Kailuo Wang's Reviews > Determined: A Science of Life without Free Will

The following is an updated version. The original one can be found below.

This book is a misguided attempt at moral reasoning based on scientific facts. Lacking a philosophical framework that can establish connections between morality and science, the author relied on his own rather lenient intuition without realizing it. One might say that he is another victim who falls on false philosophical questions.

Sapolsky conceptualizes 'free will' as a governing element inside a body, free from physical laws, thereby qualifying it as supernatural. This intuitive definition is not inherently wrong, albeit not that useful in some philosophical views (I'll come back to this later). He devoted half of the book rigorously disputing against the existence of such a supernatural free will, of which there are many useful scientific insights. This is a view that is already readily embraced by all naturalists - for whom the whole universe is governed by physical law and "natural." And for antinaturalists, it's doubtful that any amount of empirical evidence will change their mind.

What is more problematic is when the book ventures to analyze the moral implications of the nonexistence of such supernatural free will. Had Sapolsky maintained his naturalist rigor, he would have discerned the absence of an established empirical grounding for morality as well, that is, it is not something natural that obeys physical laws (unless one subscribes moral naturalism). If one rejects the whole notion of free will due to the lack of empirical evidence substantiating its existence, they would have no choice but to reject the whole notion of morality on the same ground. This would render any moral proclamations meaningless, of which the book contains an abundance.

If one wants to reason in morality with rigor, they must start with a solid philosophical foundation rather than just their own casual day-to-day moral thinking. One of the first philosophical questions the author should've asked himself might be how morality holds significance without empirical evidence substantiating its existence. Rather, his lack of awareness in this area is utterly disappointing, sometimes to the point of frustration.

Thus, the book treats the two main subjects, free will and morality with completely different attitudes - free will with one based on rigorous naturalist principles and morality with one based on lenient, casual intuitions. Upon such an uneven footing, the moral belief system it aims to build can’t help to be incoherent.

So where exactly did Sapolsky go wrong in his moral reasoning? It’s his confusion in identities. Let me explain this confusion by analyzing his statement that you do not deserve anything because you don’t have free will. To highlight the issue, let me rewrite “you don’t have free will” as “your decision apparatus isn’t free from deterministic physical laws”. I believe this is what Sapolsky means rather than “you don't have a neuron free from deterministic physical laws”. So the statement becomes “you don’t deserve anything because your decision apparatus isn’t free from deterministic physical laws.” The crucial ambiguity in this statement lies in whether 'you' and 'your decision apparatus' can be meaningfully distinguished from each other for the purposes of moral judgment. To put it another way, what is the identity of “you”? Is it just “your decision apparatus” or something else? The obvious choice for most naturalists is that there is no distinction, your identity is synonymous with the activities in your brain, i.e., your decision apparatus, parallel to how an advanced AI is indistinguishable from its software program. If this is the position Sapolsky takes, as he seems to for much of his book, then the statement should be “your decision apparatus doesn’t deserve anything because it isn’t free from deterministic physical laws,” which is apparently problematic - should we not first understand how a “decision apparatus” might have or lack deservingness before evaluating the relevance of deterministic physical laws? If someone makes such a statement about an AI program - “An AI program doesn’t deserve anything because it isn’t free from deterministic physical laws”, a natural reaction would be “why/how does an AI program deserve anything in the first place?” A naturalist such as Sapolsky has to find a justification or basis for deservingness without concerning the notion of “freedom from physical laws.” Such a justification for deservingness will render the original statement false. Conversely, if no justification is found, then there is no deservingness to begin with, which also nullifies the original statement. Therefore this is a dead end for Sapolsky's statement, which leaves him the only other choice - a meaningful distinction exists between "you" as an entity capable of being morally judged and "your decision apparatus."

Here is a thought experiment to scrutinize the identities our moral intuitions assign to notions of control. Consider a car accident caused by a malfunction in the onboard electrical system. It makes perfect sense to say that the driver was not at fault because he could not control the vehicle due to the defective system. But what if we’re discussing a self-driving car where the “car” and the “driver” are one and the same? It wouldn’t make sense to say that the car is without fault due to its lack of control over itself, as its actions would be governed by deterministic physical laws. This suggests that the claim "you don’t deserve anything because your decision apparatus strictly obeys the physical laws" only holds up if we think of "you" and "your decision apparatus" as distinct entities and that "you" don't have control over "your decision apparatus." If we eliminate the distinction like we treat the self-driving car, the deterministic nature of physical laws becomes completely irrelevant in our moral assessment. Physical laws are exactly what the brain relies on to work. Your decision apparatus IS the configuration of all the atoms in your brain plus physical laws. If one accepts that "you" and "your decision apparatus" are the same thing, like Sapolsky did in the first half of the book, then physical laws simply cannot be deemed as something preventing the “you” from controlling "your decision apparatus". The unstated premise that "you" and "your decision apparatus" are separate underlies Sapolsky's (and any many others') conceptualization of free will as the influence "you" have over "your decision apparatus." The first half of the book rigorously argues against this assumption - there isn't a "you" other than your decision apparatus, there is no driver in the brain, "you" are the brain. But when Sapolsky makes moral statements he, under confusion over identity, brings the "you" back as an entity that can bear moral responsibilities and suggests that this "you" has none because it has no controls over "your decision apparatus".

What, then, could constitute this distinction if not a form of dualism—with "you" representing the immaterial aspect, and "your decision apparatus" the physical brain? By this interpretation of "free will," humans are not analogous to self-driving cars; rather, there remains a "driver" at the helm of the brain, an identity of the person that is beyond their brain, and it is this driver who assumes moral responsibility. Thus it’s clear that Sapolsky, who consistently refutes dualism in the first half of his book, dismissing the “you” beyond “your decision apparatus” as an illusory ghost, inadvertently leans back into it due to a lack of clear identities when discussing morality.


To end this review on a philosophically constructive note, the debate between free will and determinism can be handily resolved in philosophical paradigms in the line of pragmatism. Hereafter is my perspective, influenced by neopragmatism, especially that of Richard Rorty.

Concepts are not defined based on their truthfulness, i.e how accurately they represent reality, instead, they are defined based on practical usefulness for our goals. For example, the concept of “chair” is very useful for human beings that can sit, but imagine a world with plenty of chair shaped objects and yet no animals that can sit, the concept of “chair” would be useless and not exist in the first place. With the advancement of modern science, humans have been able to introduce more and more concepts such as cell, proton and black hole, that aim to represent elements in nature more accurately. But for neopragmatists, it's a mistake to take the accuracy of representation as the end. In fact scientists themselves, especially those who work in the micro dimensions, have learned to treat concepts as tools (their end is better prediction of measurements), unbothered by the lack of representations.

Similarly, the concept of “free will” existed long before modern science, it has been very useful for individuals and societies. We can try to clarify the definition of “free will” based on its origin and how it’s being used. We can and probably should define "free will" differently under different contexts as long as the contexts don't overlap - we do not need to give it a single definition that it represents something in nature - e.g. a neuron free from physical laws. Such a definition of free will is isolated and useless in many contexts where it disconnects from the other concepts based on “free will” but yet to be also redefined to represent something in nature.

Hence, the whole conundrum between naturalist determinism and free will is a false question due to a misguided redefinition of the concept of free will (due to representationalism). It’s time to move on.


P.S.
Thanks to Arthur Zey (see comments below) I realized the relationship between the concepts of volition and free will. The following are my thoughts on how this conundrum came to be.

When determining the moral responsibility borne by an individual's certain behavior, we take into consideration two factors: the number of alternative behavior possibilities available for them to choose from and the sophistication level of their volition process - the internal process of reasoning and choosing between those possibilities. Everything else being equal, the individual is more morally responsible if their volition process is at a higher level of sophistication. A child is less morally responsible than a grown-up thanks to this reasoning. On the other hand, given the same level of volition sophistication, the individual is less morally responsible if there are fewer alternative behavior possibilities. An impoverished starving man stealing food is judged less morally responsible than a wealthy man for the same behavior. An individual facing a single possible behavior choice is not morally responsible for that behavior. These two moral intuitions suffice for all our practical reasoning regarding moral responsibility. Note that they naturally do not concern whether volition itself has “free” alternatives or is deterministic, only the level of sophistication of it.

Then why is it intuitive that a deterministic universe and moral responsibility is at odds with each other?



I think it comes from three confusions.

The first confusion is conceptual. It is the confusion between the behavior possibilities post volition process with behavior possibilities before it. To better demonstrate what I mean, I’ll use a simplified scenario of a starving impoverished man deciding whether to steal the food. The man faces two choices - to steal or to starve to death. These are the behavior possibilities before his volition process. They are what matters when it comes to moral responsibility judgment. A deterministic world means a deterministic volition process, which in turn means there is only one possible outcome of it and one possible behavior the man ends up choosing. But this did not in any way change the fact the man had two choices to begin with before he decides. It does not reduce his choices to one. People who claim that determinism renders any decision process useless since there is no choice are confusing the choice possibilities before decision with the fixed possibility of one decision outcome. They reject determinism based on the absurdity of all decisions being useless, but the absurdity is really from the confusion, not determinism. This is apparent when we consider a robot agent. Imagine a robot programmed to make behavioral decisions on its own. No one has any problem that the program, hence the robot’s volition process, is deterministic. No one would suggest that it’s useless for this robot to make any decisions due to its deterministic nature.


The second confusion is historical. It originated when Christian philosophy introduced the term “free will” (liberum arbitrium) in the 4th century, which traditionally meant the lack of necessity in human will, resembles the idea of a non-deterministic volition process. This “free will,” I speculate, was introduced because the aforementioned moral intuitions are not satisfactory (to this Christian philosopher) due to their lack of causal relationship between volition and responsibility. The “lack of necessity in human will” sounds more causal and logical to explain why one has to be responsible for his behavior (it is not). Regardless of this speculative motivation, this historical notion of “free will” was actually introduced AFTER our moral intuitions. It’s not the other way around. If one gets this order confused, they will be tempted to believe that our moral intuitions are based on such a notion of “free will” and thus require it, causing a perceived contraction between determinism and our moral intuitions.

The third confusion is linguistic. The original meaning of “free will” gradually got lost for laypersons and became closely resembling the term volition. In day-to-day language, when people say “free will” they are referring to volition - the ability to reason and choose between alternative behaviors, free from coercion but not something free from physical laws. Volition is something that can be intuitively verified through introspection. So when people intuitively believe in the existence of “free will” they often conflate with volition. The opposite happens in the aforementioned intuitions on moral responsibilities - they think the degree of their "free will", mistaken for the original sophistication level of volition, is a source of moral responsibility. In a philosophical discourse though, the definition of “free will” is "clarified" back with the historical value, while the language of their intuitions remained unchanged. Now all of a sudden they have intuitions contradicting determinism.


Update Oct 28, 2023: added the last paragraph for further clarification.
Update Oct 30 2023: reworked to improve clarity, added a constructive part.
Update Oct 31 2023: added P.S.
Update Dec 2 2023: added a paragraph to pinpoint where Sapolsky went wrong.
Update Dec 12 2023: added a paragraph to clarify the identity issue
---ORIGINAL REVIEW BELOW---------

The main lesson is that no matter how brilliant you are, avoid writing a book on a subject that you haven't studied for years. Sapolsky defined free will as the existence of a neuron free from physical laws. Of course this is contradictory to the naturalist belief that the whole universe is governed by physical laws. But when people talk about free will, they mean different things, while the definition Sapolsky gave, upon some philosophical inspection, has no practical implications and thus is useless or meaningless to discuss. One simply cannot derive any practical guidance based on the non existence of the "free will" as defined by Sapolsky. To be more specific, when we talk about morality, we do not need to be concerned with whether naturalism is real, or whether something free from physical laws exists.

Sapolsky should have noticed that there is so much linguistic nuance and complexity in the notion of "free will" that it justifies more study into the existing philosophical discussion in this area, especially post WWII.