My recent Skeptic article, “Free Will Is Real,” has prompted a response from Gary Whittenberger, who has previously written a standalone article for Skeptic in which he takes a stance against free will.1, 2, 3 Whittenberger’s response to me consists of several distinct points. A few of them are misunderstandings of my position. And a couple others are assertions made without evidence. But one of those assertions, if true, could be a legitimate challenge to my argument. I’ll take the opportunity here to both clarify my original article by rectifying some of Whittenberger’s misunderstandings, and to address a legitimate objection which did not fit within the length of my article. My first article makes three main points:
- We are the true sources of our own actions because the whole human, not the parts, is the correct scale at which to examine free will.
- Because of an asymmetry in time, the right criterion for free will is the ability to be unpredictable in principle.
- We are fundamentally unpredictable because we each have the properties of an undecidable system.
The first point is an argument against determinists like Sam Harris who say that though the subjectively felt intention to act is the proximate cause of acting, there are ultimately other physical causes beyond our control.4 But as it stands, my first point is vulnerable to objection from a philosophical position known as “epiphenomenalism.” This view proposes that a mental event (such as intending to act) is not even a proximate cause of a physical event (such as acting). According to the epiphenomenalists, mental events do not affect physical events any more than the motions of a car’s shadow affects the motions of the car. Whittenberger certainly has not provided any argument for epiphenomenalism, but his assertion of it, if true, would pose a serious problem for the first main point in my previous article. Below, I’ll give a defense against the epiphenomenalist objection. But first I’ll address several other of Whittenberger’s assertions, which are easier to answer.
I’ll start with one of Whittenberger’s basic misunderstandings. In response to my second main point (2), He writes:
This is false because Doyle, as most free will advocates do, has confused determinism with predictability. Our choices are probably determined (except perhaps when chance plays a role), whether or not our choices are predictable either by ourselves or by other people.2
The claim that I have confused determinism with predictability, ironically exhibits confusion on Whittenberger’s part. I do exactly the opposite of conflation between determinism and predictability; I delineate them. Usually, deterministic processes are assumed to be ultimately predictable, but I describe how they can be fundamentally unpredictable. By drawing on the concept of undecidability, I show that determinism and predictability are not inextricably linked. In this way, I clearly characterize determinism and predictability as two distinct and independent properties. In my first article, I spend about 600 words arguing that predictability is the right criterion for free will. Whittenberger disputes this but does not engage with the argument at all. He offers no reason for his disagreement on this main point, other than his confused accusation of confusion.
In a similar manner, Whittenberger misunderstands my third main point (3), and asserts its negation without engaging with its supporting arguments:
What you are expected to do, however, depends on what we know about how brains work and what we know about your brain and your history. We may come to a point in the future of neuroscience and behavior science where what we expect you to do will be exactly what you in fact do, especially if determinism is true, as we currently believe.2
I argue in the “Self-Reference and Undecidability” section of my article that it is in principle not possible to perfectly predict the deliberative decisions of human beings—neither now nor in the future. The case laid out in that section is meant to disprove the common assumption that all human behavior could be predictable, given enough data and analysis. Whittenberger merely reasserts the common assumption, as if I had not yet heard of the basic idea I set out to topple.
The problem here is likely that Whittenberger has not bothered to fully digest the concept of undecidability. One of the necessary features of an undecidable system is the potential to access an infinite computational medium. That is the same thing as having an infinite state-space, as I describe in my article. Whittenberger mistakenly believes that the number of available options has some kind of bearing on the issue:
But more importantly, individual decisions in real time have a finite options-space. For example, if you are going to a Baskin-Robbins store to get ice cream, your options set will probably consist of only a few specific ice creams of a few specific flavors.2
A system with only two options can be undecidable, which means that it is impossible to predict which of the two options will actualize. This is the case in the most simple textbook example of undecidability: the halting problem in Turing machines. One must lack even cursory knowledge of undecidability in order to think that a finite options-space rules out unpredictability. As the one who introduced the concept to Whittenberger, I am partially responsible for this failure.
Now for the next family of missteps, we turn to a concept introduced by Whittenberger: the “decision-making mechanism” (DMM).2 The DMM is a hypothetical part of the brain that supposedly makes decisions. It incorporates information from all other parts of the brain in order to make its decisions, yet those other parts of the brain are arbitrarily excluded from Whittenberger’s definition of the DMM. According to Whittenberger, the other parts of the brain store representations of our genetics, environment, life experiences, and their interactions, while the DMM takes those representations as inputs for a complicated algorithm that spits out a decision. So the difference between our views is that I claim decisions come from the functioning nervous system taken as one whole entity (a person), while Whittenberger insists that the decision making part is partitioned in some important way from the other parts.
I give a reason in my article for why I think the whole person is the right unit of analysis; when analyzing the will, we should be talking about what most people mean when they say that they will to do something. They are talking about desires, intentions, decisions, and commitments that are experienced. Most people would agree that the thermostat on my wall should not count as having a will. The reason is that the thermostat does not experience anything, though it does make “decisions” in an impoverished sense of the word. Will has dimensions of experience. Whittenberger equivocates between willful choosing and “choosing” in the thermostat sense. Willful choosing involves the nuanced textures of experience that can’t exist without a whole person, including a qualitative integration of memories, present senses, and imagined futures.
Whittenberger offers no argument against the whole person as the source of decisions, and he offers no argument for the DMM as the source of decisions. Such an argument would surely be needed since the concept of the DMM seems incoherent under scrutiny, as I will now show. If the DMM takes inputs from other parts of the brain where those inputs are represented, then where does the DMM represent the inputs which it receives? The DMM is supposed to integrate memories into the decision making process. Memories are stored in a distributed way, involving all of the brain areas which were involved in the original experience of the remembered event.5 There is no other way for the brain to represent memories. That means the DMM, which by definition excludes most parts of the brain, has no way to represent the memories which it supposedly uses in its algorithm.
The only apparent way out of this problem would be to say that the representational parts of the brain send compressed gists of memories, not full episodic memories with their visual dimensions, emotional dimensions, auditory dimensions, etc. to the DMM. But this isn’t really a way out. In order to construct a compressed gist of a memory that works for decision making, it needs to be determined which aspects of the memory are relevant to the decision at hand. But in order to assess the relevance of the various aspects of memory, the whole memory must be directly available to be assessed. And relevance is not binary; the more relevant and less relevant aspects of memory should be registered as such in the making of any decision. That relevance ranking is one of the main things that would be needed from a decision making algorithm. Such an algorithm can’t function without direct access to full memory representations, which are distributed throughout the whole brain. There is no way for Whittenberger’s postulated DMM to do this. The whole person is a far more plausible decision maker.
The person at least exists, which can be proven to anyone who tries to coherently doubt his own existence.6 The existence of the DMM can not be verified to this epistemic standard. It is arbitrarily delineated from the rest of the brain, yet unspecified in its physical and computational form. Its supposed informational inputs are implausible, which makes its basic functionality implausible. The mechanistic nature of the DMM is supposed to reduce away the human agent, but the DMM is itself presumably made of mechanistic parts with their own causal rules of “decision,” so it may as well be reduced away to elementary particles. All things considered, I don’t think I can believe in anything so mystical as the DMM.
Now I’ll move on to address epiphenomenalism, the philosophical view which, if true, would pose a serious problem for the first main point (1) in my first Skeptic article. Here is Whittenberger’s presentation:
Our conscious desires and intentions often precede our choices, but this hardly means that they cause them. It could be the case, and probably is the case, that both our conscious desires and intentions and our choices are caused by a third factor which precedes them both. And so, I believe the author is making a classical thinking error known as “post hoc, ergo propter hoc.” As we all learned in graduate school, correlation does not necessarily indicate causation, even when the correlation is sequential.2
Of course, taking such a hard line skeptical attitude toward causation would obliterate the validity of Whittenberger’s own proposed “test of determinism,” which completely depends on correlation.3 The only way Whittenberger’s DMM hypothesis could ever be supported by research on brain and behavior would be by observed correlations between localized brain activity and behavior. When scientists find that certain patterns of activity in a monkey’s dorsal premotor cortex correlate with the monkey’s subsequent choice,7 does Whittenberger declare “post hoc, ergo propter hoc”? Certainly not. Suddenly correlation does indicate causation, when it’s convenient. And it has to. As David Hume pointed out 275 years ago, there is nothing else but correlation to observe when looking for causation.8 My presumption of causal mental states is no more threatened by charges of “post hoc, ergo propter hoc” than is the presumption of causal action potentials.
The epiphenomenalist denies that the mental can affect the physical. But he seems unreasonably comfortable with the fact that the physical can affect the mental. Such one-way causation would be special in the world of causation. A bowling pin is the kind of thing that can be affected by a bowling ball because the bowling ball is the kind of thing that can be affected by a bowling pin. Actions and reactions are the rule in causation. Even a car’s shadow exerts a tiny vacuum force on the car.9 Epiphenomenalists unwittingly postulate the only exception to actions and reactions. Why should such an exception be expected? How does it work? Why aren’t there more exceptions? These questions are probably not any more tractable than the question of how the mental can affect the physical. And without answers to these questions, the epiphenomenalist is committing the fallacy of special pleading.
The draw of epiphenomenalism seems to be our mere lack of knowledge about how the mental might affect the physical. We can’t see any mechanism which connects the two. But this lack of apparent connection works both ways. That’s what’s so hard about the so-called “hard problem” of consciousness; there is no apparent necessary connection between matter and conscious experience.10 But since we know that conscious experience does exist, we should conclude that the lack of apparent necessary connection between mental and physical is only a feature of our ignorance. It can not imply that there is no connection. We should not be fooled by our mere failure to see a connection in either direction. But we do have a good reason to reject epiphenomenalism: it relies on special pleading, failing to justify its needed exception to the rule of action and reaction. This is my preliminary argument against epiphenomenalism. Other writers have already made many other arguments for and against, with the dispute showing no signs of final resolution.11 So I don’t expect this brief treatment to conclude the issue for those in the field.
But my argument should be enough to answer Whittenberger’s unargued claims. Nothing in either of his articles should lead to any doubt about the realities of free will:
- We humans are the true sources of our own actions because the whole human, not his parts, is the correct scale at which to examine free will.
- Because of an asymmetry in time, the right criterion for free will is the ability to be unpredictable in principle.
- We humans are fundamentally unpredictable because we each have the properties of an undecidable system.
About the Author
Stuart Doyle is a Force Recon Marine who has deployed to the Middle East, East Asia, and Central America. He has written philosophy and psychology papers published in The Journal of Mind and Behavior and The Journal of Libertarian Studies. He has also written for Quillette and Merion West. Stuart has degrees in neuroscience and behavior from Columbia University and in criminology from the University of Pennsylvania. Besides contending in the world of ideas, Stuart also contends in the physical world as an MMA fighter. Some of his fights can be found on YouTube.
References
- Doyle, S. T. (2023) Free will is real. Skeptic. https://bit.ly/3U8HuGm
- Whittenberger, G. (2023) A Skeptical Analysis of Doyle’s Defense of Free Will. Skeptic. https://bit.ly/3Uc6Jro
- Whittenberger, G. (2023) Testing Determinism. Skeptic. https://bit.ly/40XiUuV
- Harris, S. (2021). Sam Harris: Consciousness, Free Will, Psychedelics, AI, UFOs, and Meaning. Lex Fridman Podcast #185, 1:19:20. https://bit.ly/3zybDp2
- Wagner, I. C. (2016). The Integration of Distributed Memory Traces. Journal of Neuroscience 19 October 2016, 36 (42) 10723–10725. https://bit.ly/3U982am
- Doyle, S. T. (2022). Cartesian Dualism Does Not Commit the Masked Man Fallacy. Preprints. https://bit.ly/3Uc4KTL
- Peixoto, D., Verhein, J.R., Kiani, R. et al. (2021). Decoding and Perturbing Decision States in Real Time. Nature 591, 604–609. https://bit.ly/3ZItCne
- Hume, D. (1748). An Enquiry Concerning Human Understanding. Oxford University Press, Oxford, U.K., 2007, edited by Peter Millican. https://bit.ly/3GemOXK
- Wikipedia (2023). Radiation Pressure. Wikipedia. https://bit.ly/3GhyMjo
- Van Gulick, R. (2022). Consciousness. The Stanford Encyclopedia of Philosophy (Winter 2022 Edition), Edward N. Zalta & Uri Nodelman (eds.) https://stanford.io/3nOPnVb
- Robinson, W. (2019). Epiphenomenalism. The Stanford Encyclopedia of Philosophy (Summer 2019 Edition), Edward N. Zalta (ed.) https://stanford.io/3ZHnPP1
This article was published on April 5, 2023.
In agreement with the two comments above, I doubt that we can freely choose between any outcome, simply because our decisions are based both on past experiences and on our underlying genetic and epigenetic constitutions.
For example, let’s assume a job comes up that I am both perfectly well qualified to apply for and succeed in. However, I may have doubts about my ability against other candidates; or I may have doubts about the management or other people who work there. Now, while I am free to choose to apply or not, my prior experiences constrain my choices.
Therefore, there is a difference between what I could do and what I choose to do. Free will could be taken as the former; or in a more realistic sense, be taken as the latter, which involves constraint.
Specifically, at the biological and social levels, we carve niches, which are the sum of our underlying biology, but also a consequence of our ability to mold our environments to fit our “comfort” – niche construction.
Moreover, past experiences shape the genome in our neurons, at the genetic and epigenetic levels, influencing how neuronal circuits respond to their stimuli. Consequently, “free will” for person A is never the same as free will for person B. Everyone is constrained in the choices they make, even if such constraint is unconscious.
Addressing each point: “We humans are the true sources of our own actions because the whole human, not his parts, is the correct scale at which to examine free will.” Whole human or its central nervous system? I am unclear, here.
“Because of an asymmetry in time, the right criterion for free will is the ability to be unpredictable in principle.” – only valid in principle, where prior constraint is included.
“We humans are fundamentally unpredictable because we each have the properties of an undecidable system.” Not entirely. We are constrained by the genetic and epigenetic consequences of prior experience and inheritance.
Any confusion here stems from simply asking the question of ” otherwise ” backward. The question is not could but would and in the past if we could rewind the clock. I think it gets confusing with the word could. We feel like if an option was technically available that that amounts to “could” and therefore free will. Like I could have chosen the flavor of desert I don’t like , I just prefer the other. The fact that you prefer the other is the point. You don’t know why you prefer the other, you just do. We only can choose what we want to do most and from the options available and we don’t have control over either. So the question is , given the same circumstances, might we have chosen otherwise . I don’t see how we would have. We do have personal will, but it isn’t free.
While we may be “free” to make choices, this does not constitute free will. We routinely make a distinction between “conscious” decisions and “subconscious” or “unconscious” decisions. Where is free will in a subconscious decision? In addition, where in all these discussions is a usable definition of or explanation for “consciousness” in the first place?
“I had a dream my life would be so different from this hell I’m living, so different now from what it seemed, now life has killed the dream I dreamed.”