Distant Causation and Compatibilism

The problem that the funny mind control cases illustrate is basically this: compatibilists say that free will is compatible with determinism and/or ultimate outside causation of our inner states. But there are SOME forms of ultimate outside causation/determination of our inner states which are clearly not compatible with free will in any sense; the fantasy/science fiction mind control cases are examples of such.

So the compatibilist has to either offer a distinction between 'legitimate' and 'illegitimate' ultimate outside causation of our mental states to handle these cases or else she has to show that the kind of mind control in question is logically impossible. The latter seems doubtful; what about the former?

The only move the compatibilist seems to be able to make here is some kind of 'proper function' theory of the mind, whereby a properly functioning mind does/can have freedom of the will, whereas a disturbed/broken/stuck one doesn't. We'll talk about these theories in class - there are a few out there already, though I don't know if they are motivated by mind control considerations or not. But the idea is that addiction, compulsion, mind control, etc. interfere with the mind's proper function and in so doing make it un-free.

External Control

The idea that if we identify with something we can't be unfree with respect to it is interesting, but I thought of an example that may show it to not be quite correct.

The Stoic or Spinozist may want to identify with the universe as a whole, and by so doing they may in a certain sense increase their freedom, or even gain freedom, with respect to it. But it seems important even in this case that there is a free individual acquiescence to nature.

Consider the following case: Brakmorn loves the sorceress Xanthippe with all his heart, and she loves him in return; their purposes are intertwinable, and each loves the other so thoroughly that every object of desire of each is an object of desire for the other, reinforced by second-order desires to make the others' desires desires of their own.

Xanthippe wants to gift Brakmorn with a magical sword from an ancient tomb, a sword Brakmorn himself has long coveted. Only Brakmorn's body has the physical strength to loot the tomb, but without Xanthippe's knowledge of sorcery, he won't be able to defuse the ancient lich-king's magical traps.

So, Xanthippe possesses him and uses his body to navigate the tomb's dangers, in the usual science-fantasy way.

Now: Brakmorn wants the sword, wants Xanthippe's help, wants to be mind-controlled by Xanthippe, etc. One may suppose that they are as one in love and unity of will and that there is no respect in which their desires diverge from one another even slightly.

But in spite of all this, Brakmorn does not act freely when he gives himself over to Xanthippe's sorcery to be mind controlled. He gives up his freedom freely and wholeheartedly, but he still gives it up, and thus is not free.


It seems that whatever our theory of freedom involves, we need some kind of condition ruling out external control.

This, in turn, seems like it might be a problem for compatibilism in general, insofar as determinism looks a lot like external control. Why is Xanthippe's sorcery out if physical laws are in?

One way around this which has been explored in the literature is a proper function condition on the brain/mind. (I'll have an article on this for the class a little bit later.) That is, one might hold that a mind appropriately situated in its environment and properly functioning is free, but that OCD, drug addiction, or Xanthippe's sorcery interfere with that proper function, and thus undermine freedom. I don't think proper function is sufficient for even compatibilist free will, but I do think that external control cases might show us that it's a necessary condition that needs to be added on.

Assuming it works to overcome them, which I'm not sure it does.

Another tactic one might take is to argue that 'the idea of mind control' presupposes libertarian free will so that the fantasy/science fiction cases aren't relevant. You never know I guess.

Morality and Libertarian Free Will

This is just a 'how things fit together' note, but I wanted to get it up here.

Here is one dilemma that faces the partisan of free will:

1. If our will is determined by (reasons, beliefs, desires, etc.), then it is not free, because something prior to it determines it.

2. If our will is undetermined by (reasons, beliefs, desires, etc.), then it is not free, because acting on the basis of anything besides (reasons, beliefs, desires, etc.) is senseless/pointless/irrational, and any freedom of will worth wanting requires rationality.

3. Our will is either determined or undetermined by (reasons, beliefs, desires, etc.).

Conclusion, as usual: There is no free will.

Now, one option is simply to deny (1), and say that if the reasons/beliefs/desires etc. are 'ours' in the appropriate sense, acting on their basis is too free action. This response is most easily articulated within a compatibilist framework; it seems to a first degree of approximation anyway to be Frankfurt's answer.

The denial of (2) is also possible, and I take it that this may be where the Kant/Campbell/Wolf idea that morality is a necessary condition of free action comes in. That is, let's say that we have this completely undetermined ability to choose that the libertarian wants. Aren't its movements effectively arbitrary, by definition?

Certain conceptions of morality, particularly deontological ones, seem to provide a possible 'no' answer to this question. You can assess your situation, determine what morality dictates ought to be done in it, and then will that, entirely indepedently of the actual reasons/beliefs/desires you had going into the situation. You subordinate your will to the moral law and act in accordance with it, and in this you show your freedom.

It is actually rather hard to understand this response. First of all, the person who acts this way is effectively sublimating their beliefs and desires to the moral law - they would seem to have to want to do good in order to act on this basis. So aren't they actually just showing that the moral law is for them the kind of reason/belief/desire etc. that it is not always for everyone else? And isn't their will then in this case also fundamentally un-free because it is being determined by prior reasons, etc.?

I think the idea here has to be that, if we do indeed have an undetermined ability to determine what we do irrespective of our own reasons/beliefs/desires/etc., the only way to avoid arbitrariness is to find a principle for action that would apply to everyone in that situation. (Kant uses this kind of formulation all the time, of course.) We may make our action depend on anything we like and of course we may fill it in directly with our actual current beliefs and desires, irrespective of the moral law. Some people arguably never do anything else. But if there are general moral laws which govern our situation, following them is not arbitrary, just because they are the laws that everyone ought to follow in those situations. Thus in following them, I think the thought must go, our hypothesized libertarian free will can be undetermined without being arbitrary or irrational, because there is a rationality beyond that of satisfying our own reasons/beliefs/desires out there upon which we can act.

It is often said that we need free will for morality to make sense. This line of thinking is interesting because it seems to suggest we need morality in order for free will to make sense.

A Note on Wholeheartedness

In “The Faintest Passion”, Harry Frankfurt wrote: “It is a necessary truth about us, then, that we wholeheartedly desire to be wholehearted.”

Wholeheartedness has long been on the list of proposed virtues, with supporters as diverse as Augustine, Kierkegaard, and Nietzsche. We are more accustomed to being exhorted to it: “Leave off your wretched ambivalence! Put aside distractions! Tend to the one thing needful!”

But if Frankfurt is right these exhortations are in a way beside the point. If we necessarily desire to be wholehearted, then we should not need exhortation. Perhaps the exhortations are merely reminders; but it might nonetheless strike one as odd that some respond to them with ambivalence, or at least seem to. How can we be ambivalent when reminded of our own desires, desires that we cannot fail to possess?

Ambivalence Frankfurt defines as follows:

Ambivalence is constituted by conflicting volitional movements or tendencies, either conscious or unconscious, that meet two conditions. First, they are inherently and hence unavoidably opposed, that is, they do not just happen to conflict on account of contingent circumstances. Second, they are both wholly internal to a person’s will rather than alien to him; that is, he is not passive with respect to them.

The second condition will be trivial if one counts all a person’s desires and beliefs as internal to them; Frankfurt needs it because he does not. It also seems clearly right, though, if we follow Frankfurt in positing an ‘inner’ or ‘true’ self (not his chosen terms) which consists of the beliefs and desires with which one is identified: if there is such a self it is the one whose ambivalence concerns us. So either way this condition should be granted.

The first condition is too weak, however. Consider the following passage from William James’ Principles of Psychology:

With most objects of desire, physical nature restricts our choice to but one of many represented goods, and even so it is here. I am often confronted by the necessity of standing by one of my empirical selves and relinquishing the rest. Not that I would not, if I could, be both handsome and fat and well dressed, and a great athlete, and make a million a year, be a wit, a bon-vivant, and a lady-killer, as well as a philosopher; a philanthropist, statesman, warrior, and African explorer, as well as a 'tone-poet' and saint. But the thing is simply impossible. The millionaire's work would run counter to the saint's; the bon-vivant and the philanthropist would trip each other up; the philosopher and the lady-killer could not well keep house in the same tenement of clay. Such different characters may conceivably at the outset of life be alike possible to a man. But to make any one of them actual, the rest must more or less be suppressed. So the seeker of his truest, strongest, deepest self must review the list carefully, and pick out the one on which to stake his salvation. All other selves thereupon become unreal, but the fortunes of this self are real. Its failures are real failures, its triumphs real triumphs, carrying shame and gladness with them. This is as strong an example as there is of that selective industry of the mind on which I insisted some pages back. Our thought, incessantly deciding, among many things of a kind, which ones for it shall be realities, here chooses one of many possible selves or characters, and forthwith reckons it no shame to fail in any of those not adopted expressly as its own.

Many of these do in fact conflict because of contingent circumstances: the fact that the active part of most human lives is rarely more than 80 years, and very often much less, is in fact the chief driver of the supposed conflict here. There are imaginable cases of genuinely necessary conflicts of this kind in a human life – a person who genuinely desires both to be a lifelong meditative ascetic and to star in pornographic films, say – but these are mostly not the important kinds of ambivalence most of us face. We are ‘forced’ to choose, in whatever sense that word is appropriate in cases like these, not by necessity logical or metaphysical or physical, but simply by the dumb statistical unlikelihood of being able to do something worthwhile along more than one or two lines in the limited amount of time we shall have if we are lucky.

But I want to grant the idea that ambivalence is concerned with desires that necessarily conflict regardless, for the sake of argument, because I think that even with this artificially strong condition in place, Frankfurt’s claim does not follow.

What is the argument? Frankfurt writes: “Is it possible to be satisfied with ambivalence? A person may certainly come to accept the fact that he is ambivalent as unalterable. It seems to me, however, that it is not a fact with which he can possibly be satisfied. No one can be wholeheartedly ambivalent, any more than someone can desire unequivocally to betray himself or to be irrational. That someone accepts his ambivalence can only mean that he is resigned to it; it could not mean that it satisfies him. Perhaps conditions are imaginable in which a person might reasonably regard ambivalence as worthwhile in order to avoid some even more unsatisfactory alternative. But no-one can desire to be ambivalent for its own sake.”

Why not? The idea here seems to be that

-         To be ambivalent is to desire x and desire y even though x and y cannot both be achieved.

What does a person gain and lose by abandoning either the desire for x or the desire for y? Their gain is that they no longer must, of necessity, have at least one desire that will not be satisfied. Their loss, though, is that now there is only one thing which can satisfy them, whereas before they had two, of which either one would have given them at least some satisfaction.

Is having a desire unsatisfied worse than not having the desire at all? I am not sure. If we weight these two things equally, though, it seems as though either one is OK. That is, we have x & ~ y, ~x & y, and ~x & ~y as our three possibilities. The wholehearted person yields satisfaction in the first case and not in the other two; the ambivalent person receives a half-yield of satisfaction in the first two cases and none in the third.

With no information about their concrete circumstances, what can we say about this?

If we value satisfaction and don’t care about frustration, the ambivalent person is actually better off: he has three possible outcomes, two of which pay off and one of which doesn’t, whereas the non-ambivalent person has two outcomes, one of which pays off and one of which doesn’t. The expected yield of ambivalence is 2/3, as opposed to the 1/2 which comes from wholeheartedness.

If we only care about frustration, then Frankfurt is right: the ambivalent person will always feel some frustration, whereas the non-ambivalent person will only feel it half the time.

If we care about both, then the non-ambivalent person is better off, but only slightly so. His average satisfaction-frustration is 0, whereas the ambivalent person’s is -1/3. With no information about the situation at hand, then, I suppose we can conclude that a person who cares both about having desires satisfied and not having them frustrated is better off betting on wholeheartedness than on ambivalence.

Is there some a priori reason to suppose that ambivalence will tend to skew the probabilities here? One could argue, for example, that more desires mean more directions for using energy and therefore less likelihood of getting any of the things one wants. And there would be some truth to such a supposition in many real-world contexts. But not all: for example, young persons, who are unclear about their actual abilities, often learn what things they are good and which experiences they actually value by pursuing a variety of different goals, some of which may be incompatible. For a young person, then, a certain degree of ambivalence might actually make long-term satisfaction more likely, inasmuch as making first stabs at satisfying many different desires might help to teach her which ones she was likely to be able to satisfy and which ones had component acts which were satisfying to her in their own right, as opposed to perhaps painful intermediate steps to a prima facie satisfying long-term goal.

I think there is more to be said about this, and this is all I have time for for today. But it does seem that if we want even necessarily incompatible desires to generate a desire for wholeheartedness, we need to say something about the value of satisfied and unsatisfied desires. For now we can at least say that without that, there is no theorem here.

Naturalized Libertarian Free WIll - What You Need

This is the list from class, reproduced for convenience. If anyone sees things that need to be added or has questions, let me know!

(1) Object-causes which (a) do not reduce to event causes, and (b) which have originating events 'in time'.

(2) An account of objects taking on states which keeps the object which takes on the state as an object-cause of the state. (This might be easy, actually, since an object's taking on a state (e.g. putting your bike into 3rd gear) clearly involves the object, and if you have (1) and (3) and (4) worked out that's probably all you need to get this too. But it seems as though you might need it if you think that human beings aren't 'brute' causes so much as beings that cause things by getting into particular states).

(3) There are 'emergent' objects which satisfy (1) and (2) and which do not 'reduce' to their constituent parts, because they introduce 'new causal powers' into the world that cannot be fully analyzed in terms of the causal powers of their parts.

(4) Persons are objects of type (3).

(5) 'Willings', 'intentions', 'tryings', 'attemptings', or some subset and/or aspect thereof - the agent-component of actions - are among the states of type (2) that persons take on.

It seems as though if you have these five things you have the best an emergentist can do for naturalized free will. Human beings bring new causal powers into the world which do not fully depend on their constitution and only depend on what has come before them in the sense that whatever assembled the persisting emergent physical structure which leads them to emerge was the cause of their coming-to-be, but not (somehow - this is the point pressed earlier) of their actions.

So then, can we get (1)-(5)? And if we do, is this free will, or is there still something missing? What is it?

Side Note - Davidson's Art

Donald Davidson in "The Material Mind" discusses the case of one Art, a synthetically constructed human being, who is ex hypothesi exactly like one of us in all psychologically relevant respects, and suggests that we should/would not deny such a person title to free will just because their origin was entirely determined by outside processes. Davidson thinks that the possibility of Art does something close to providing a knock-down argument for Compatibilism, as far as I can tell. I agree that there is a strong compatibilist intuition that Art pumps in us, but it would be nice to have the argument a little more precise than Davidson makes it (or than I made it in the 'quantum sperm' example).

As was pointed out in class, if we make an exact copy of a person, including whatever gives him free will, then trivially the exact copy has free will. But an exact copy of e.g. a dualist self would include a duplication of his soul. Maybe this doesn't matter; but we don't our claim to come to nothing more than "If you make an exact copy of someone who has free will, the copy will have free will", because the worry is that the result is then trivial because the free will itself got copied.

I think if we assume some sort of underlying physical substrate, as Davidson does, the argument gets clearer.

1. If we make an exact physical copy of a physical being, that being will have all the mental properties (if any) of the being thus copied.

2. Free will is a mental property of some physical beings.

3. Therefore, if we make an exact physical copy of a physical being who has free will, that being will have free will.

4. Exact physical copies may be produced deterministically or indeterministically, naturally or artificially, and with or without purpose or guiding intelligence.

5. Therefore, beings who have free will may be produced in deterministic or inderrministic ways, naturally or artificially, and with or without purpose or guiding intelligence.


You pretty much have to either be a dualist or claim that actual histories are essential to the nature of physical beings to deny 1 and 2; worries about psychological structure and/or physical law don't really matter to endorsing the conclusion.


So if you don't take one of those approaches, this argument seems to show that the actual causal history of organisms is irrelevant to whether or not they have free will. So there's a sort of antinomy here between this argument and arguments that distant causation undermines free will. Which is what we were trying to get to earlier.


A note about anomalous monism. The above considerations seem to show that there must be at least one psychophysical law if every event is (among other types, perhaps) a physical event: namely, every exact copy of the entire physical universe is also an exact copy of the entire mental universe. Furthermore, if we accept various sorts of commonly accepted views to the effect that minds are local and discrete - that mental properties supervene, even if only in a token-token way, on physical properties - or perhaps that events occurring within a particular mind supervene, even if only in a token-token way, on physical events occurring within that mind (or even within that mind's backwards light-cone) - then it seems that we have a second psychophysical law as well, namely, that every exact copy of the relevant physical organisms (perhaps, plus their environments) is also an exact copy of the minds that go with those organisms.

I do not see how to escape this, and Davidson's intuitions about Art actually require it - otherwise we can simply shrug and say that the total lack of psychophysical laws makes it completely unsurprising that we have free will (if we do) and this Art who is identical to us but who came to be in a different, fully deterministic way is not - even if he is an exact clone.

I don't think this is a special problem for anomalous monism, though - the view is still easily stated as claiming that there are no type-type mental-physical superveniences, or as Davidson put it no psychophysical laws, other than the laws of identity stemming from events under multiple descriptions and sets thereof being the same events and/or sets of events.

'Naturalized' Libertarian Free Will - The Basic Idea

This is a summary of our last class, with details for the future. If people want to work on this for their final papers, let me know; if a lot of you do we will go into it more later, but for now I think we should keep canvassing different views and get the compatibilists their fair say.

O'Connor made the intriguing suggestion that 'emergent' selves might serve as plausible loci of agent-causation within a naturalistic worldview. The idea here can perhaps be drawn out by parallel to dualistic free will. For the dualist, the non-material mind is automatically 'on another level' or 'outside the physical universe' in important ways; so it can be an uncaused cause of things within the physical universe. Aside from worries about causal interaction and unity, though - which are not intrinsically 'physicalist' or 'naturalist' in character - the dualist picture also seems to simply push the questions back to another level. What creates the non-material mind? Is it determined by prior non-material causes, or is it free? In what does its freedom consist? If the non-material mind has beliefs and desires, do those determine its actions? (The analogy here to questions about what is the ultimate cause of God if God is the ultimate cause of the world is exact, in my view.) I don't say that these questions are insoluble, except perhaps the causal unity issue - which I have never been able to see my way around - but they are serious.

The 'emergent self as the locus of free will' approach gets around the hard questions the dualist must answer by making the 'other level' within the same causal network as the other stuff. Emergent selves would be  entities which are made up of or constituted by ordinary physical stuff (atoms, genes, etc.) but whose persistent structural organization introduces new causal powers into the world which do not reduce to the causal powers of their constituent parts. Such selves would be constitutionally dependent on the way the world is, but not causally dependent on the way the world is, because the actions they caused would be traced back to them rather than to their various constituents.

The defender of this view still has to answer the 'Propson objection', namely that if these emergent selves have something like a particular self-character that emerges from particular details of their structure, then even if the selves bring about new causal powers, it will be plausible to suppose that those new causal powers depend on prior causes which led to the whole to be organized in precisely the way it is. Further, denying this would seem to entail that the 'self-structure' must be shared among all human selves - but is it? I suppose there's a line of thought through Husserl and Heidegger that might at least offer some support to this claim, but the idea of structural phenomenological invariants of human experience is controversial, to put it mildly. Another possibility would be quasi-Kantian in nature: there is some kind of transcendental structure of selfhood which is filled in through our particular history and decision-making but which is nonetheless somehow prior to it. This is going to be hard, but I think it would be interesting to see how the going goes.

Also, though, there are going to be questions about emergence that the dualist does not have to answer, as well as questions about object-causation and the metaphysics of personhood that the dualist will also have to answer. In the next post I'll put up a list of things that I think the naturalized defender of libertarian free will has to provide us to make out a case for naturalized libertarian free will.

A Possible Picture of Object-Causation

Consider the imaginary substance indifferentium, which always comes in uniform masses with no internal structure whatsoever. At some temporal intervals, regular or otherwise, indifferentium emits small particles called quiddons. We call this an emission because the quiddon forms at the surface of the indifferentium fragment and proceeds outwards from there. Whether the quiddons are made of indifferentium or not does not concern us, and we may specify further that neither mass nor momentum nor energy are conserved over quiddon-emission events. (This condition is there so that the mass of indifferentium emitting the quiddon will be exactly the same in every respect before and after the emission-event. Probably looser conditions of object-identity would allow for conservation principles to hold, but since we are trying to establish a conceptual point about causation we will keep everything exactly the same.)

What in this scenario is the cause of quiddon-emission events?

Ex hypothesi there is no event within the indifferentium mass which proceeds the emission. There certainly could be such: a buildup of etheric fluid meeting some critical threshold and coalescing around the indifferentium to form a new quiddon. But this is not the case; the case is just that, periodically, indifferentium masses emit or shed or throw off quiddons – they form at the edges of indifferentium masses and go their way – and there is no other event, process, etc. to be found proceeding this one.

This substance seems conceivable, and therefore logically possible. Indifferentium masses emit quiddons. But what is the cause of their so emitting?

We might specify the coming-to-be of the indifferentium mass itself as the event which causes it. But we can imagine cases where this sort of answer would not be available. Imagine universes with circular time with chunks of indifferentium that persist unmodified for the entirety of it; universes with time extending backwards infinitely, ditto; and perhaps even universes like ours, if certain Leibnizian interpretations of time offered by relativity theorists are correct, where time itself begins with certain initial material interactions, and where among the materials present at the very outset (and thus, in effect, ‘eternally’) are persistent indifferentium chunks. So there may be no prior event to consider in such cases.

If all this is indeed conceivable, the indifferentium’s emission of quiddons is very like something coming from nothing, in that there does not appear to be any prior event that causes it. But there is a difference, in that here something is coming from something.

Consider voidons, particles emitted spontaneously from the void with no prior event causing them. The void’s emission of a voidon is an event, but has no prior event-cause, and no prior object-cause, since the emission comes from the void.

When indifferentium emits a quiddon, we still may have no prior event-cause, depending on how and whether chunks of indifferentium come-to-be, but we do now have an object-cause: the indifferentium mass that emits the quiddon.

There hence seems to be a place in our conceptual scheme for object-causes that don’t entirely reduce to event-causes. This place is fairly clear when we have eternal objects, but even with temporal objects, there seems to me to be a point to talking this way. For suppose that indifferentium masses come to be and pass away. Even in this case, there is a great difference between events like the emission of a quiddon, whose prior event-cause is the creation of the object which emitted it, and events like breaking a window, whose prior event-cause is some other event in the window’s vicinity.

Of course all events causally depend on the existence and thus coming-to-be of the objects which feature in those events. But many events causally depend also on sequences of events involving those objects, and here we don’t have that.

We can attempt the following definition (though one can see various problems): x is an object-cause of e iff x is a component of e and x exists prior to e occurring. On this definition we can ‘pick out’ object-causes in series of event-causes just by tracing the object through the chain. But by the above argument, there are conceivable (therefore logically possible) events which have prior object-causes but need not have prior event-causes. Therefore, although the notion of an object-cause is (and must be, if it is to be relevant to our universe) intimately connected to that of an event-cause, it is false to hold that there is nothing to be said about causation that can be said in terms of event-causes alone. For there are conceivable events which have object-causes but no event-causes.

1. If an agent has free will, then a being identical to that agent with a different causal history has free will.
2. Beings identical to agents with free will with different causal histories are possible.
Therefore, if there are agents with free will, it is possible for there to be agents with free will with different causal histories than them.

This argument seems to suggest that the causal histories of free-willed agents are irrelevant to their having free will.

What should we make of that? We could deny premise 1; this would amount to saying that the causal history of a free-willed agent is an essential feature of such agents, I think. Is this true? I am not sure. I have some inclination to think for example that if I have free will, a being identical to me but produced ex nihilo by a capricious deity would also have free will. That being would have the same subjective states, the same beliefs, desires, hopes, plans, and cares, a materially indistinguishable body, etc.

It's true that some writers make causal histories essential features of agents: for instance Kripke claims that if the actual Queen Elizabeth was born of two parents, it is an essential feature of her that she was born from those two parents and not others; she would not be the person she is without that specific biological organism. So one might claim that it's essential to being the self one actually is that one has a particular causal history of some kind, and then we would deny (1) by saying that two qualitiatively (and physically) indistinguishable persons might nonetheless in one case have and in the other case lack free will just because of their difference in origin. This seems wrong though, both psychologically and in terms of basic modal intuitions. The psychological wrongness comes from the fact that if we have freedom we ought to be able to exercise it through choice, reflection, intelligent action, etc., and qualitatively indistinguishable persons would have the same psychology even if divergent causal histories converged on those persons in the past. Modally speaking, imagine that at the moment of your conception, a quantum mechanical mishap caused your father's sperm to be obliterated, but as a matter of freak chance created a bit of sperm with exactly the same DNA, etc. inside of it, which conceived you. Everything else proceeded in exactly the same way it has and there is now a person qualitatively identical to you with the same DNA as you raised by the same people in the same society etc. except that this person is actually a product of immaculate quantum-mechanical conception. Is there any reason to think that there is a difference between you and this person with respect to free will?

(2) likewise seems hard to deny, unless one thinks everything proceeds by necessity, or that causal histories are essential to agents. A pyramid made of three bricks arrayed in exactly the same way seems to be the same pyramid whether the bottom level went left brick - right brick or right brick - left brick. As long as you wind up with the same structure it's, normally we think anyway, the same pyramid.

Another way one could try to go after the premises is to allow that different causal histories are possible but to restrict them to the right kind of causal histories: if an indeterministic history e.g. is essential to the free-willed agent, than only other indeterministic histories yielding qualitatively indistinguishable beings would yield free-willed beings, or something like that. But why think this? Again, does it really make a difference to a person's psychology or freedom what the inputs going into that person were?

If it does not, and something like this argument goes through, this is good news for the compatibilist. For it's easy to suspect on the basis of the conclusion above that

3. Whether or not a being has free will does not depend on the causal history of that being.

which, to the degree that it has the status of an 'intuition about free will' or whatever label we like to put on philosophical premises that we imagine to be grade A in character, directly contradicts the one I have claimed the traditional libertarian defender of free will wants to assert, and which the problem of distant causation seems to bring up more generally. That is, both

4a. Whether or not a being has free will depends on whether that being is not always (or, necessarily?) ultimately entirely controlled by events external to that agent, and

4b. If an agent has free will, that being must not have behaviors and mental states that are distantly caused/determined/controlled by factors outside that agent.

seem to entail

4. Whether or not a being has free will depends on the causal history of that being.

and (3) and (4) can't both be true without some disambiguation (depends in this sense, does not depend in this other sense, etc.).

Any thoughts?

Is There a Genuine Contradiction in the Notion of Free Will?

The problem with introducing a fundamental indeterminism into the self/agent/mind is that while such beings may be causes of their actions, even ‘first causes’ in some fairly deep senses, they are not controllers of their actions. This may be conceptually forced on us – if ‘controlling’ implies determination at any level, and we are forced to believe in an indeterministic determination in order to believe in free will, then we are in some trouble here. Kane’s often thoughtful discussion of responsibility is in a certain way an attempt to paper this problem over; I think that deeper consideration is showing that one can be a responsible cause (at least under certain assumptions) without therefore being a controller and without therefore being a free-willed agent in the full traditional sense, if we are right about what this sense entails.

Let’s then put indeterminism to one side and consider what the prospects for a determined self are. The difficulty here as I see it is that if the elements of self are determined by prior elements, we seem to be forced into the familiar choice between absolute origins, circularity, and infinite regress.

Infinite regress might work for God’s freedom of will, if there is a God. Could it work for human beings? Could we somehow be at the cresting edge of an infinite regress of causal control with no point of origin, all within us? This would be an interesting line to take. I have trouble seeing how it would work out at this point.

Circularity, where our will is somehow made free by itself, likewise seems more the province of a deity than of human beings, if it is possible. Still this idea resonates with certain themes in the history of philosophy, and perhaps if free will is internal to some kind of ‘performative’ rather than ‘substantive’ self – a self which emerges through actions (though what is acting in such cases?) rather than a self which is prior to its actions, say – the circularity option might be made out.

An absolute origin in space and time seems to be the more natural, pretheoretic assumption about selves/agents/minds of the sort many of us normally take ourselves to be. If I am perhaps being too open-minded about the above options, it’s because this one does seem to bring us pretty close to impossibility. For we seem to have two choices: either the self/agent/mind is created at least in small part ex nihilo, or in some way which is radically causally independent of whatever exists prior to it, or it is caused to be wholly by things which came before it. In the latter case ultimate control seems necessarily to be beyond the self; the self’s internal determinations will always trace back to things that came before it. What about the former? A character appears, or a will, in some way untouched by the things before it: it is an emergent whole with new causal powers, or an extradimensional self which reaches into this reality from another one. Is such a being free (at least if it has the right kind of will, as we do)?

It’s not clear to me that it is. Here’s an argument. Assume a world of believing-desiring beings who are as like us as beings can be without free will. Let’s say that, just before one such being is born, or at just the stage of development that ‘personhood’ is going to emerge, or whatever, there is a quantum-mechanical hiccup in the universe that interrupts the normally deterministic processes of believing-desiring being production with an initial indeterminism. This being, and this being alone, has a character which was not determined by what came before him in the universe. Maybe he even wears plaid with stripes. Is he free, if the others are not? It seems to me that he is not. The fact that a random quantum mechanical accident introduced an indeterminism into his nature and thus rendered him causally independent of the universe that came before him does not seem sufficient to make him free if the others are not free.

So the argument is: if the others are free already, then the introduction of a fundamental indeterminism to cut this man off from the external causal sources of his selfhood is unhelpful. And if the others are not free already, then adding the fundamental causal indeterminism doesn’t seem to make a difference which really matters. So a causal indeterminism which cuts agents/selves/minds off from their origins, however it does it, doesn’t seem to get us free will.

Likewise, consider two beings with an identical psychology, one of whom was created ex nihilo and the other of whom was created through deterministic causal processes. Is the fact of origin sufficient to make one free and the other not?

I am not sure these arguments are conclusive. What we wanted to know was whether beings like us could have free will and we thought that, in order to have this, we had to show that we were not in fact completely causally dependent on things external to us. Now we are presenting some alternate arguments to the effect that two identical beings could be produced by completely different causal processes (standard issue, indeterministic randomness, miraculous ex nihilo) and that this shouldn’t ultimately matter to whether they had free will or not. It seems to me that you can’t have it both ways; as of now though I am very sympathetic to both of these lines of thought.

It would be very nice if we could make both lines of thought a little more precise. But anyway, to conclude the line of thought on external determination for now, it seems as though what the defender of traditional free will would want would be a self that was (a) a determiner of its own actions but (b) was not ultimately entirely control-dependent on things outside itself. So introducing some sort of indeterministic or ex nihilo element into the events which create a self, but not to the events within a self, would seem to be the best one can do here. A non-physical, dualistic faculty of choice which determined one’s attempts to act would be one way to get this, although just as when we specify God as the creator of the universe we may wonder who created God, we may also wonder what the ultimate origins and causal sources of this non-physical mind are. (How do we know we have not taken it out of one causal order only to put it in another?) O’Connor postulates an emergent self as perhaps another, although DP raised an interesting objection to this line of thought last class.

Can we generate a straight contradiction/antinomy out of the free will problem? This morning I’m inclined to think that if we rephrase things just slightly we can, but I don’t have it yet. There are loopholes (which we are exploring in the class!) in the traditional arguments against free will, but maybe they can be closed by rephrasing things a bit. I think that discovering some genuine contradictions in the neighborhood of the traditional free will problem would help us get clearer on the notion, whether they undermined free will in some broader sense or not.

The Antinomy of Free Will

I think a sort of converse of the Dilemma of Determinism highlights the degree to which the self is central to free will debates. I am using 'caused' rather than 'determined' here, however; probably the D-word can be substituted without loss if desired. Consider the following:

1. If an agent is to have free will, her actions must be caused.
Argument: Free will involves causal control of one's own actions; a free willed agent must be able to cause his or her own actions.

2. If an agent is to have free will, her actions must be uncaused.
Argument: Free will involves causal control of one's own actions; a free willed agent must not be totally constrained by things external to him or her.

If we take (1) and (2) straight we get an antinomy - but it's not much of an antinomy, since the arguments for the opposing pair suggest exactly where the issue is. Free willed agents or selves (on this understanding anyway) must be causes of their own actions without having been caused to act in those ways by anything outside of them.

If you look at the problem in this way, it's possible to see various forms of compatibilism as 'widening the gap' between the inner and outer causes of behavior in various ways. Hume (on one reading) is the narrowest gap possible: the outside causes form us, and our inner states cause our actions, so 'we're both free and determined'. Frankfurt widens the gap by introducing second-order desires, which create more internal system complexity. The idea of widening the gap also might be used to characterize some of Dennett's ideas about freedom.

What the libertarian wants is a self in which the gap has been widened to infinity, in a certain sense.

Following out this line of thought one might explore a couple of things:

(a) There's a sort of 'game of freedom' approach possible here which might allow compatibilism to approximate libertarianism in the same way that the infinite limit of discovery by a rational community of inquirers is sometimes used to approximate truth. In other words, one might suggest that while the freedom we _have_ is always compatibilist in character, it is always open to us as human beings to widen the gap farther by introducing additional complexity and/or higher-order operations to our mental life; to abstract. This approach might also be used to connect freedom to mind in other senses, since the human ability to abstract is often taken to be pretty central to our intelligence, self-consciousness, rationality, etc. by various authors.

(b) Kane's view, or even a mini-Kane view without the psychological elements (SFAs), seems like an important test for revealing what else the libertarian might need. The reason for this is as follows: let's say that there's an indeterminism in the agent which leads somehow to the agent's actions. If we allow indeterministic causation, then the agent can cause his or her actions by virtue of such an indeterministic state. And since it's an indeterministic state, the actions are not caused by anything outside the agent. Therefore, the mini-Kane self counts as a free self in the sense of resolving the tension between 1 and 2 above. Is it a free self? In effect Kane adds a bunch of compatibilist-style psychology to this basic mini-Kane position to make it more palatable - but I'm not sure that stuff is what's central in the end. What's central is whether containing a locus of indeterminism within ourselves is really a source of freedom in the end. Things like this have been thought before - the idea that we discover our freedom in madness, Dionysian spontaneity, Kierkegaardian irrational leaps of faith, whatever - but do we really want to accept them? It seems like we need an argument for or against. Van Inwagen offers the 'replay argument' against - should we be persuaded by that or not?

Thoughts?

Metaphysical Aside on Substance Causation in General

Whether or not 'agent-causation' as distinct from 'event-causation' is comprehensible is supposed to be a problem. Some think that what is puzzling about it is not something to do with agents, but something to do with 'substance-causation' more generally. Randolph Clarke develops this theme at some length in Chapter 10 of Libertarian Accounts of Free Will.

I do not find this notion puzzling at all, so I suppose I should say something about why. I would be very interested to find out what was wrong with this line of thought, since both this question and the more general question about whether objects can be causes have always irritated me: they seem like non-questions. Consider:

First, as far as I know, there are no 'bare' events. Events involve objects: the bat strikes the ball, the boiling water turns to steam, I type the word "word" on my keyboard. Barring, then, the elimination of the distinction between objects and events, or the demonstration that there are bare events and/or no objects, it seems as though objects and events are part of the same 'universe' of entities, and there is no principled reason why types of object-causation might not be defined in relation to event-causation.

Second, there is a fairly clear way of doing this. Consider a potassium-40 atom undergoing radioactive decay. Every so often, the atom emits a beta particle. The potassium-40's emitting of the beta particle is surely an event.

Let us assume, as appears to sometimes be the case, that there is no cause external to the potassium-40 atom for this decay. And let us imagine, this time contrary to the facts, that the potassium-40 atom is an undivided whole with no internal structure; or, alternately, let us assume that for whatever reason we have regimented our speech to strictly avoid any discussion of the internal structure of the atom.

What in this case can we say about the cause of the emission? Ex hypothesi there is no event to trace it back further to. But does that mean there is no cause? I'd rather say that the potassium-40 atom is the cause, that the potassium-40 atom causes the potassium-40 atom's emission of a beta particle.

Most of us are trained to think here: OK, but there must be something going on inside the potassium-40 atom that causes the emission to occur, and those internal events are the real causes of the omission. I used an example of radioactive decay to conjure a vague sense of prima facie plausibility for the idea that this might not be the case, but since this is metaphysics let's put the real world to one side. For the question will repeat itself with respect to the entities that constitute the internal structure of the potassium-40 atom, or whatever thing we are talking about. Do those events always have outside causes? Or are there some objects which just do things?

We could look at fundamental particle physics at this point to get a sense of how our world works, although there are going to be some confusing issues - for instance, the entity status of quarks, which never appear bare, and also how to interpret the second formulation of Heisenberg's uncertainty principle and/or the idea of the universe as a kind of energy field in terms of objects. There are some fairly intelligent people who think that quantum field theory might ultimately imply some sort of extreme holism of the kind more often associated with, say German idealism than the kind of quasi-mechanistic philosophy that often naturally fits with the kind of object and event talk that we are doing here. On the other hand, at least some fundamental particle decays sort of look like cases where a thing just exists for a while and then transforms into a different kind of thing by emitting a photon, say. It might be that there's a cause, but it might just be that that's what particles of that kind do, in which case the causal chain either stops with the particle itself or goes back to the event which brought that particle into being in the first place.

But those are at least quasi-physical questions. Metaphysically it seems that we can simply restrict ourself to the following: any event will involve one or more objects. Their causes will come from one or both objects themselves or from events external to it/them (or both). Cases where only one object is among the causes of an event are possible (vibrations, particle emissions). If the causes do not come from events external to the object then they must either come from the object considered as a whole or from the parts of that object. If it comes from the object considered as a whole in some way, then I think I would say and others would say that that object is the cause of the event in question. If it comes from the parts of that object, then we may run the same set of options, until one of a few things happens.

(a) it turns out that the universe is infinitely divisible, in which case (I think, this is sketchy) it will always be somewhat arbitrary where we want to draw the line between whole objects as causes and invoking events involving their constituent parts, which again seems to suggest at least a limited place for object-causation in blocking regress; or

(b) it turns out that there are events causally traceable to partless objects and/or whole objects but not their parts, in which case the causal lines either terminate on that object or go all the way back from that object's current state to the event that brought it into being in the first place; or

(c) all events all the way down, even those involving partless objects, are caused by something external to them, so that there is never an event without two distinct objects being involved (as e.g. the vibration of an energetic particle might be imagined to be).

If case (b) is conceivable it is a conception of what sort of universe we might need a special place for 'object' or 'substance' causation in. If there are partless wholes and/or irreducible wholes that cause events, such that the only prior event that can be sensibly invoked (assuming that the partless and/or irreducible wholes are not eternally existent) is the event which brought those wholes into being in the first place. Since normally events proceed in a sort of web, and here we have an at least conceivable locus of event-generation which remains the same over time, we likewise have a potential role for object- or substance-causation to play in an event-causal universe.

We might put this a different way by saying that while normally causes occur in strict temporal sequence - the bat's swing takes up all the time until the bat strikes the ball, and the ball's flying out of the park begins after it is struck by the bat - object-causes involve 'time jumps' between event-causes, where the only prior event causing, say, the beta particle emission is the event that formed, say, the potassium-40 atom in the first place. Or, if a certain conception of free will is correct, the only prior cause of my writing this note is my having been conceived by my parents.

Here is another set of considerations that makes a plea for object-causes along similar lines. We often say things like

Bonds' swinging the bat caused
the bat to strike the ball, which caused
the ball to fly out of the park.

So, the bat's striking caused the ball's flying, and Bonds' swinging caused the bat's striking. What caused Bonds' swinging? Well, the pitch, external social factors about the game, but also presumably Bonds himself. So what caused Bonds to swing? It would be natural to just say "Bonds" here. This may simply be a marker of our ignorance - the hard determinist thinks so, for example - but we use these markers all the time, when we treat objects, agents or otherwise, as wholes whose special nature contributes in some way to their actions but whose parts and internal structure we don't really want to consider. The metaphysical question is whether there must always be such structure and the type of event-linking that goes with it; I have tried to sketch out some reasons for thinking that it is concievable that there is not. There is also of course the physical question about which of these our universe most ultimately resembles. But we can at least sketch out a metaphysical possibility in which we can mark out object-causes distinct from event-causes.

If readers are so inclined I would be interested in what you think about how to relate these reflections to the perdurantism/endurantism debate. The pessimistic view is that the above discussion requires endurantism to be true; a more optimistic assessment might be that we have here the seeds for indicating the kind of causal structure needed to pick out 'enduring objects' in a world of time-slices, and thus an outline of a sort of argument for endurantism, or alternately an indication of what sort of things we need to know about the world in order to decide whether it's time-slices all the way down or whether enduring objects have a special role in its causal structure. But that is all even more sketchy than the preceding, so it's time to stop.

Agents, Wills, and the Locus of Freedom

Before we move on to arguments for and against free will, there is another issue we should probably consider, one relating to the structure of agents or of the will.

With what should we identify an agent’s will?

One standard answer from ‘folk’ or ‘commonsense’ or ‘philosophical’ protopsychology – the language in which we ordinarily discuss, describe, and analyze human action - is “with that agent’s beliefs and desires.”

(I will not consider skepticism about these notions, or more technical replacements for them from the various branches of scientific psychology and cognitive science, at present. It would be interesting and potentially quite valuable to try to recast the traditional problem of free will in the terms of a better developed and/or scientifically ‘alive’ psychological theory. However, it seems doubtful to me that, whatever models ultimately prevail in our attempt to understand our own minds, representational, inferential, attentional, and motivatory states of human beings will not both exist and be important parts of how we conceptualize human action; and these are, essentially, beliefs and desires, no matter how they are cached out in terms of genuinely better theories than we find in the protopsychology of ordinary reasoning.)

This is a troubling assumption if one hopes to establish the genuine existence of traditional free will. Not because – as one sometimes finds written, presumably due to lazy composition by philosophers who otherwise basically understand the underlying issues here – if the actions of an agent are determined by that agent’s beliefs and desires, such an agent cannot be free. This is clearly an error in the absence of further elaboration: if there are any, beliefs and desires are parts of agents, and so there is no cause external to an agent determining his or her will simply because his or her actions are determined by them. Having beliefs and desires which were genuinely owned by and in some way original to us would indeed be exactly the sort of thing we would need to establish the existence of free will.

The problem is rather that at least the great majority of an agent’s beliefs and desires do not seem to be internal to that agent. One problem stems from the fact that a great number of beliefs and desires are beliefs about and desires for objects, or for doing things with objects, which are in many cases external to agents as we ordinarily conceive them.

(Alternately, we might reconceive the boundaries of agents in such a way as to include the objects of their thought; whether this sort of move might lead to extended minds or even panpsychism, and how making this kind of move might affect our perspective on the free will problem, is an issue I wish to consider later on.)

A second problem comes from what might be called the ‘regress of finitude’, which is a version of one of the main arguments sometimes given against free will, and which has both theological and scientific analogues to the protopsychological version we are considering here. If you consider the beliefs and desires you have now, it would seem that they must have been adopted on the basis of some combination of innate factors, social conditioning, and the choices of the agent him- or herself. The first two involve external causation, so if freedom is to be found in our beliefs and desires it must be found among those which were chosen. But choosing to adopt a belief or desire is itself an action, presumably itself based on beliefs and desires that were held prior to that choice. (It makes no difference whether we add plans, cares, conscious attention, or other factors to the protopsychology here, unless those are structured in such a way as to have been adopted by agents without a determining external influence.) And where do these prior beliefs and desires come from? The same three choices present themselves; but gradually, as we fancifully work our way backwards through the history of an agent’s beliefs and desires, we reach infancy, the embryo stage, and so on; a finite origin for the temporally bounded beings that we are. And presumably at some sufficiently early stage in this development, whatever beliefs or desires an agent has can no longer be credibly held to be freely chosen; an infant does not freely choose its desires for parental comfort, milk, a clean diaper, and a tummy relatively free of gas; and so on. So we are left with only nature or nurture, generally conceived as external to us, as the only originating sources of our beliefs and desires; thus, with a will which is completely ultimately determined by external factors, and which thus is not free.

Roderick Chisholm, a defender of traditional free will, once wrote:

 “If we are thus prime movers unmoved and if our actions, or those for which we are responsible, are not causally determined, then they are not causally determined by our desires. And this means that the relation between what we want or what we desire, on the one hand, and what it is that we do, on the other, is not as simple as most philosophers would have it. We may distinguish between what we might call the ‘Hobbist approach’ and what we might call the ‘Kantian approach’ to this question. The Hobbist approach is the one that is generally accepted at the present time (1966 – SS), but the Kantian approach, I believe, is the one that is true. According to Hobbism, if we know, of some man, what his beliefs and desires happen to be and how strong they are, if we know what he feels certain of, what he desires more than anything else, and if we know the state of his body and what stimuli he is being subjected to, then we may deduce, logically, just what it is that he will do – or, more accurately, just what it is that he will try, set out, or undertake to do. Thus Professor Melden has said that ‘the connection between wanting and doing is logical.’ But according to the Kantian approach to our problem, and this is the one that I would take, there is no logical connection between wanting and doing, nor need there even be a causal connection. No set of statements about a man’s desires, beliefs, and stimulus situation at any time implies any statement, telling us what the man will try, set out, or undertake to do at that time. As Reid put it, ‘Though we may reason from men’s motives to their actions and, in many cases, with great probability,’ we can never do so ‘with absolute certainty.’”

The epistemological language here is unfortunate, but despite that Chisholm’s basic point, that the existence of traditional free will requires there to be some factor other than beliefs and desires which determine an agent’s actions, seems right. Beliefs and desires, though important parts of agents, do not seem to be intrinsic to agents in the right sort of way to ground freedom. And a similar point would hold for any descriptive theories in philosophy or scientific psychology or cognitive science whose architecture of the agent contained terms which could be traced back though a similar regress.

To have traditional free will, we need something in the agent which is ultimately undetermined by external causal factors. The choices here are either (i) that there is some part of the agent which is completely free, or (ii) that, while all the parts of agents are ultimately completely determined by outside causes, nonetheless there is something about the whole agent which is free even though none of its parts are free. (I take this to be exhaustive because any instance of (i) would entail the kind of free agent also given by (ii), making the appeal to the agent as a whole redundant.)

Solutions of type (i) include appeals to a fundamental faculty of ‘choice’ or ‘free choice’ or ‘free will’ or ‘volition’ which exists alongside beliefs and desires and selects between them arbitrarily; also certain pseudoscientific notions such as Eccles’ ‘transempirical power centers’. The existence of any of these seems very difficult to establish, to put it mildly, but if anyone has an idea for where to look I’d love to hear it.

Solutions of type (ii) include dualistic theories of the agent, wherein we are conceived as non-physical beings who nonetheless are able to initiate attempts to act in our quite physical bodies. I am an ultimate kind monist, so this solution is not open to me; I do not believe that there can be two fundamentally different ultimate kinds of thing. Even Malebranche’s occasionalism, in addition to being silly, is not truly up to solving the difficulty of mind-world interaction: if we were purely mental beings for whom God initiated physical miracles in accordance with the determinations of our will, we would not be able to so much as perceive these miracles, given that they would have occurred in a realm entirely alien to our nature. Berkeley may have been wrong, but his philosophy is nonetheless a whirlpool into which many bad assumptions inexorably lead, as anyone who has taught a certain kind of introductory course to undergraduates well knows. To my taste these solutions are even worse than the type (i) solutions mentioned above, so I will consider them no further.

A more promising approach along these lines would simply to insist on agents or persons as part of the basic furniture of the universe, alongside the fundamental particles of physics and whatever else one thinks there to really be. (If I recall correctly, van Inwagen holds a view of this sort.) If we take this approach, though, we need some sort of theory of the agent to assuage very natural worries about just where these agents happen to be in the world, and how it is that they interact with the quarks and leptons and so forth. Such theories are very rare in current philosophy, but we have at least one good historical example, in Kant. For Kant, agency may be prior to external determinations of it – though we may never observe such a determination occurring – because it is the agent’s own spatializing and temporalizing of parts of the world that imposes the necessity of causation upon it. Thus, while we cannot ever observe freedom or know that we are free, even in our own minds, it is nonetheless possible that the ultimate springs of our action are not determined by the causal order which is always necessarily revealed in our spatializing and temporalizing cognition. So while we can’t actually know that Im Anfang war die Tat, it’s at least possible.

Kant’s idea is brilliant and I think that it may contain a kernel of insight that will be useful to us later – we will see. I am not a transcendental idealist, though, and without his metaphysics in place we would need another way to unhook the agent’s actions from the prior causal order, or else at least to show that the springs of will and causation were in some way cotemporaneous. (Again, more soon.)

But in general it is not enough for a ‘basic agents’ approach to become plausible for a savvy metaphysician manages to defend it against common objections, or shows it to be a deep and fundamental assumption of our thinking. (In general showing that something is a deep and fundamental assumption of our thinking is less valuable than it is often taken to be: so much the worse for our thinking if it turns out untrue for all that.) The location of agents is coextensive with that of various physical and biological beings and structures; furthermore, there is no physical or biological entity over and above those to which one can point when indicating the agent. So if we are to say that agents nonetheless exist in some important and fundamental way, we must not only defend them against objections; we must say something about what they are and what entitles us to think of them as distinct from the physical entities which make them up.

This leads us to a third type of effort along these lines, one which has been at least sketched by O’Connor: to consider the agent as an ‘emergent’ entity, one which ‘supervenes on’ but ‘does not reduce to’ more basic physical and biological structures. (Perhaps social ones as well, if social structures can supervene on e.g. biology without a prior determination of individuals within that society.)

On this sort of view, as I am able to understand it anyway, agents would be a kind of entity which came about when the right physical (biological, etc.) entities were assembled in the right sort of structure – which structure itself was a fundamental cause of certain events occurring when other entities interacted with it. Perhaps a crystal lens would be a useful analogy: the crystal lens is made up of atoms, and the property of those atoms determine what can be made of them; nonetheless, it is the crystal structure as a whole which makes possible certain kinds of focusing of light, etc. through the lens, and this structure is not reducible to the single or combined powers of the atoms that make up the crystal, without this structural assembly as well.

This is ground on which sensible philosophers fear to tread, but I think e.g. the famous two-slit experiment makes the importance of structure even more apparent: even a single electron may be caused to interfere with itself by the opening of a second gate in an apparatus; and so it need not be the way that e.g. every individual photon interacts with every individual atom in the crystal that determines the type of focused light that comes out of the crystal; it may genuinely be a property of the whole structure that determines the output of the system, along with the properties of its parts. This is only an analogy, and I would love to see this discussed in more detail by a knowledgeable person, but anyway this is the kind of thing that I take it a successful ‘supervenientist’ about anything would need: properties of structures that do not reduce to the properties of their parts. Arches are sometimes put forward as simpler examples – the combined forces exerted by all the stones on one another would not suffice to hold the whole up were the stones not arrayed in the appropriate way.

These arguments that structural wholes have causal powers distinct from the causal powers of their parts have always seemed somewhat plausible to me, but only in a fuzzy-headed sort of way. All of these issues about supervenience need more discussion, especially if we're going to try to see how far we want to develop this kind of solution.

Assuming that the notion of supervenience I have sketched above can be made out in a systematic way, it seems not only possible but quite likely that human beings are agents of this type; it is not merely the neurons, neurotransmitters, hormones, etc. inside us doing their individual causal things that make us physically what we are, but, surely, the way that all of those are assembled into a structure. One could accept this even while still admitting that some of our abilities depend on parts of the brain and go away when those parts are damaged.

But how is this supposed to help with agential causation?

As far as I can see the idea is this. x supervening on some y’s may make those y’s into  causes of x’s coming to be. But then there can also be some things that x does that do not depend on any of those y’s individually and cannot be said to be caused by them. Thus: the arch supervenes on the stones; but the arch stays up because it is an arch, and not because of anything the stones do singly or jointly irrespective of being organized archwise. So the staying-ups at t of the arch are caused by the arch, and not by the stones that make it up. On the other hand, there is no arch without the stones, and so if one stone is smashed, the arch, which supervenes on the stones, will cease to be. The crystal cannot be without the atoms that make it up, but its focusing of light depends not on those atoms, but on their being arranged crystalwise in a way which focuses light. It is the crystal which causes the light to be focused.

Likewise, I as an agent depend on the existence of various physical and biological structures: I would not be without them. But my willings come not from any of those parts, or at least not always (it may be that some do); they come from me, the agent, and not from any independent entity outside of me. And thus I may be free, in the sense that I am that agent which acts, and although I may die or my parts may become damaged in a way which deprives me of agency, still the things that agent does are not reducible to the causal sources of its parts; they are whole and intact, and make new contributions to the causal sequence of the universe by virtue of the new structure they introduce into it.

This approach needs more investigation. It has a little bit of the feel of being a cheat and also a little bit of  the feel of being ingenious-if-plausible. Our existence depends on the existence of parts which are causally dependent on something external to us, but our actions and the events we initiate are not so dependent, because they trace back to us as a whole and not to our parts. 

So: 

1. Is this sort of move plausible?

2. If it works, does it really provide us with the kind of free will we were looking for? If our existence is contingent on physical things outside of us but our actions are not, that seems like a pretty good outcome for standard views of human agents anyway. But perhaps the ultimately external determination of our natures will still raise problems for some. Or perhaps free structures, even if they exist, aren't suitable candidates for full-blooded selves or agents on some other ground. Lots more to think about if you want to go this way.

In conclusion, it would be nice to have a solution to the problem of freedom of the will that didn't depend on anything more than a very general description of the will itself, since any such description brings with it additional dubious entanglements and premises. But if one does want to go on and describe the structure of agents or of the will as a way of fleshing out the details of one's account, or defending it against objections, I think this post brings out at least some of the important options one can take in so doing.

Preliminaries

1. If something is an action, then it is a causing of an event by an agent.

2. Will is the faculty by which agents attempt to act.

3. If an agent wills an action, the agent attempts that action. Willings are exercises of the will, or attempts to act. Successful willings bring about the action that the agent attempted.

3.1 If we understand exercises of will in this way, they are similar to what some philosophers call intentions,   which if successfully realized in one’s environment yield intentional actions. For now we leave open the question of what factors within an agent contribute to an agent’s will.

3.2. Willings are temporal events and thus able to cause other temporal events. (This is intended to meet Broad’s objection to the very idea of agent causation.) An agent wills A at t and, if this is successful, A occurs. Thus: S wills A at t and, if S is successful, A occurs at t + Δt.

4. A free will, in the traditional (libertarian) sense, is one whose willings are not always entirely ultimately caused by events external to the agent.

4.1. Not always: a free-willed being may sometimes fail to exercise that freedom without thereby losing his or her free will. We may still possess free will even if many or most of our willings are entirely ultimately caused by events external to us.

4.2. Entirely: it is inescapable that any finite agent’s willings will be partly caused by factors external to that agent. Freedom of the will is not jeopardized by the fact that my will may include factors such as beliefs and desires which were formed in relation to and logically depend on actually existing entities in the world outside of me. The question is whether there is any element of my willings that does not so depend.

4.3. Ultimately: without this condition, we would trivially possess free will any time we acted after reflecting or deliberating over potential actions we might take. For then mental processes internal to us would occur and terminate in willings which did not directly causally depend on anything outside us; only immediate and un-thought actions might be un-free. Free will in the traditional (libertarian) sense requires that at least one of the causal factors leading up to an event has its ultimate origin within that agent.

4.4. Defining free will in this way is consistent with an ‘ultimate origination’ view of free will, as described e.g. by O’Connor (http://plato.stanford.edu/entries/freewill/). While other views of free will are possible (because actual), I do think this rough analysis captures the ordinary thoughtful person’s understanding of free will. This does not imply that this is the only or only correct conception of free will, much less that free will exists.

4.5. It is often suggested that, in addition to an ultimate control condition, genuine free will also requires alternate possibilities for action on the part of the agent. However, this seems false, as ‘Frankfurt-style’ examples seem to me to show. Consider a person who deliberates and acts as we do, but who is followed by a guardian angel who would without detection cause that person to will the right thing in any circumstance where that person would will otherwise. This person is essentially trapped in a fatalistic micro-universe where he will always do the right thing: there is no other possibility of action for him. However, in the case where this guardian angel has been assigned to a saint, or even where dumb luck produces an entire lifetime of willing the right actions, the guardian angel never actually does intervene, and the person does the right thing through his or her own will alone.

This person has no possibility but to will right actions, but if the angel does not intervene then these willings will spring from him or her rather than from any outside source. And if we have free will, and this person is like us, it does not seem to me that sending such an angel in any way deprives this person of free will. The angel does in a certain sense rob the person of freedom of action, since she cannot do otherwise than as she does, but so long as all that agent’s successful actions spring in part from his or her own will, and so long as that will is not itself determined by any outside factors, it would seem to me to remain free.

The belief that free will requires alternative possibilities may indeed stem from the belief that free will requires an ultimate, jointly sufficient cause of action within the agent. If one had this sort of control, and if the universe one inhabited developed differently over time depending on which prior events occurred – if the future were open, and fatalism were not true – then ultimate control would in such an environment entail that alternative possibilities were open to the agent.

This observation, if correct – and this entry is sketchier than some of the others in this post – would seem to support the idea that a could-have-done-otherwise condition is in no way fundamental to the traditional (libertarian) notion of freedom of the will. For on the one hand we can give cases where agents cannot will otherwise and yet their will is still free, and on the other hand we can explain why someone who believed in a free will involving ultimate control and did not accept an extreme form of fatalism would automatically conclude that alternate possibilities would be available to a free-willed agent.

4.5.1. If the argument sketched above holds up, it would follow that a family of arguments made against incompatibilism (by Ginet, van Inwagen, O’Connor, and others) on the basis of an alternative possibilities condition are irrelevant to the real issue of whether traditional (libertarian) free will exists or is possible. The defender of traditional libertarian free will can jettison alternative possiblities as cheerfully as the compatibilist. The real issue between them would then remain what it has seemed to most thinkers to be historically: whether it is possible for (a) any agents, (b) any physical agents, and/or (c) human agents to possess such a faculty or not, and, if it is possible in case (c), whether human agents actually do possess it under any conditions.

Note. This is a preliminary stab at (one way of) defining free will. Many people think that free will understood this way is in some sense impossible. The next steps in developing the discussion seem to me to be to see what can be said to motivate a belief in the existence of free will in this sense and to consider the main arguments opponents of free will have brought against it.

Traditional Free Will via Nontraditional Conceptions of Human Agency

Welcome!

In this corner of the internet I hope to pursue a variety of thoughts about free will over the next four months - or, if it's a great deal of fun, perhaps longer. What particularly interests me is the attempt to defend something resembling a traditional agent-causal libertarian conception of free will by way of nonstandard conceptions of human agents or selves. This will also entail some thinking about what agents and/or selves actually are.

Comments and critique are warmly welcomed. The posts will start going up in earnest within a couple of weeks, probably sooner.