March 5, 2010

A Case against the existence of Free Will

The term “free will” has essentially two meanings. The first definition of free will is that it is that state in which one’s decisions can be realized in physical actions. In this sense, if one is physically constrained by devices, disease or other externally induced circumstances one is, to the extent of the constraint, deprived of free will. The second definition of free will is that it is that state in which one’s decisions constitute a first cause. In other words, the possessor of free will does not make decisions because (or at least not wholly because) he or she is caught in the middle of some inexorable physical process, but rather the decision maker is the actual originator of physical processes. While I will deal briefly with the first definition in closing, my chief interest is in the second.

My argument depends on certain assumptions that, while admittedly arguable in themselves, are by no means weakly held positions.

My first assumption is that the order we observe in nature is not illusory, but rather refers to, however imperfectly, a body of stable relationships that exist in a fully ontological sense. If one assumes that we live in a chaotic universe in which the apparent laws of nature might suspend themselves at any time, then arguments of any sort are futile. Any understanding of states of affairs, no matter how tenuous, must admit at least some constrains to be coherent. When I use the term “physics” in my argument, I mean just this set of constraints and relationships, whether they are presently known to us or not.

My second assumption, closely related to the first, is that nature is fundamentally causal. This assumption alone does not preclude the existence free will. Indeed, to assume that anything can be a first cause one must certainly admit to the existence of causality. By “causal” relationships, I mean to describe relationships in which objects and events are bound together by necessity and not merely by accidental, albeit stable, correlation.

These preliminaries being assumed, let’s dissect the concept of free will a little further using a thought experiment.


Imagine a flying bird. I don’t want to engage in a muddy debate about whether nonhuman animals have free will or not, so let’s just accept that our thought experiment bird, for sake of argument, does have free will in the sense of having the power of first cause. We can describe innumerable trajectories our bird might take across the sky, including many that plainly violate the laws of physics. For example, the bird may not fly a path that would require changes of direction too rapid for the aerodynamic forces it is able to exert with its feathers. Neither can it fly straight up for very long, for a variety or understandable physical reasons. Nevertheless, so long as our bird might fly along at least two alternative paths, however constrained by gravity and aerodynamics, we might still consider it free.

Compare the flight of our bird with that of a thrown ball. Notably, we can imagine exactly the same innumerable set of trajectories for the ball that we could for the bird. Unlike the bird, however, the ball clearly does not possess either free will, or any physical means of altering its own trajectory. Its path is, without question, wholly predetermined by physics. With our knowledge of physics we can predict its trajectory with impressive accuracy. Moreover, even if we knew nothing of the exact physical laws that govern its trajectory, merely watching the ball’s motion would give us an intuition that it is a “thing” and not a “being”. Apart from the relatively static properties of its mass and its shape, there is nothing about the ball which determines its trajectory. It does not “choose” to do anything. It is a neutral participant in an inevitable physical process.

The bird’s behavior differs from the ball’s in at least two ways, one from our point of view and one from its own. From our perspective, the bird is unpredictable; the ball is not. The bird appears to have free will, not because it is unconstrained by physics, but because we can imagine it taking any of any number of plausible paths. From the bird’s perspective (which we are privileged to know only because this is a thought experiment) it has the conscious perception of free will – which is to say, it is aware of having choices. The ball, of course, has neither plausible options nor any capacity for awareness.

Now consider an entity whose status is somewhere between a bird a ball – a heat-seeking missile. For those who are unfamiliar with such things, a heat-seeking missile is essentially an autonomous robot whose function is to intercept and destroy aircraft. It detects infrared radiation (heat) with a special camera and adjusts its course toward the source of that radiation using a rocket and aerodynamic surfaces. Such missiles are “smart” enough to fairly reliably distinguish between aircraft and heat sources that are not aircraft.

Like both the bird and the ball, the missile’s trajectory is limited to a considerable degree by external physics. It is subject to gravity, for example, and its ability to turn is limited by the aerodynamic forces it can exert with its control surfaces. While its trajectory is theoretically predictable (given one has knowledge of the characteristics of all the heat sources in the range of its camera) that trajectory would be much more difficult to either predict or describe than that of a merely ballistic object like a ball.

More significant than the missile’s brute obedience to physics is the fact that its behavior calls into question what it is that constitutes a “choice”. If there are at least two heat sources in front of it on which it might potentially home, then, in at least some sense, the missile’s behavior is the result of a “choice” between alternative imaginable paths. While no suitably educated observer would say the missile has “free will,” we nevertheless get into trouble when we attempt to explain the distinction between its actions and any real bird’s.

Assuming we have access to the knowledge of both the missile’s internal physics (its camera, actuators, rocket, software, etc.) and the environment in which it is operating, we can predict its behavior up to the level of precision of that knowledge. If it does something we did not predict we can infer that there is either something in the external environment we didn’t notice, some variance between the missile’s components and our assumptions about them, or, perhaps, some aspect of physics we simply don’t understand. We would not assume an inability to specifically account for the missile’s aberrant behavior constitutes an argument for it having “free will.” We assume, in short, that entities like balls and missiles behave in a way that at least would be entirely predictable if our knowledge of the relevant physics and states of affairs were sufficiently complete. We do not resort to endowing such entities with extraphysical sources of causation.

When we speak of other entities as having “free will,” whether they are birds or human beings, we are in effect denying that we might be simply facing problems of enormous physical complexity. In place of an unknown (and perhaps even unknowable) physical solution to the problem of behavioral unpredictability, we are postulating an explanation which is little better than magic. While we cannot logically disprove the existence of such extraphysical entities as “free will,” there is nothing necessary about them either. In fact, there are no bona fide examples of physical events that could not have their origins in some purely physical cause. There is much about the nervous systems of animals we do not understand, but there is nothing about those nervous systems that clearly renders them incapable of being the sole mediators between an organism’s environment and its intentional actions. Anyone who has ever been hungry or has consumed an appreciable quantity of alcohol can attest to the fact that material factors strongly influence choice, and it is at least plausible that all decisions are ultimately reducible to the net influence of various environmental and neurological factors.

We can’t say with logical certainty that “free will” does not exist, but since its existence is not necessary to explain behavior we can say that the assertion that it exists is a violation of the principle of parsimony. It is more parsimonious to look for explanations in some as-yet-unknown physical process, or even in some collection of physical processes so complex as to be unknowable, than it is to put forward the untestable assertion that there are extraphysical first causes.


Up to this point, I do not believe I’ve covered any new ground. One can make damaging arguments against the existence of free will in entities external to oneself, but the strongest practical argument for free will is probably the introspective one -- the claim that “I know my own actions to be free.” This is a harder nut to crack, but I believe it is possible -- without resort to peculiar quasi-dualism proposed by epiphenomenalists.

When one says “I have a choice” what does this actually mean? The proponent of the free will hypothesis takes such an assertion to mean that there are multiple possibilities for one to select from – and I am using the term “possibilities” in a strict sense: something that can be a state of affairs, not something which is imaginable but impossible. What I am proposing as a counter to this view is that, while the capacity to imagine multiple “possibilities” plays a role in decision making, any actual decision itself can only be understood as ultimately reducible to some physical process. In simple terms, we perceive ourselves as free because we can and do imagine alternative actions before making a decision, but we inevitably choose from among those apparent alternatives the one which suits our predispositions best.

If I may take the liberty of generalizing my internal experiences to those of other human beings, I must conclude that most of our actions are straightforwardly mechanical. When I type the letter “m” for mechanical, I do not think to myself “maybe I’ll press the ‘m’ key to produce the letter ‘m’ instead of trying to do it with the ‘q’ key.” In truth, there is nothing in the process of typing the letter “m” that has even the appearance of a choice. I just press the “m” key and an “m” appears. Likewise, when I eat my dinner I do not generally entertain the “possibility” of eating it off the floor – though I am at least arguably “free” to do this. Most of our actions are driven by the well established reflexes we have accumulated over a lifetime. Some are driven by genetics. The sort of weighty decisions ethicists like to talk about are comparatively rare events.

When we do act in a way that involves making recognizable choices, what is it we are doing that differs from our reflexive involvement in the causal world? When faced with what we actually perceive as a choice, we weigh the relative merits of each alternative. We attempt to predict the future. We imagine alternative futures for the purpose of selecting the most attractive one. We never fail to select the most attractive one, “attractiveness” being defined by our unique set of heuristics, our background, our genetics, our current state of knowledge and awareness, and any number of other factors than manifest themselves, ultimately, in the state of our neuroanatomy. There is nothing about this process that requires any special causal powers.

Consider the trivial choice of selecting a snack from a vending machine. We approach the vending machine not merely with a handful of change, but also with a huge collection of memories and other sorts of predispositions. Typically, we eliminate from consideration all of those items experience has shown us we don’t like. We gravitate toward those we know we like the taste of, perhaps toward those we think might have some nutritional value, and perhaps toward those which hold some pleasant associations unique to us. We might also have acquired a heuristic that inclines to trying new things, or a heuristic that inclines us to avoid them. We might have cravings related to our metabolic state.

When we choose one snack over another, we may or may not be able to explain our decision. One might say, “cheese crackers are my favorite snack, and that is why I picked them.” Such an answer erodes the case for free will on its face because the chooser is substantially aware of the dominant causal factor in the choice. The chooser is predicting that the cheese crackers will taste better than the other alternatives, and “tasting better” is almost certainly reducible to some biochemical process.

One might say, on the other hand, “I could not make up my mind so I chose the cheese crackers.” This answer admits to no known cause on the part of the chooser, and might be explained in either of two ways. The choice is either genuinely random, or it is the result of some process of which the chooser is simply unaware. If a choice is genuinely random in some quantum statistical sense, then it can hardly be considered an act of free will. On the other hand, if a decision is the result of subconscious motivations (or something of that sort) then it is still the product of an antecedent cause, so the decision cannot be a first cause in itself. There is no more reason to ascribe special causal powers to the subconscious than to the conscious, and even if there were, the possession of extraphysical subconscious powers is clearly not what we mean when we postulate free will.

Yet another (and quite atypical) response our decision maker might make would be: “I chose the cheese crackers because I am free to do so, even though I would have preferred the cherry pie.” Although superficially this sounds promising, on analysis it is either a lie or a self-deception. To make a choice for the sake of proving one’s freedom is just to evaluate the goal of the selection task differently. It is merely to find attempting to make a particular philosophical point more attractive than selecting an attractive food. It illustrates not freedom, but merely the working out of different causal sequences that happen to be granted precedence at the moment. The decision to put a belief before a physical pleasure is no proof of freedom either, as it may also be easily explained in terms of yet other antecedent conditions acting on the decision maker’s neuroanatomy.

Ultimately, it is simply unintelligible to talk about a decision as an isolated cause. In practice, we know decisions always exist within the unique context of the decision maker’s cognitive world. It is always legitimate to ask why a decision maker made a particular choice, which would not make any sense if decisions were uncaused, spontaneous events.

Relating the vending machine example above to our earlier missile example is revealing. Both the behavior of a heat-seeking missile and that of a hungry human being can be explained (at least in principle) by wholly deterministic physical processes. Both entities, too, behave in ways which vary according to the circumstances they detect in the external environment. They differ, most fundamentally, at the level of intentionality. Where the human being makes a series of predictions and inevitably seeks the most attractive, the missile is not “attracted” to any outcome in the same sense, nor is it capable of anything that answers to the term “prediction”. It does not pursue its target because it has some internal conscious purpose which it is deterministically compelled to carry out. The machine operates by a comparatively simple and wholly unconscious algorithm, more-or-less as I do when I press the “m” key to type the letter “m”. What sets human beings, and no doubt a good many other animals, apart from machines is not some special power to initiate causation, but rather the ability to attach action to meaning and meaning to action. A bird may fly a course wholly determined by physical causes, but one of those causes is its particular purpose at any given moment – a purpose that entities like balls and missiles do not possess. Current artificially “intelligent” machines may respond to external events as if they were predicting the future, but the only actual predictions that can be inferred from their behavior are the predictions of their designers.

To state my position another way, the distinction between a mechanical reaction and a decision is not the interposition of free will, but the interposition of an awareness of causality itself. To make a decision, in the sense that human beings make decisions, is to model at least two imaginary causal sequences in imaginary space-time, and to pursue the outcome of one or other of these sequences. The predicted outcome is the goal we pursue in making a decision. Further, to make reasonably reliable predictions one must have a substantial working knowledge of the physical world in which one is immersed. Without such knowledge, predictions would simply be wrong too often to be useful. While nothing about this process requires free will, it is no mere brute reaction either. Making decisions based on such understanding, no matter how deterministic the actual mechanism of the decision might be, is certainly no small achievement.
Ironically, it is this very ability to model the future in a variety of ways that creates the illusion that we have special causal powers. What we call freedom is nothing more or less than the general belief that each of our predictions really could represent some future state of affairs.1 It is a byproduct of the impressively complex, evolutionarily advantageous, but ultimately deterministic, way our nervous systems respond to a varied but not wholly unpredictable environment.

I think it is necessary, at this point, to clarify the distinction between the position I have outlined and the position of the epiphenomenalist. The epiphenomenalist would agree that our decisions are fundamentally deterministic, in other words, that the nervous system behaves like an elaborate machine and that the entity we identify as “the mind” cannot exercise the power of first cause. However, the epiphenomenalist goes further in taking the position that our cognitive activity is inconsequential, playing no role whatsoever in our actions. My position is that, since our cognitive processes are inseparable from the neurochemical processes that constitute them, it is wrong to assume that our cognitive activity is inconsequential. The epiphenomenalist position is analogous to saying that a regular lattice of silicon and oxygen atoms somehow only contingently appears to be a solid at our level of perception, and that it might just as easily be a liquid. This is contrary to experience. Without intentionality we would not, and could not, participate in the causal universe in the way we actually do. My assertion is that what we perceive as a decision and what we define (and, to the extent that our instruments are capable, detect) as a brain process are merely different expressions of the same state of affairs. “Thoughts” are no more capable of being first causes than the neurochemical processes which constitute them, but, as they are features of physical processes, the subjective experiences we know as “thoughts” are logically bound to causality. I think epiphenomenalists make their mistake because they believe that “thoughts” really are first causes (even if only for themselves) and must therefore be causally quarantined to protect the integrity of a material universe which they concede is deterministic.

I would like to make clear, too, that the elimination of a first cause interpretation of our capacity to make decisions does not require that the universe be wholly deterministic. If quantum theory is substantially correct, some states of affairs, at least at the level of subatomic particles, are statistical rather than deterministic. While it is at least imaginable that some unknown causes underlie this apparent randomness, even if the statistical nature of states of affairs at that level is a brute fact it does nothing for the “mind-as-first-cause” hypothesis. Genuinely non-deterministic processes would render the future less predictable, but would not require, or even imply, any special causal powers of the mind.2
Finally, I would like to make clear (if I have not done so already) that while my position implies that decisions are, at least in a broad sense, computations, I do not take the position that a capacity to perform computations is a sufficient condition for intentionality or consciousness. John Searle’s Chinese Room argument seems to cover this issue forcefully enough, and I will not re-cover the same ground.3 My purpose in this essay is to show that we have no reason to suppose we possess the power to truly originate causal sequences, not to show that all of our cognitive processes are reducible to computation. Indeed, I don’t believe either intentionality or consciousness are computational.


Inevitably, any attack on the idea of free will must come to terms with our own sense of self-identity. If we are not free beings – what are we? Well, from a starkly materialist perspective, we are merely the loci of a particular kind of causal complexity. Being a sort of “thick spot” in the field of causation may not be a very satisfying self definition for most people, but there doesn’t seem to be anything inaccurate about this view. A perhaps more satisfying but no less accurate self assessment is that we rare instances consciousness in a universe that is largely unconscious. Consciousness, while a difficult entity to define, is at least ontologically less problematic than free will. While freedom isn’t necessary to explain our experiences, consciousness certainly is. None of us can honestly entertain the notion that he or she is absolutely unaware. To engage in any sort of reasoning at all requires an awareness of something. Taking the position that the elusive entity I identify as “me” is fundamentally a conscious entity is therefore a more secure stance than assuming what is fundamental to my nature is freedom. It is perfectly possible to act and make decisions without pretentions to magical causal powers, but many of our actions would be impossible (or at least wildly improbable) without a conscious capacity to make predictions.

The other sense of free will, that of being unimpeded in carrying out one’s intentions, is wholly independent of the first cause definition. Whatever the nature of our decisions might be, our ability to carry out those decisions is subject to circumstances that are external to the decision making process. To say that I am free in the sense that I can carry out my deterministically derived decisions without external impediments is not a meaningless claim. Here too, however, our perception of freedom pivots on our assessment of our own identity. The deterministic processes that constitute my decision making are still identifiable as part of “me,” just as the deterministic processes that make my muscles function are identifiable as part of “me,” but any deterministic processes in my environment that prevent me from carrying out my decisions are not a part of “me.” While in some doggedly holistic sense this distinction may be trivial, it is probably inevitable that we organize cognition from the perspective of our unique identities. Evolution has predisposed us to model the rest of the universe as somewhere we live, not something we are.

The sense in that we are free when we are not externally impeded can be readily explained in terms of the determinist perspective I’ve proposed. Remember, to make a decision is to pursue the most attractive option among some set of imagined options. To suffer a loss of free will in this more external sense, then, is just to have one’s options constrained to some subset of unattractive choices – or at least to be denied some plausible more attractive choice. Unless one is utterly paralyzed, one always has plausible choices of action. Even if one is utterly paralyzed, one still has plausible choices in regard to one’s own thoughts. A slave does not feel enslaved because he or she lacks imagined possibilities, but only because even the most attractive of those imagined possibilities is unpleasant.4-------------------------------------------

1 This erroneous belief may even be necessary to the decision making process: if we were constantly aware that most of the causal sequences we imagine before making decisions simply could not occur, we might tend to engage in futile, and rather paradoxical, searches for the possible rather than the attractive.

2 I owe this observation to John Searle.

3 See:

In theory, one might be severely constrained by external circumstance and yet feel completely free. Consider the following thought experiment. Imagine a vending machine with the magical ability to predict our decisions perfectly. Before we push a button to make our selection, the vending machine disables all the other buttons. Thus, we are constrained not merely by our own internal deterministic processes, but by an external one as well. Provided the machine made perfect predictions, however, we would be oblivious to the constraint. While it is tempting to imagine that we are made less free as the number of options available to us is reduced, we are in fact only less free, in the sense of being materially constrained, when we perceive the actual constraint.


  1. Am example I like to use is temperature: there is no such thing as temperature at the molecular level. Individual molecules are not hot or cold. There is only more or less Brownian motion. And yet, if you try to live without the concept of temperature… you’ll get burned.

    Temperature/Free Will are labels for heuristic rules that allow us to function, not statements about physics.

    Although you're a little bit wrong at the end - people feel constrained and unhappy when they are not given enough choices, even when the choices they have are all good. People know they are supposed to be exercising free will, and will suffer psychologically if not allowed sufficient pretension of doing so.

  2. Yahzi,
    No, I suspect we are still essentially in agreement. I said: "To suffer a loss of free will in the second sense, then, is just to have one’s options constrained to some subset of unattractive choices – or at least to be denied some plausible more attractive choice." The last clause covers your instance. I am also careful to avoid the question of happiness altogether. The relationship between happiness and freedom (sense 2) is a messier question than I care to tackle at the moment. While there seems to be a certain correlation there, one must admit that some people apparently thrive under relatively harsh constraints. I may wade about in the nasty little subject of psychology now and then, but I always like to keep the comforting shores of logic well in sight. ;)

  3. Good point - I missed the "plausible."