pleased what otherwise would not have pleased you.”
Here I think it is evident that Leibniz is right. Descartes did not (nor wanted to) show that we are not determined in our actions. Yet Leibniz is wrong in what he inferred from that fact. What he inferred was that „A mind that would have the property of willing, and being able, to do and will the opposite of what can be predicted of him by whomsoever, belongs to the set of entities that are incompatible with the existence of the omniscient being, i.e. with the harmony of the things, and hence to the set of entities that have not been nor are nor will be.”
Leibniz is wrong because we know by experience that it is possible for men to do the opposite of what is predicted of them, if only they know what is predicted of them and want to prove their liberty. Children in their many spite phases of development demonstrate this capacity much to the trouble of their stressed parents. Because of this capacity, our actions are unpredictable if only we know the prediction. This is an unpredictability that results not from an overcomplexity of causes but from the fact that our deliberation about the prediction enters the causes. Though in normal circumstances we can rely that people do what they promise and what suits their characters, they become unpredictable if they are challenged by a prediction or the claim that they cannot but act the way they are supposed to.
According to traditional determinism, earlier world states determine later ones with necessity and without exception. It implies predictability of later states by a Laplacean demon or an omniscient being and desists from factual predictability or predictability by man only in (the probably factual) case of an overcomplexity of causes of earlier world states which precludes finite human intelligence from predictions of later world states. Yet the unpredictability of human action does not result from any overcomplexity. The same property that renders human actions predictable under normal circumstances, their controllability by promises, education, reason and reasons, makes them unpredictable when actors feel they are supposed to be necessitated to act as they are expected to. In that case, they can, but need not, use the expectation as a cause or reason to do the opposite. They can act on the reason not to be determined by pretended unchangeable causes.
How are we to describe this curious situation? Deliberation, it seems, brings in a reflexivity of action causes that decouples our actions from potential past causes. Acting after deliberation would be determined by the reasons we follow, yet would not be predetermined. Free will would include determinism but exclude predeterminism. Yet can there be a determinism that is not predeterministic? What is sure is that since the time of Lucretius, determinism has been understood as predeterminism, as a theory that whatever happens is not only somehow determined but is determined by the very first state of nature or the will of its creator. Let us have another look at Descartes’s argument to learn the conditions of a non-predetermined determination of an action.
The first condition is that the action result from a deliberation in which the agent gets conscious of impulses and motives for actions he inclines to. The second and critical condition is that in deliberation the agent suspends his inclinations. As we have seen, Locke admits that we have a power of suspension and yet sticks to the idea of actions being predetermined. Up to this day, many thinkers follow him. Their argument is that in the end, necessarily, we stop suspending action and decide either for the strongest motive or the most convincing reason or argument. In either case, they conclude, we are predetermined.
What is most important in Descartes’s rejection of this approach is that he analyses what it means to be determined by a reason. Once we suspend an impulse, Descartes argues, we cannot but decide for a reason that we must always be capable of replacing by another one. Once we enact the power of suspension, we can no longer follow an unreflected cause or a motive. We turn the predetermining power of unreflected causes and motives into the determining power of a subset of causes, namely, reasons. But it would not be reasons we then must follow, if their power were necessitating or determining our action in an inescapable way. Rather, it is their essence always to be replaceable by other reasons. This essence gives actions that result from deliberation the peculiarity of being determined without being predetermined.
Descartes is well known as the author who gives much prominence to the inescapability of the argument that because I doubt the existence of anything I therefore must exist myself. Therefore, it may seem paradox that the same author nonetheless maintains the escapability of any argument. Yet Descartes relied on sound empirical evidence that men can act against the reasons they judge the best. We know from ourselves and other people that we may reject the most convincing reasons without ceasing to follow reasons. The reason we follow when we reject the most convincing one may be the reason just to try an unconvincing reason or to prove our negligence or coolness or capability of acting against reason or our independence of reasonable reasons or our power to prefer sentiments or passions to reason or to replace reason by pure arbitrary will. But rejecting reason does not mean escape from acting on reasons or a reason. What it implies, though, is that acting on reasons is inseparable from a decision for a reason which yet always might be a decision for another reason.
If it is true that we cannot act on reasons without being able to replace a reason by another one, then actions that result from deliberation cannot be predetermined. But then, aren’t they undetermined? No, they are determined by the agent’s choice among the reasons he can produce to himself. An action that results from a deliberation is never something that happens to the agent. Hence, Hume’s argument that unless actions are determined, they are at best actions of lunatics, proves ambiguous. If we understand him as he wanted to be understood, namely, that actions must be predetermined, he is wrong. If we understand him as we better understand him, he is right: unless actions are determined, they are not actions at all, but events that happen to the agent. But such actions need not be predetermined if they result from deliberation. Determination of actions by the agent is of a different kind from determination of billiard balls by causes.
This result exposes the scholastic conception to another objection. Isn’t it an obscure idea of agent causation and perhaps even of a timeless self that it depends on? What or whom are we to understand as the agent that decides for or against a reason? To neuroscientists this question does not pose a serious problem. They will find out an excitatory brain pattern that represents what folk psychology calls self or agent. The pattern will be excited when a subject deliberates; it will be connected to loops and inhibitions in neuronal processes and be determined by the subject’s history and endowment. Yet if it is given, it will in principle exclude the scientist from predicting which of alternative possibilities a subject will choose. Its determination of efferent neurons may be called agent causation. But this term can be explained by a description of the kind neuroscientists have recourse to when they describe excitatory patterns, loops and inhibitions.
The notion of agent causation is problematical only if we presuppose that the agent has an absolute spontaneity. By contrast, starting from a conditioned spontaneity we need not be puzzled how there can be a self or responsible agent. Rather, we can think of chains of causes that end up in a peculiar effect. The effect is a gap in the chain that otherwise would predetermine the future. The gap is performed by a particular animal’s faculty of suspension, of imagining alternatives to the original determination, and of choosing among them by reasons, i.e. in the end arbitrarily and autonomously. If we must take account of such a faculty, we imply that there is non-predetermined determination and a self or agent with a conditioned spontaneity. Hence, if we think there is conditioned spontaneity, we must think that there is non-predetermined determinism, and vice versa.
So the answer to our third question must be this. The scholastic conception of free will, unlike the Kantian one, is compatible with determinism but, like the Kantian, is incompatible with predeterminism. Therefore, it is compatible with science. But it does not presuppose that science is the only nor even the best way to describe or understand the world. It can be accepted by both dualists and monists. This independence of an influential metaphysical watershed is another point in the list of its virtues.
Let me condense the lessons of history of philosophy on free will to four theses:
1. Free will is the capacity not of initiating something absolutely new but of acting after deliberation.
2. Acting after deliberation is acting on reasons.
3. When acting on reasons, we are determined by causes that we can replace by other ones, even by causes that did not belong to the original potential causes deliberated about, e.g. by the wish to prove our freedom.
4. When acting on reasons, we are determined by reasons without being predetermined by them.
 Kant, Kritik der reinen Vernunft A 445, B 473: „eine besondere Art von Kausalität, nach welcher die Begebenheiten der Welt erfolgen“.
 ibid. A 534, B 562: „eine Kausalität, ... eine Reihe von Begebenheiten ganz von selbst anzufangen“.
 ibid. A 445, B 473: „ein Vermögen, einen Zustand, mithin auch eine Reihe von Folgen desselben, schlechthin anzufangen“.
 ibid. A 533, B 561: „eine Spontaneität, die von selbst anheben (kann) zu handeln, ohne daß eine andere Ursache vorangeschickt werden (darf), sie wiederum nach dem Gesetze der Kausalverknüpfung zur Handlung zu bestimmen“
 David Hume, A Treatise of Human Nature bk 2, pt 3, sec 1, ed. Selby-Bigge, Oxford (Clarendon) 1978, 404
 Aristotle, Metaphysics Λ 1071b 3 – 75a 10
 Kant, Kritik der reinen Vernunft A 536, B 564: “Wirkung in der Welt … aus Freiheit“, and ibid. A 533, B 561.
 Luis de Molina, Liberi Arbitrii cum Gratiae Donis ... Concordia, Antwerpen 1695 (Lisbon 1588), quaestio 14, art. 13, disput. 2, S. 8 (my translation). For the scholastic conception of proposition and will, cp. Anthony Kenny, Will, Freedom and Power, Oxford 1975.
 E. Tugendhat, Vorlesungen zur Einführung in die sprachanalytische Philosophie, Frankfurt (Suhrkamp) 1976, 110.
 J. Habermas, Theorie des kommunikativen Handelns, Frankfurt (Suhrkamp) 1981, I 370, vgl. II 113f.
 Thomas Hobbes, Leviathan ch.21, ed. Macpherson (Pelican) 1968, 262f
 Th. Hobbes, Of Liberty and Necessity. English Works ed. Molesworth vol. 4, London 1840, 253
 Aristotle, The Nicomachean Ethics III 1111a 26, transl. H. Rackham, London 1962 (Loeb Library).
 ibid. 1110a 17
 ibid. 1112a 33
 ibid. 1110a 17. Rackham translates incorrectly “power to do it or not”. Aristotle curiously gives the description when he discusses not the deliberately chosen, but the voluntary. But if Aristotle was consistent, he cannot have wanted to say that all voluntary actions spring from the “power to do it and not to do it”. He wanted to say that only deliberately chosen actions spring from it.
 ibid. 1109b 35
 John Locke, An Essay concerning the principles of human understanding bk 2 ch 21, § 47; ed. Nidditch, Oxford (Clarendon) 1979, 263
 ibid. bk 2 ch 21 §23; p. 245.
 G.W. Leibniz, Nouveaux Essais, on Locke, Essay bk 2, ch 21, § 49; ed. C.J. Gerhardt, Die philosophischen Schriften, vol. 5, Berlin 1982, 184.
 René Descartes to Father Mesland, February 9, 1645; Œuvres de Descartes ed. Adam et Tannery vol. IV, 173
 G.W. Leibniz, loc.cit. on Locke, Essay bk 2, ch 21, § 25, p. 168; this and the next translation of Leibniz are mine.
 Leibniz ibid. p. 84f: „A mind that would have the property of willing, and being able, to do and will the opposite of what can be predicted of him by whomsoever, belongs into the set of entities that are incompatible with the existence of the omniscient being, i.e. with the harmony of the things, and hence have not been nor are nor will be.”
 Lucretius, De rerum natura, bk 2, ll 251-7: „Again, if movements always is connected, New Motions coming in from old in order fixed, If atoms never swerve and make beginning Of motions that can break the bonds of fate And foil the infinite chain of causes and effect What is the origin of this free will Possessed by living creatures throughout the earth?” (transl. R. Melville, Oxford UP 1997)
 Cp. Descartes’s allusion to Medea in Œuvres ed. Adam et Tannery vol. IV, 174, and Ulrich Steinvorth, Freiheitstheorien in der Philosophie der Neuzeit, Darmstadt (Wissenschaftliche Buchgesellschaft) 1994, 48
 The notion of a
gap has been used by John Searle in his Rationality in Action,
Cambridge/Mass., 2001. Although I share his criticism of deterministic
models of action, I do not share his conception of free will.