Is the common notion of human free will compatible with the natural laws of the universe? To so called ‘compatibilists’ who see free actions as emanating from the wants and reasons of human agents, free will looks perfectly plausible. However, ‘incompatibilists’ claim to see the more ultimate sources of human action. The wants and reasons of agents are said to actually be caused by physical processes which are themselves mere natural results of the previous state of the world and the natural laws which govern it. This paper argues that the incompatibilists make a mistake in appealing to such non-agent sources of human action. They fail to realize that free will may exist at one scale, but not at other scales. The non-human sources of action do not exist at the scale where free will may exist. So the non-agent sources of action are not relevant to the question of free will. When free will is considered from the correctly scaled perspective, it does seem compatible with natural laws.
Most people feel like their decisions and actions follow from their own free will. But all actions in the world evolve according to natural laws which do not follow from anyone’s free will. This apparent conflict has led to a large literature where some philosophers have argued that natural laws (or more often, determinism) are compatible with free will. Others have argued that natural laws are incompatible with free will. The general terminology of ‘compatibilism versus incompatibilism’ now frames most of the debate around free will and natural laws. In this paper, I will argue that compatibilism is likely the right way to go, and that what many consider to be the strongest incompatibilist argument is really not strong at all. I will reach that conclusion through consideration of scale; consideration of how the facts change when we zoom in or zoom out from a given perspective. But first, I have to be a little more specific in mentioning the kinds of compatibilism and incompatibilism I will be talking about.
There are a couple of very influential compatibilist theories which describe freely willed choices and actions as resulting from either the desires of the self, or from reasons that the self has for acting. These are ‘source’ compatibilist theories, in that they focus on the source of action. Harry Frankfurt put forth the theory which explains freely willed action as issuing from desires that mesh with important elements of a person’s psychology. The idea is that a person who acts of her own free will acts from desires that are nested within elements of herself. Frankfurt says that his actions emanate from him rather than from something foreign (Frankfurt, 1971).
John Martin Fischer proposes another influential source compatibilist theory, which has become prominent. He speaks of ‘reasons’ as the source of freely willed actions. He claims that action is responsive to rational reasons in a free agent; that if the reasons were different, the action would be different as well. His view is not merely that an agent would be responsive to reasons in some counterfactual situations, but rather that her responsiveness to reasons in the counterfactual is evidence that her actual conduct is also responsive to rational considerations (Fischer, 1994).
I am not arguing that either of these source theories of free will is exactly right or complete. I am making a case in this paper against the negation of the kind of theories exemplified by Frankfurt and Fischer above. That is, I am arguing against source incompatibilism and the thought experiments meant to support it. Michael McKenna and Justin Coates aptly describe source incompatibilism in their review:
If determinism is true, then for any person, what happened in the past prior to her birth that, when combined with the laws of nature, provides causally sufficient conditions for the production of her actions. But if this is so, then, while it might be true that an agent herself provides a source of her action, that source, the one provided by her, itself has a further source that originates outside of her. Hence, she, as an agent, is not the ultimate source of her actions.
. . .
1. A person acts of her own free will only if she is its ultimate source.
2. If determinism is true, no one is the ultimate source of her actions.
3. Therefore, if determinism is true, no one acts of her own free will.
(McKenna and Coates, 2018)
In support of source incompatibilism, some thought experiments called ‘manipulation cases’ have been developed in order to crystalize the incompatibilist intuition. These manipulation cases ask us to imagine an agent whose thoughts and actions are manipulated by some evil scientist. Though apparently not free, this manipulated agent could meet the free will criteria put forth by either Frankfurt or Fischer. So manipulation offers a counterexample to source compatibilist theories. And manipulation cases have been taken farther by Derk Pereboom in his Four-Case Argument for incompatibilism, in which he tries to show that normal reality is not different in any relevant way from a manipulation case (Pereboom, 2001). Pereboom’s argument has been widely regarded as very effective. McKenna and Coates say of manipulation cases, especially Pereboom’s: “Why is determinism any different from a manipulation case? The burden, it seems, is on the compatibilist to show how it is that manipulation cases differ from a normal deterministic history” (McKenna and Coates, 2018). One of the conclusions of this paper will oblige McKenna and Coates by doing exactly that: I will show how manipulation cases differ from a normal reality. And more generally, I will show why source incompatibilism is malformed. All of this will rely on a certain understanding of what I am calling scale, which I will now describe.
Scale Changes Things
Many facts and properties depend on scale. Take, for example, the color of a red apple. At macro scale, where we usually think of apples, it is truly red. But if we zoom far enough into the micro scale (technically to the pico scale), where atoms dominate, the surface of the apple is truly not red. None of the atoms are red. The concept of ‘red’ does not even apply to atoms. Likewise, we can zoom out to the mega scale, where the apple is just a part of some grand entity or process. The agricultural industry is not red; the consumer demand which brought our apple (with all of its color) into existence is not red.
So we see that a property as simple and real as color can not always survive analysis which jumps between different scales. I do not think this should lead us to conclude that the apple is not actually red. It should tell us that when trying to answer whether or not the apple is red, our analysis should stick to the macro apple scale; not the micro atom scale, and not the mega system scale. We should avoid the common assumption that there is truth or insight to be found by zooming in or zooming out. There is a tendency to act as if to investigate means nothing more than to zoom in or out. But the apple proves that there is also truth to be lost by zooming. Truths of one scale may not apply on other scales. The truths of different scales can contradict each other if lifted from their homes, and clumsily mashed together in a scale-free theory.
If we want to answer the question of whether or not people have free will, we have to figure out where in the world of scale we should be looking for truths, from which we can build our answer. For the question of apple color, the right scale for the answer was simply the same as the question scale: the apple scale. I contend that the same is true of the free will question; that we need to look for answers on the same scale as that of the question: the human agent scale. I will also argue that the prominent compatibilists (Frankfurt and Fischer) have operated on this correct scale, while their source incompatibilist opponents have been inappropriately zooming between scales in order to build their theories.
First, I have to argue that free will is like color; that there is such a thing as right and wrong scales at which to look for it. Then I will have to make a good case that compatibilist theories like Frankfurt’s and Fischer’s look for free will at the right scale, and that source incompatibilists look at wrong scales. If I can do this, it will constitute good reason to favor compatibilist theories. Though there may be other issues with the major compatibilist theories, the landscape will become much more favorable for compatibilism in general after giving good reason to reject the source incompatibilist critique, which has so far not been satisfyingly answered.
Free Will is Scale-Dependent
Behavior has causes at many different scales which are simultaneously real and true. Here are a couple of rough examples:
- Agent Scale: I moved to a new city across the country because I wanted to. I wanted to move because I perceived a good reason to; and because I am the type of person that responds to reasons, and that does not fear the type of change entailed in moving.
- Neurological Scale: I moved to a new city across the country because a specific combination of motor neurons in my brain and spinal cord fired action potentials in a certain sequence that caused certain muscle fibers to contract in a certain sequence. Those motor neurons were caused to fire by the neurotransmitters released from the firing of other neurons, which were also caused to fire by the firing of other neurons.
Let us think for a minute about the neurological scale so that we can see that free will is scale-dependent, that there are scales at which we should know not to look for free will, much like we know not to look for color at certain scales. Everyone already knows not to look for free will in a single brain cell. Though I am not aware of any compatibilists nor incompatibilists who have bothered to spell this out, neither camp seems to ever explicitly set out on the project of finding the cell with free will. This is because a single cell does not experience action or potential action as the object of desire, want, or contemplation; it does not experience anything at all. So if we look for free will in cells qua cells, we will never find it, no matter how large the collection of cells we search. Note that I have said nothing of determinism (or probabilistic natural laws) and the mechanistic nature of cells. We already know not to look for free will at the neurological scale before we ever start thinking about compatibilism and incompatibilism. The concept of free will simply does not apply at that scale, much like the concept of color simply does not apply at the pico scale. The concept of free will only applies once we look at the collection of brain cells as an emergent unified subject, which is to say, once we zoom out of the neurological scale, into the agent scale.
Since there are some wrong scales at which to look for free will, any theory which deals with free will needs to be checked for reliance on wrong scales. If any of the key steps in a theory depend on some intuition mined from one scale and misapplied to another, then the theory does not work. This is what is going on in the case of source incompatibilism. When one declares that for free will to obtain, the agent must be the ultimate cause of action, he implicitly demands that free will should make sense at scales where free will was never supposed to make sense. This becomes clear when we ask: What are the non-agent causes, which disqualify the agent from being the ‘ultimate’ case? That is, which causes are more ‘ultimate’ than the agent? There must be some ‘more ultimate’ causes implied. Otherwise, the agent herself could be declared ultimate.
So what are these more ultimate causes? Philosophers seem to avoid enumerating them. But there are some that can be inferred. Most of them, especially in real life, are versions of, “Our brains are made of molecules; those molecules must obey the laws of physics; our decisions derive from brain activity” (Coyne 2019). This is the popular formulation of source incompatibilism, which is circulated outside of philosophy journals by ‘public intellectuals’ such as Jerry Coyne and Sam Harris (2012). Clearly, this version of source incompatibilism jumps between scales in order to trace causality. That would be fine, if this were merely an issue of causality, which meanders up and down between scales, maybe jumping, or maybe branching. But no, this is an issue of causality and free will. And free will is scale-dependent. Free will is supposed to evaporate when we look at the neurological scale; we should expect it to, without causality and natural laws ever entering the question. So when the incompatibilist directs us (explicitly or implicitly) to look at the neurological scale and find no free will there, we should not conclude that there is no free will at agent scale. Remember the lesson from apple color: a property can be perfectly real (or descriptively true) at one scale, but not at others.
Now that I have shown what is in principle wrong with the source incompatibilists’ implicit reliance on ‘more ultimate’ causes of action, I will make clear where this error shows up in their best argument. As mentioned above, the consensus as seen by McKenna and Coates is that ‘manipulation cases’ pose the greatest challenge to compatibilists. And the most forceful of these manipulation case arguments is Derk Pereboom’s (2001) Four-Case Argument for incompatibilism. As briefly described earlier, a manipulation case is a thought experiment where the more ultimate cause of an agent’s action is another person, who manipulates the agent. The idea is that we are supposed to intuit that a manipulated person is not free, yet she might fit compatibilist criteria such as wanting to act as she does, or having reason to act as she does. In a good manipulation case, the agent is manipulated into wanting to do, or having reason to do, the action which she is manipulated to do. Pereboom ties such a case to normal reality by using four cases which get progressively more normal and realistic.
First is electronic stimulation of the subject’s brain, causing her moment to moment decisions and actions in such a way that she still satisfies (according to Pereboom) what various compatibilists contend is sufficient for free will. Pereboom then presents a second case in which the manipulator tinkers with the agent ahead of time by “programming in” the relevant psychological details that will later lead to the agent’s actions. We are meant to find no relevant differences between the two cases in terms of free will. For the third case, Pereboom takes another small step toward normal life: rather than direct electrode stimulation, the manipulation consists of indoctrination of the subject in her childhood, which causes her specific actions in adulthood. this process is said to not be different from the pre-programing case in any relevant way. And so here too, we are meant to conclude that the agent is not free. Pereboom argues that the best explanation for why these first three agents are not free and responsible is that the process of manipulation deterministically caused each of them to act as they did. And so, finally, the fourth case is that in which a normally developed agent in a determined world acts (basically a description of normal reality). This case also is said to be no different from the others. In all four cases, what underwrites the lack of freedom, Pereboom concludes, is the fact that the source of the agent’s actions can be traced back to originating conditions that were completely beyond her control.
Note that Pereboom’s Four-Case Argument makes at least two points. The first is, like in other manipulation arguments, that compatibilist theories are shown by counterexample to be flawed. The second point is that normal reality is, in all relevant senses, just like the manipulation cases. The first point, I think is right, in that it at least shows that some compatibilist theories are incomplete. The second point is wrong. By considering scale, we can see why the second point is wrong and, in the same stroke, hint at what the compatibilist theories need to add in order to become more complete.
Pereboom’s jump in scale, his inappropriate zooming, happens between cases three and four. The first three cases all manage to remain at the same scale in a subtle way. In the first three, causality flows perfectly from the manipulator’s desires and reasons into the subject’s actions. Where person-level things like desires and reasons can definitively, with perfect fidelity, cause actions, we are more or less looking at agent scale. This is where will can make sense. Here, we can appropriately talk about whether or not a will is free; because here, there is will. Depending on the character of the will we observe at this scale, we might judge it free or unfree. In Pereboom’s first three cases, we judge the will of the subject unfree because there are desires and reasons other than the subject’s own desires and reasons, which cause her actions. We can say this without jumping scales. To clarify by way of my analogy, we judge the color of an apple to be non-red if there are colors other than red (such as blue) all over the apple. We make this judgement without jumping scales; without making color a meaningless concept.
Staying in the analogy a little further, Pereboom tells us to imagine three blue apples, which we agree are in fact not red. Then he points out truthfully that all three non-red apples are composed of non-red atoms. And finally, he says that since all real apples are likewise composed of non-red atoms, they are likewise not red apples. Along the chain of his argument, he surreptitiously disposes of the blueness which was the entire basis of our agreement about the non-redness of the first three apples. Without blueness, we have no good reason to agree that all real apples are not red.
Now exiting my analogy, Pereboom tells us to imagine three controlled agents, which we agree are in fact not free. Then he points out truthfully that all three unfree agents are directed by deterministic mechanisms. And finally, he says that since all real agents are likewise directed by deterministic mechanisms, they are likewise unfree agents. Along the chain of his argument, he surreptitiously disposes of the controlledness which was the entire basis of our agreement about the unfreedom of the first three agents. Without controlledness, we have no good reason to agree that all real agents are unfree.
Pereboom smuggles the intuition of controlledness from the manipulator at agent scale, missaplies it at other implicit scales, then takes the deformed product of that misapplication back to agent scale without the controlledness on which the intuition depends. So reality is not, in all relevant senses, just like the manipulation cases. Pereboom’s conclusion is refuted. And compatibilists now have a hint of what is needed to make their theories more complete: something to accomodate a concept of unfree will which can exist alongside free will, much like blueness exists alongside redness.
Scale gives us a clear answer to the question, “Why is determinism any different from a manipulation case?” It is different in that normal deterministic (or natural law) causes of action other than agent-causes do not exist at agent scale. If they did exist at agent scale, they would be agent-causes. Non-agent causes exist at non-agent scales, where free will was never supposed to apply anyway.
There is certainly more to develop in compatibilist theory, but now the strongest argument against compatibilism does not look so strong. When proper scale is considered, compatibilism shines. When compatibilists say that your choices and actions are caused by your reasons or desires, and that your desires are caused by your character, these are real causes, and they are the relevant kind of causes. As far as free will is concerned, they are the only kind of relevant causes. And isn’t there something ‘ultimate’ about that which is the only relevant kind? We are as free as an apple is red.
Coyne, Jerry. (2019) “Why We Shouldn’t Bet on Having Free Will—A Reply to William Edwards,” Quillette, July 17.
Fischer, John Martin. (1994) The Metaphysics of Free Will. Blackwell Publishers.
Frankfurt, Harry. (1971) “Freedom of the Will and the Concept of a Person,” Journal of Philosophy, 68: 5–20. Reprinted in Fischer (ed.), 1986; Frankfurt, 1987; and Watson (ed.), 1982.
Harris, Sam. (2012) Free Will. Free Press.
McKenna, Michael and Coates, Justin. (2018) “Compatibilism,” The Stanford Encyclopedia of Philosophy (Winter 2018 Edition), Edward N. Zalta (ed.),
Nahmias, Eddy; Morris, Stephen; Nadelhoffer, Thomas; and Turner, Jason. (2005) “Surveying Freedom: Folk Intuitions about Free Will and Moral Responsibility,” Philosophical Psychology Vol. 18, No. 5, October 2005, 561–584.
Pereboom, Derk. (2001) Living Without Free Will. Cambridge University Press.