1. Introduction

1.1. Compatibilism

I feel more comfortable with the free-will-existence debate than any other topic in philosophy. I personally have convinced several people to, if not change their mind, at least reconsider their position. The people I convinced were not philosophers, but still.

My position on free will is the following: I am a compatibilist. In the debate about the existence of free will, there are 4 main positions:

Hard Determinism: Physical determinism is true, and free will does not exist.

Free-Will Libertarianism: Physical determinism is false, and free will exists

Hard Incompatibilism: Physical determinism is false, but free will is also impossible 

Compatibilism: Physical determinism is true, but free will also exists. The two things are “compatible” with each other.

Just for reference, Wikipedia defines determinism:

Determinism is a philosophical view, where all events are determined completely by previously existing causes.

In my own words, determinism is essentially the view that every state of reality that happens or will happen is and will be an inevitable consequence of the circumstances that lead up to it and caused it.

In the debate about the existence of free will, I am essentially conceding determinism. That is to say that, whether or not we live in a deterministic universe, my opinion of free will (that it exists) holds regardless. This position may sound a little unintuitive but I think it is actually the best description of reality.

1.2. Views of Philosophers

The view I am advocating is far from fringe among philosophers. According to a survey of philosophers, this was the breakdown:

Hard Determinism (no free will)   114 / 931 (12.2%)

Free-Will Libertarianism    128 / 931 (13.7%)

Compatibilism:    550 / 931 (59.1%)

Other:    139 / 931 (14.9%)

I am not saying that my conclusion is true just because a lot of philosophers believe it. That would be fallacious. The rest of this paper is for my actual argument. I give these statistics simply to show that the conclusion I’ve reached is not incredulous.

1.3. Target Audience

I suppose the target audience for this article is someone acquainted enough with philosophy to understand the basic gist of the topic, and to follow and engage with the arguments, but who has not yet heard a thorough defense of compatibilism.

I’m not an academic philosopher. Much of what I said may be a repeat of what philosophers have written. I will go against the consensus of some philosophers who otherwise share my position. There is some chance that I will improperly use some terminology or not be aware of certain terms I should be using. I accept that; just keep that in mind.

My purpose in writing this piece is 1) to argue my position in an approachable and understandable way, and 2) to archive my opinions so that I don’t forget my own main arguments.

1.4. Further Comments

I have written about this topic in a previous post. However, I don’t think my previous post (probably now deleted) was very convincing. It consisted of a hodgepodge of vague points that assume philosophical pre-work and existing agreement on the part of the reader. It is less “here is a thorough argument” and more “here are some things to consider that helped to convince me”, and even on that measure it’s limited. For literally years, I planned on writing a more philosophically vigorous explanation of my position, but I put it off because I knew I had to start from scratch.

One of the main ways I differ from many compatibilists is that I don’t think I am trying to “redefine” anything, as many hard determinists (here on, “incompatibilists”) criticize. I believe in the existence of the kind of free will that the general public believes in. And I will try to convince you of what that constitutes.

You probably will develop some rebuttals while reading, but the rejoinders to some sections are answered by other sections. In an ideal world, readers would withhold full judgment until reading the paper. I hope that at the very least, readers will skim all sections to see if their rejoinder is addressed. I may be very long-winded or repetitive at points, but only because I want to be as clear and thorough as possible.

This is the kind of paper that naturally invites 100 counter-points, and a different 100 counter-points for each distinct person who reads it; if 1,000 people read it, I will hear 100,000 counter-points. And I doubt I will convince anyone with a paper. But I need to write this, otherwise I’ll go mad. 

2. The Freedom-to-do-Otherwise Thought Experiment

2.1. AKA The “Rewinding Time” Thought Experiment

There is a common thought experiment I will call the “freedom-to-do-otherwise” thought experiment, “FTDO” for short

FTDO V1: If you were transported back in time into the exact situation you were in [X] number of minutes ago, then would you be able to do differently than what you did last time, or would you have no choice but to repeat your exact actions?

I know exactly the point that this thought experiment is trying to make. However, for the sake of vigor, I am going to be intentionally anal and make objections that are easily answered.

First objection: If I went back in time, I would definitely be able to do differently, because I have the memories from what happened last time. But of course, the thought experiment implies that you do not have those memories. Let’s note that.

FTDO V2: You are transported back in time into the exact situation you were [X] number of minutes ago, additionally, your thoughts and memories are reset to exactly what they were [X] number of minutes ago. The rest of the thought experiment is the same.

Hm. I’m still not satisfied. Even if your thoughts and memories are on paper and in broad strokes the same, that doesn’t go far enough. Because of the butterfly effect, it’s plausible that even a tiny number of differently positioned neurons can lead to different thoughts in the end. Heck, even different compositions of gut bacteria can have knock-on effects for your mind and cause you to develop slightly different thoughts. In the end this can manifest in different decisions. Let’s note this.

FTDO V3: The same as version 2, but every cell of the body is reset to its state [X] number of minutes ago, including every single neuron.

FTDO V3 may seem like a mere clarification, because it is. But it is an important clarification.

I’m still not satisfied. Because of the butterfly effect, even the different positioning of a tiny number of atoms can in theory lead to a cascade of differences which result in a different decision in the end. Let’s note that.

FTDO V4: The same as version 2, but not only is every cell reset to its state [X] number of minutes ago, but that entails also that every atom in the body and brain is reset to its exact previous position. We can go farther and say that this is true down to subatomic particles, even subatomic particles yet to be discovered, such that every physical property of you known or unknown is exactly reset.

Here is where I interject that this is a physics question. If it was a biology question, then FTDO V3 would be sufficient. If it was a psychology question, then FTDO V2 would be sufficient. If it was a sociology question then FTDO V1 would be sufficient. But it’s not. The fact that all of those clarifications are necessary means that it is a physics question. The more applied academic fields are not useful for answering the thought experiment.

I would never go into a physics conference and try to make a point about free will, because I don’t believe compatibilist free will is relevant to that setting. Frankly, physicists don’t talk about “free will” either; they instead refer to specific scientific theories such as the No-hiding theorem.

The point of this is that libertarian free will is only relevant in a specific level of abstraction. That level of abstraction is where physics lives, and where the truth or falseness of determinism is tested. Thought experiments like the one above tell us little about other levels of abstraction.  

2.2. Quantum Indeterminacy 

People take different positions on the truth of quantum indeterminacy, but it is unclear what if anything it would imply either in favor of or against free will.

For free will: if quantum indeterminacy is true, then the world is not deterministic, thus undercutting the claims of incompatibilists who use determinism as an argument against the free will position. Only, the existence of quantum randomness does not get you to free will. Quantum effects, especially such as this, are generally stochastic, canceling out on larger scales. Moreover, even if it did manifest on a human scale, mere randomness is not sufficient for the quality I am calling free will. In any case, I am granting the assumption for the purpose of my arguments that determinism is true, or “for all intents and purposes true”, because the respects in which it may not be true (quantum) will not be used to support my case.

Against free will: if things are bottom are random, then in what respect do individuals have a choice? After all, you don’t “choose” things that are thrust upon you by random chance, right? My first response is the same as before: these quantum effects are stochastic and therefore do not bear on the ability to effect larger-scale changes. Let’s put that aside. The question presented here is not a weak challenge to free will. However, iit is also redundant, because the logic of the challenge is essentially the same as the argument that determinism implies the nonexistence of free will. Just swap out “things are unfalteringly random, therefore individuals do not choose” for “things are predetermined, therefore individuals do not choose.” I’ve made arguments intended to work equally well at rebutting both.

I am not a quantum physicist, nor do I closely follow debates in quantum physics, so I have no idea whether quantum indeterminacy is true. However, the implications of indeterminacy, but for and against free will, don’t amount to much.

3. Structure of the Debate

3.1. This is a Semantic Debate 

It is possible to make a distinction between the (in my opinion poorly named) “libertarian free will” and what I call the “compatibilist free will”.

Libertarian free will posits that a force not entirely caused by the brain’s input’s and and the interrelated brain functions would have to exist. This force would be not entirely caused by former things, nor would it just amount to something like quantum randomness – but it would somehow allow people to make decisions not bound entirely by set physical forces (in a way I would characterize as supernatural).

Libertarian free will is something that doesn’t exist, and it is what free will rejecters are typically referring to when they say (correctly, granting that definition) that “free will doesn’t exist”. Compatibilist free will is something that (I argue) exists, and it is what free will supporters are referring to when they say that “free will exists”.

Whether or not the compatibilist free will is even a coherent concept is a topic that I will address later on. For now, let us just grant that both types of free will describe valid concepts.

“Does free will exist” is ultimately an ambiguous question, as it does not clarify what kind of free will it refers to. The debate about free will, between two people who grant the truth of determinism, is ultimately an entirely semantic conflict. In the end, the empirical question (whether the universe is deterministic) is not the fact in contention.

I have seen many debates between compatibilists and incompatibilists, where both start out conceding to all factual matters, declare it a semantic debate, and then continue to argue over the problem for another hour plus. But as long as the debate is about whose model is better, what’s wrong with a debate about semantics? Just because something is a semantic debate doesn’t mean that it is not a debate worth having. 

When people say “free will”, they usually do not specify what type of free will they are talking about, unless the speaker happens to be a philosopher who is vigorous about defining their terms. Furthermore, it is not very convenient when communicating facts to the public (more on this later). It is worth debating, when people use the term “free will” as a shorthand, what it should be shorthand for.

Just an aside, some people less familiar with these kinds of debates will say, “just look up the term in the dictionary!” To answer those people, that doesn’t work; philosophers do not use dictionary definitions for critical terms. That’s because dictionary definitions are too vague to be philosophically useful. The dictionary isn’t wrong in this case (although it can be), it’s just not specific enough.

Most of the time, when someone uses terminology, we can infer based on usage what the term means in context. However, here we run into trouble. There is a “prevailing context”, a most common usage of the term, and I disagree with incompatibilists about how to interpret that prevailing context.

We are deciding which concept should take precedence in the process of defining a shorthand term. I can only think of three factors that can be used to mediate that dispute:

1. Define the term descriptively based on prevailing usage on the part of the public

2. Define the term descriptively based on prevailing usage on the part of contemporary academic philosophers (I will skip this, because I have already given a survey of philosophers)

3. Define the term prescriptively based on which concept has the most utility to philosophers

4. Define the term based on historical usage, and taking into account how the concepts have impacted the history of ideas

I will go over each of these. However, let me say on the outset that I will be taking the perhaps controversial approach of assigning the most weight to #1. The reason for this is that, as just stated, the main reason this debate matters in the first place is because the question poses a challenge when it comes to communicating with the public. If the term “free will” was unknown outside the pedigree of professional philosophers, then it would not be important to solve, because professional philosophers can be expected to define their terms precisely in each paper they write anyway, which, although inconvenient, obviates the need to ascertain the “prevailing context”.

3.2. Structure of this Paper

Section 1 is the introduction. Section 2 is the first part of my argument (but still preliminary). Section 3 is is setting the stage.

Sections 4 and 5 take a brief detour; I’ll speak more generally about knowledge disciplines, clashes between them, and how they relate to philosophy and the topic at hand.

Sections 6, 7, 8, and 9 dig into the “semantic debate”, the work of deciding how we should define free will. This will cover the checklist list I just gave from 3.1. The most important sections are 6 (who’s stealing who’s definition?), and some of 9.

Sections 10 and 11 start to delve into grounded, real-life manifestations of free will.

Sections 12 and 13 cover more technical, meaty, cognitive science topics where I’m taking a risk by even hypothesizing. They are necessary because for my argument to hold water I need to say what I think free will is.

4. Semantic Differences Between Fields

4.1. Fields Can be Ordered by Different Levels of Abstraction

You might have seen this XKCD comic ranking fields by level of purity.

If I had to stack-rank the scientific fields from most fundamental to most emergent, I would put them in this order:

A. Physics (the study of everything)

B. Chemistry (the study of matter)

C. Biology (the study of living matter)

D. Psychology (the study of how living things think)

E. Sociology & economics (the study of how thinking animals behave in large numbers)

F. Applications of sociology, e.g., criminal justice, public policy, etc.

As we go “down” the stack, we move from the more broad (yet somehow more concrete) fields to the more narrow (yet somehow more fuzzy) fields. Each listed discipline is a slightly narrower application of the discipline listed right above it. Chemistry is a slightly narrower application of physics, and so on. I skipped intermediary disciplines like biochemistry etc.

In this paper, I will refer to the fields higher-up the stack as more “granular”. I think that word is appropriate because you arrive higher-up by analyzing phenomena in more fine-grained detail. For example, if you want to understand the motions of balls, you can deconstruct them to individual atoms. An atom is more granular than a ball. This should not be confused with taking “granular” to mean a focus on individual data points at the cost of abstraction. 

4.2. Dice Analogy

There are times when it is difficult to tell the definition of a word from context. I am going to compare something to free will, but be careful, it is possible to take too much from the following analogy. Pay attention to what I’m comparing (that both terms run into semantic problems) and what I’m not.

The analogy is to the term “randomness”. I pose the following question: is a die roll random? This is surprisingly difficult to answer.

From the perspective of physics, the answer is no. According to physics, that is to say, according to determinism, a die roll is not random. If we perfectly knew everything about the momentum of the die, the physical properties of the die, air resistance, the friction of the table, etc., and had infinite calculation capacity, you could perfectly predict, from Newtonian physics, exactly what the outcome of the die roll would be.

From the perspective of board game players, the answer is yes. This is because, for all practical purposes, if someone shakes 6 dice in their hand and tosses them, it is not realistically possible to predict what is going to happen. To call the action “random” is simply a description of this state of reality.

The academic fields of information science and signal processing can readily formalize the logic of the board game players. In this model, “randomness” is to be defined as a relative property, relative upon certain specific, measurable qualities of the die roll. As per chaos theory, how chaotic is action, ie., how sensitive is the outcome to small differences in starting conditions? Furthermore, how much information do the players have about the starting conditions, and what is their capacity to compute the result? Based on these questions, the resultant randomness is derived, as broadly random, broadly predictable, or something in between. In other words, “random” is used as the antonym of “predictable”, which is necessarily conceptually contingent on what or who is being called to make the prediction.

It is not that one definition of randomness is right or wrong. It is merely a semantic conflict between the fields of statistics and physics.

It is not the case that the die roll is actually non-random, but board game players merely are operating under the misconception that it is random. No. According to their definition of randomness, a die roll is truly, literally random. A die roll is both random and non-random. How can this be the case? Because there are multiple different definitions of “randomness”.

It is a problem of homonyms. Some words mean different things in different disciplines. For example, the term “cell” means something in biology. However, the term “cell” also means something in criminal justice, as in a prison cell. The listener can easily derive what is meant from context. In the same way, we can infer what “random” means in context.

Some people strawman my position to be that “It feels like we have free will, and therefore we ought to treat it as though we actually do have free will.” To be clear, how people subjectively feel may be relevant, but my belief in free will is not based on some detached “feeling”, any more than I would say that dice are random because they “feel random”. It doesn’t feel like anything to be dice; they do not have a subjective experience. It is a property of the dice, and how they behave materially, which I identify; so too with free will.

5. Physics-Bias and Philosophers

5.1. Physics Envy and Mindset

I have written quite a few words thus far justifying the notion that privileging the implications of determinism implicitly privileges physics as a discipline over other disciplines. This is connected in no small part to the phenomenon of “physics envy”. Because, in many ways, fields are judged as important or legitimate in relation to perceived similarity with physics, many fields (practitioners therein) may feel a sense of inferiority, and try to emulate physics to compensate.

I should note that physics envy is a bias that I hold. That is to say that the legitimacy judgements I just described are the kinds of judgements I am primed to make. This is where it gets somewhat personal. Me, and many people with similar psychological dispositions to me, share such a bias. But that should not distract from the fact that it is a bias.

What is the psychological disposition I refer to? In my view, there is a specific personality type that prefers the specific over the abstract, and likes to break things down to their “fundamentals” which in practice actually means “smaller constituent components”. But missing the forest for the trees is not a virtue; it is myopic.

The mindset I describe pertains to physics, certain kinds of mathematics, and philosophy. The reason for this is that those are granular, analytic, deconstructionist fields. Hence, someone with a granular and analytically deconstructionist mindset will be attached to those kinds of fields.

However, critically, most of the general public does not share such a granular mindset. Rather, they are more comfortable with abstractions, including those that are at first glance too high-level to be subjected to analytic evaluation. That should be appropriately factored into the discussion in section 6: “what best represents the general public’s use of the term?”

Yes, I called the layman’s view more high-level and abstract. I’ll go more into this in section 5.3.

5.2. Picking Favorites

To restate the question, what terminology should philosophers favor when there is a semantic conflict between fields? An easy answer is that philosophy is about that which is most abstract, and therefore should favor the most abstract field. However, I find this answer unsatisfactory. Whereas physics is in some sense more fundamental, criminal justice (or even biology) is in some sense more emergent. Fundamental or emergent: which more means “abstract”?

There is some sense in which disciplines are indeed hierarchically ordered, which I laid out in section 4.1. They are ordered according from most all-encompassing to domain-specific. For instance, chemistry can be viewed as a narrower, domain-specific application encompassed within physics, just as biology is to chemistry. As such, physics is “earlier in the stack” than biology.

The next question is: semantically, should philosophy prioritize fields that are earlier in the stack? I think the answer is no. Provided that all claims are correct (ie, the disagreement is in terms of semantics or emphasis rather than facts), philosophers should not bias claims that are made earlier in the stack over those made later. Regardless of the perceived “philosophy-ness” differences, philosophy should not pick favorites.

Philosophy itself is not placed on the stack. There is a meaningful separation between the empirical (science) fields and the deductive (logical) fields. The empirical fields are how we gather data about the world. The deductive fields are math and philosophy, which build proofs to evaluate what conclusions follow from certain premises. Undoubtedly, the empirical fields utilize the deductive fields (you could not have science without math), and vice versa.

The extent of the overlap is actually an active debate in philosophy. Nonetheless, I believe it’s useful to distinguish e.g. math from physics.  Although physics is generally perceived as more “math-y” than, say, sociology, this is not set in stone. Math can be used in the field of sociology (e.g., statistics analysis of human behavior), and that is no less a valid use of math than any other. Mathematical proofs are tautological and agnostic to where those conclusions are being applied.

If how fundamental an empirical field is does not determine how philosophically it should be analyzed, as I assert, then what, if anything, does? I would say that what makes something “philosophical” has to do with the level of explanatory clarity and logical validity present. Forget what level of abstraction I appear to be touching; what really matters is whether I can be descriptively specific and logically consistent.

5.3 But Shouldn’t We Privilege the Fundamental?

I hear this criticism:

There is an order to it: the more fundamental fields are what establish the logic for the lower fields. How the “lower” fields operate is determined by the characteristics of the “higher” fields, so we should prioritize the higher fields to get to the root of things. Physics and philosophy is precisely about abstracting from the granularity, to get to the very top of the tree diagram of causes to reach a single axiom. At a less-abstract level, things may appear to be a certain way that logically contradicts the highest axiom under which it falls. This does not mean there are two shared contradicting realities. It is just that the appearance is incorrect, because it does not take into account the various other branches of the causal tree (to which the single granularity is ignorant, but the overall axiom is not). Going back to the dice analogy: physics should get the final say, and statisticians should indeed be disallowed from saying they are random. We can let them use the term as a shorthand, but it’s foolish to make too much of that.

This argument appeals a lot to me instinctively. The “higher” fields are about “how the world really is”. (The order is arbitrary; I could just as well have put physics on the bottom and sociology at the top. The only important thing is that the fields do come in that order.) With contemplation, however, I decided it is wrong to prioritize certain fields over others in semantic conflicts. 

I like this topic because it relates to the broader topic of emergence, and how it is sometimes useful to talk about complex structures as more than the sum of their parts. For example, suppose someone says, “The company published a report”. What does that actually mean? Well, the person who actually clicked “submit” on the report was probably some miscellaneous administrative assistant. Maybe the real question is, “Who authored the report?” Well, the report was filled-in using a pre-existing template. Ok, who wrote the template? It was written by a team of MBA hires, who in turn were taking instruction from managers, who were taking instructions via email from higher-ups… it goes on and on and on. Potentially a thousand people were involved in some way, all motivated by a web of incentive structures implicit in the larger organization. So to try to drill down on a question like, “What person is responsible” is less useful than simply conceding, “The company published the report.”

In asking, “do you have free will”, it is possible to drill down on the constituent components (which is necessary to establish determinism). I don’t think that’s very useful for any conversation topic more general than about free will itself. Words exist because they allow us to effectively communicate some facts. Those facts may be emergent, high-level, fuzzy, socially contingent, or whatever; the important thing is that information is being communicated.

6. What Definition Most Closely Matches Public Understanding? (Or: Who is Stealing Who’s Definition?)

6.1. Descriptivist Linguistics

The arguments below may seem a bit strange, like argument-ad-populum. However, this portion of the argument is over descriptivist linguistics, which is fundamentally based on public consensus. Because of that, within the field of descriptivist linguistics, argument-ad-populum (bandwagoning) is not fallacious.

The fundamental motivation behind this: I am attempting to refute the all-to-common refrain that compatibilists are “trying to change the definition of free will” to match their claims. I argue that no, “free will” has always meant something of compatibilist nature according to descriptivist linguistics, and in fact, if anyone is trying to change the definition of free will to match their claims, it is the incompatibilists.

6.2. Exploring a Public Refrain

From the general public, it is extremely common to hear some version of the following refrain. I don’t feel like digging up a bunch of examples of people stating it, but the fact that it’s common is critical. If you disagree that it’s common, tell me and we can set up some sort of empirical experiment. But I think most people can be honest and say they’ve either thought it or heard someone who’s thought it before. I’ll call it “The Layman’s Restrain”:

If we have free will, what’s the point in doing anything? What’s the point of getting out of bed? Why even avoid bad behaviors if you had no choice?

Incompatibilists have a few different ways to respond to these questions. At best, they methodically walk through the rationale of their model. They explain how the superficially confusing state of affairs actually obeys a coherent philosophical logic, and how the above questions are therefore not the smackdown-objection people think they are.

Incompatibilism does have answers to The Layman’s Refrain. But before I address those, I want to comment on the relevance of it, and why I call it that.

It is tempting for incompatibilists to respond to The Layman’s Refrain in a dismissive way. If someone asks the above questions, without anticipating the natural best replies, it indicates that that person is not aware of the natural best replies. Therefore, to be frank, the refrain typically indicates that the questioner is a layman in the sense that they are not very familiar with the ins-and-outs of the debate. They represent a relatively surface-level critique of the no-free-will position.

But that is exactly why I choose to highlight The Layman’s Refrain here: the fact of being a layman is useful and pertinent to the question at hand, which is how words are understood in a descriptive sense. Most people are laymen. By calling them that, then, I am if anything assigning more importance to their view. I seek a sample of the statements of the general population in order to explore the usage of language broadly.

Now, we can analyze responses. This is one possible response:

Even if you stay in bed, it will be of no free will of your own! Just the same, if you get out of bed, it will be of no free will of your own! 

I usually hear this in the tone of, “I’m joking but I also really believe this.” That’s not the best response because it does not actually answer the question. The question is about what is the point of getting out of bed, which this response does not answer directly.

The following is a better, and incompatibilist, response:

On a physical level, there is no “point” to doing anything, because to do something for a “point” implies an external intent, which does not actually exist. It would be like asking: “if the wind can’t control the direction it blows, what’s the point of it blowing mostly to the right, when it can instead blow to the left?”

This is better, but someone could still ask, “Then why should I get out of bed?” To that, an incompatibilist can say:

Even if we have no free will, it remains the case that one’s actions affect their life circumstances, and therefore it is good advice to work hard, if the aim of the advice is to improve the life circumstance of another. The possibility of being motivated by advice does not refute the position, because the act of giving advice itself, and mechanisms by which one responds to it, is all deterministic. This does not take away from the fact that your actions determine your circumstances, and it feels as though you decide your actions. So when operating within your world, you should still be productive.

That is a fine answer, and correct within the incompatibilist model. How does the public respond to it? It is often met with confusion:

Hey! You just said we don’t have free will! And now you’re saying that people can affect their own lives?

I contend that the source of the confusion on the part of the public is that they are operating under a definition of free will quite similar to the compatibilist one. When one says, “we don’t have free will”, people interpret that as saying, “one cannot make self-directed choices”, which perhaps in some narrow, down-to-the-atom and all-of-history sense they don’t, but that’s not pertinent to the essence of the question at hand.

Under the compatibilist model, by contrast, the essence of the Layman’s Refrain is maintained. It is not rendered easily dismissed, or nonsensical, or irrelevant. Rather, the refrain is sensical, because it touches on free will as defined by the compatibilists. In short, the ubiquity (the layman nature) of the refrain is an indicator that the general public is operating under a definition of free will which most closely maps to the compatibilist framework.

6.3. Other Common Refrains

Let me deliver some cases-in-point.

This tweet reads:

If we don’t have free will in the ultimate sense, as some philosophers claim, does that mean I never consent to any of my thoughts or actions? Isn’t that problematic?

And in this video, at 5:40, the speaker says:

For example, our justice system depends almost wholly on the fact that we consider people personally responsible for their actions, but if I were to disseminate to the world the rather robust neurophysiological evidence that free will does not exist, this could be a great “unveiling hazard” to a great many people and institutions.

I give these examples not because they are extraordinary, but rather because they are ordinary. The views presented are not particularly interesting. I present them because they are good representations of how the general public understands the concept of free will.

The general public believes that the non-existence of free will would pose a problem for the notion of consent (in any respect). The general public believes that the non-existence of free will would pose a problem for the ability of the justice system to hold people responsible for their actions.

In other words, the claim “there is no free will” is true given the hard-determinist definition, but the general public seems to understand “free will” to have the compatibilist definition.

I have seen many more examples than the two presented above. However, I only started saving links to them when I started writing this article, to say nothing of IRL claims. So I only have those two examples, sorry!

The incompatibilist may respond to those questions, that the public is drawing the wrong conclusions. They could say “no, the nonexistence of free will does not necessarily imply that there’s no such thing as consent.” They could say, “no, the nonexistence of free will does not necessarily imply that there is no utility in putting some criminals in prison.” And they would be right, but they are fighting an uphill battle, and dodging the very real issue of what is causing so many people to develop those misconceptions. The issue is semantics. The compatibilist framework/definition is, it seems to me, more natural for the general public; it’s what they initially grasp.

6.4. An Anecdote

At one point in college, I believed that free will did not exist. In a philosophy-ish class, the topic came up, and I started making the anti free-will case. In these kinds of classroom discussions, I was always the obnoxious student who loudly voiced their opinion on everything and disagreed with the professor. Some of the students on the other side of the classroom who had been discussing quietly amongst themselves finally spoke up. One of the students said, “You have contradicted yourself! At one point, you said stridently that free will doesn’t exist. But at another point, you said that we should hold people accountable for their choices. What is it!”

At that moment, I was tempted (and started) to launch into a weaselly explanation about how the incompatibilist nonexistence of free will was, in fact, compatible with a criminal justice system that holds people accountable. However, I soon realized that my insistence in the nonexistence of free will was causing more confusion than it was worth. My insistence upon the incompatibilist model was, at best, over-philosophizing / bloviating, and at worst, communication of an untruth given the other students’ understanding of terminology. I didn’t have the time for that, literally, because class ended before my classmate even finished listing off his disagreements.

In my view, it is easier to disambiguate a concept “determinism” from “free will” than it is to disambiguate a concept of “free will” from “accountability”. That is truly the most critical thing. Most proponents of the no-free-will view do admit to the utility of certain forms of accountability. In so doing, they already subscribe to a kind of compatibilism: compatibilism between no-free-will and accountability. I believe there are many fascinating cognitive properties sociologically associated with accountability that are worth discussing. That is why, compared to an incompatibilist starting point, I advocate for definitions which shift the “x is compatible with y” case up-the-chain, to the point in the model where “free will” is first introduced.

6.5. Concessions to “Freedom-to-do-Otherwise”

I have to make a concession about the freedom-to-do-otherwise thought experiment, specifically V4. And that is that most members of the general public would be surprised to learn about the deterministic consequences of FTDO V4. It is worth exploring how much that matters and why that is. This section is at an even higher risk than 4.4 at sounding pretentious, so be prepared.

FTDO #4 is quite unintuitive to most people. This, in my opinion, is because most people are understandably unaccustomed to thinking through abstract things down to the level of physics. I explored this more in section 5. Although the laws of physics govern all of our lives, general day-to-day activities do not actually require knowing the laws of physics at a granular level, because most people interact with the world at a higher level of abstraction. FTDO does not actually require knowledge of any specific laws of physics (except perhaps the no-hiding theorem, but even that is debatable, because one needs not have heard of it as such), however, FTDO does require the mode that deconstructs reality down to the most fundamental components.

The real question is: how much does that matter when it comes to our interpretation of their use of the “free will” term?

Most people believe they have freedom-to-do-otherwise. However, there are many types of FTDO (see section 2): V1, V2, V3, and V4. We established that they certainly do not have V4. However, I contend that FTDO V3 & V4 are more of a property of free will (a false one), rather than a core aspect of its definition. I will give an analogy.

There is a popular misconception that The Great Wall of China is the only man-made structure visible from space. This is not true; there are manmade things visible from space, but The Great Wall of China is not one of them. If you asked someone, “what is The Great Wall of China?” they might tell you, “you know, it is the big wall in China, the one to keep the Mongols out, that is visible from space.”

One option is to take the definition literally as it stands. We can unthinkingly define “The Great Wall of China” as “noun: that wall in China that is visible from space”. Then, as a consequence of that, we have to conclude that The Great Wall of China does not exist! No object matching that definition exists. This is the consequence of a prescriptivist approach that crowdsources the definition from the public rather than observing public use and correctly characterizing it.

It would be foolish to do the above, and conclude as a result that The Great Wall of China does not exist. The fact that the public believes a misconception about something does not mean it doesn’t exist. We can solve this by distinguishing between properties that are part of the definition of the object, and properties that are descriptions of the object. “Visible from space” should not be part of the definition (accurate or not); it is a description after-defined. Likewise, I believe FTDO V3 & V4 is merely a description of free will after-definition.

FTDO V4, specifically, is something I believe should not hold too much bearing on the definition of free will. The reason is, put bluntly, I doubt most people care so much about that philosophy/physics-heavy question, as opposed to other questions I will explore in other sections which they might consider more important.

In this way, speaking candidly, I view FTDO V4 much like a parlor trick. It is an interesting philosophical question for certain reasons. For minds inclined towards philosophy, reasoning through it is actually a formative experience. It sure was formative for me when I first realized, after considering FTDO V4 (among other thought experiments), that determinism was likely true. Unfortunately, it may be more interesting for some than others; even for me, compatibilist free will (as elucidated in section 12) wins out as slightly more interesting. And so, as much as an intrigue FTDO V4 is, I don’t like how incompatibilists wheel it out as a shiny object with which to say, “free will refers to that!” To do that is to take a staple of intro-to-philosophy material and simply stop there, rather than progressing farther, onto the rich and largely distinct domain that is free will research.

7. What Definition Carries More Utility for Philosophers?

7.1. Redundancy of Incompatibilist “Free Will”

What is “free will” to the incompatibilist? The term “determinism” already exists, so if an incompatibilist definition of free will is to distinguish itself, it need mean more than simply the negation of determinism, else it be largely redundant.

Of course, those who believe free will does not exist do not only disagree with the free-will libertarians. They also disagree with the compatibilists like me. However, here there are nuances. For our purposes here, there are two kinds of rejects to compatibilism, which I will call type 1 and type 2.

Type 1 rejection of incompatibilism says that what compatibilists describe may be a real phenomenon, however that phenomenon should not be (or is not) called free will. The compatibilists are not necessarily misidentifying properties of the world, they are simply using the wrong labels; free will is a term that should be reserved for other purposes. From type 1 we hear the argument, “compatibilists are trying to change the meaning of words.”

A type 2 rejection says that compatibilists fail to describe any real phenomenon. They would make the claim: the emergent property which compatibilists characterize, even understood on compatibilists own terms, simply does not exist as a fact of the physical world.

A type 2 reject is much more fundamental than semantics. It concerns a factual matter which I hope to argue through in large sections of this article. To accept the type 2 critique would be to presuppose they are correct on more than semantic matters, which is off-topic for this section.

I feel that, so long as there is a sufficiently important concept that is adequately served by a term “free will”, then it is advisable to lend the term to that concept, as opposed to denying the term much meaning outside of the claim that it doesn’t exist. That is what, in the long run, conveys more utility of philosophers.

I don’t want to say that the incompatibilist terming of free will is totally useless, because I can think of contexts where it might grant some utility (like if you’re discussing the intersection of determinism and decisions), but those contexts are highly specific, and we’re better off just using different language in those cases to avoid confusion. I consider compatibilist terming more generally practical for a broad array of discussions. 

7.2. Supposed Redundancy with “Will”

This is to address a criticism: 

The term “free will” is redundant because we already have the term “will”, and the term “will” already sufficiently captures whatever compatibilists mean by “free will”. Therefore, a compatibilist terming of “free will” is unnecessary.

If you don’t have an objection similar to the above, you can skip this section.

I counter that they do not mean the same thing. Let’s suppose that the CEO of a company sends an email to his subordinates, instructing them to “liquidate holdings XYZ.” One can accurately say that “it is the will of the CEO to liquidate holdings XYZ” because his authority is behind it. However, one cannot say that it is the “free will of the CEO”, because that is not known. It could be that the CEO does not want to do it, but is being forced to by his board of directors. He could resign in protest, but perhaps he needs the money to cover some debts, and the selloff would happen anyway.

It makes sense that “will” should be connected to “free will”. After all, one half of the words in “free will” is “will”. Certainly, they are related! But they communicate different concepts. “Will” is a straightforward descriptor: it connotes simply that someone is putting their authority behind something, eg, they want something to happen not just authority, but fact of decision. “Free will” must also consider one’s internal psychological state.

There are some situations that become “edge cases” only under the guise of free will. Someone can have a “will” while not being of sound mind (e.g., being mentally indisposed), but it is debatable whether they have free will. Large organizations like countries and corporations can have wills, but they do not have psychological states. Corporations debatably do not have free wills, at least not in the same manner that a human mind does.

8. Defining Free Will Based on History

8.1. Cursory Look at History of Compatibilism

In one argument that I had with an incompatibilist, my debate opponent brought up how free will was talked about historically. The thing is, compatibilism has a long philosophical history. 

Compatibilism was advocated early by Stoic philosophers: 

The Stoics put forward two theses that apparently look contradictory: on the one hand, they maintained that the world was governed by providence and, accordingly, that all the events were fated to happen. On the other hand, they were willing to assert that individuals are responsible for their own actions

Boethius in 523 gives the example of a man walking vs the sun setting. Whether or not those events take place in the past, the present, or the future, the former takes on a different character than the latter. He advocated for compatibilism: a man can freely choose to walk down the street, despite the fact that God’s perspective – being able to see all of history at one moment – would seem to imply it was inevitable.

Compatibilism became huge under Thomas Aquanas, who was a big advocate of his own brand of religious compatibilism. They believed that human action could be free, despite the fact that they technically could not have done differently according to the more fundamental laws of the universe. Compatibilism continued throughout the Medieval era, advocated by the followers of Thomas’s philosophy. 

In the early modern era, some prominent compatibilist philosophers were Thomas Hobbes, David Hume, and John Stuart Mill. Someone could write an entire book about the topic of the history of compatibilism.

But this isn’t entirely fair, because ideas of compatibilism have adapted through the ages. Compatibilism historically is going to differ from contemporary compatibilism. But that is normal and expected with every philosophical theory. 

The argument against me is that I can’t use early advocacy of religiously motivated compatibilism towards my modern definition. But I’m not so sure of that. See how a branch of religious view, Physical Premotion, is described:

More broadly, according to this Thomistic theory, physical premotion is the causal influence of any principal cause upon the respective instrumental cause (such as the influence of a scribe upon his pen) by which the instrumental cause is elevated so as to be capable of producing an effect which is beyond its natural powers (e.g., the pen is enabled to write a poem).

For all that, this sounds quite naturalist, as opposed to mind-body dualism.

A mind-body dualist form of compatibilism held by many theologians is contrary to my view. However, many ancient philosophers, including theologians, advocated for naturalist views of free will. Therefore, it is not true that a usage of the term “free will” like mine is some modern invention.

8.2. Importance of the History of Philosophy

I tend to think of contemporary Philosophy and History of Philosophy as two different disciplines. Contemporary Philosophy is debate on philosophical topics using philosophy. History of Philosophy is a branch of history, one that can help one’s thinking about the former subject, but history nonetheless.

Some elevate the History of Philosophy discipline to the principal position. I do not. This paper is actually a contemporary philosophy paper; I am concerned principally about how the term “free will” is used within contemporary Philosophy discussions. I am less concerned with how historians may characterize particular past thinkers.

Past thinkers should be labeled, in most cases, based on how they labeled themselves. If historical self-labels are outdated or mismatched with modern terminology, their positions should be described for clarification. But mismatches are allowable provided that clarifications are made.

In discussion about history of philosophy, the meaning of the term will depend on the philosopher being discussed. When describing a compatibilist philosopher, it will refer to a compatibilist definition; when referring to a hard determinist philosopher, to a hard determinist definition. I do not know of any controversy of how to employ the term “free will” in history of philosophy discussions. This paper is about otherwise-ambiguous cases in modern discussions.

9. But What is Free Will?

9.1. Definitions

I have come a long way without actually defining free will. Here is an attempt.

Free will is the ability of an agent to act in a motivated yet non-coerced manner. An action is motivated if an agent has a motivation or motivations for doing it. A motivation is a reason to take an action that the agent consciously or unconsciously considered (in short, that the action was the result of a decision-making process).

An action is coerced if the causes for the action were reducible to forces outside of the agent. I define my concept of reducibility in section 12. “Reducible to” is not the same as “caused by”. Technically all mental processing was caused (set into motion) by forces outside of the mind if we trace the chain of causes far back enough. Reducibility is different, residing closer to the domain of information theory concerned with how one clusters information about systems.

9.2. Some Immediate Implications

If you do not have free will, it means you were “forced” to do it by external circumstances (those external circumstances need not be conscious). Those external circumstances could be some other agent (some other person forced you to do it), or by circumstances more generally (you were forced to do something to escape falling debris). Here “forced” just means coerced.

In both cases, the cause for your actions is reducible. In the first case, your sequence of events, including your actions, are reducible to (fully explained by) the actions of external agents. In the second case, the sequence of events (the debris fell, and you did what you could to escape) is reducible to the falling debris, with little options for free choice on your part. 

In both cases, there existed such a strong reason or reasons to do the action such that you were definitely going to do it (you were “forced” to do it) regardless of any sociologically-defined conditions that could otherwise be the basis of alternate counteracting motivations. “Sociologically defined” is a for-now ambiguous category that I use more so to refer to a level of complexity/abstraction/ambiguity, and less so the current state of sociology as a discipline. A hermit with no social contact can still have free will. This will be explored more precisely later, but I’ll tentatively simplify the model, and say there’s some indeterminately-high level of complexity/opaqueness, exceeding which grants potential for free will. But what I am here referring to “complexity/opaqueness” does not apply to machines, because more than just being internally complex, it is also about sensitivity to a complex external environment.

9.3. My Multiple Definitions

At different sections of this paper, I may seem to present free will differently: defined based on freedom to do otherwise, defined as an emergent sociological phenomenon, defined based on response to deterrence, defined as a sensation, and so on. Although these may seem to be competing, they are actually descriptions of different aspects of the same concept. I believe it is important to describe the phenomenon from various angles in order to develop a holistic picture. Not all sections are meant to pinpoint-define free will. My actual definition is based on coercion, as I’ve recently put it, where my definition of coercion is based on criticality vs reducibility, as I will describe in section 12.

9.4. How to Communicate Meaning

There are many ways to communicate meaning. Definitions are one way. However, definitions are not actually how we learn most of the words we know. We learn most words through gesturing, eg, your mom says, “that is a cat”. I will flesh out the concepts of my definition, but my larger goal is to communicate the concept through many means and many methods, so you grasp my meaning fully. Whether through examples, or by speaking to the implications, or to get it into your head by some other means, I am indifferent to the method of intake.

It could be that, for concepts of sufficient philosophical abstraction, it is difficult to impossible to communicate the full meaning simply by succinct definition, because of the limitations of the language itself. For instance, it is hard for me to create a succinct definition for “sociologically defined”. Nonetheless, I’ll try to make my argument coherent with the language I have available.

My aim over all of this is simply to convince you that free will exists. Therefore, if your definition of free will differs from mine in the end somewhat, or perhaps one of the ways I approach the topic doesn’t resonate with you, then I have not failed so long as you believe in the end that “free will exists”. It is not about having a perfect description of the property, merely about conceding that there is some property that rises to the level of “free will”, and that that is a relevant distinction.

9.5 Competing Theories of Compatibilism

Some portions of this article are dedicated to making the case for a specific conception of free will. Other portions of this article are laying out a broader conceptual framework challenging the notion of indeterminism as the one-and-true view.

I acknowledge competing explanations, other “versions”, if you like, of compatibilism. I will not put virtually any work into dispelling them, because we agree on the fundamental claim of “free will exists”. Of course, that claim means different things to them as it does to me.

If you disagree with “my” compatibilist free will, but hold to “your” version of compatibilist free will, we can have that discussion separately. I suspect that our versions are likely overlapping in some respects anyway. The scope of this article remains simply: #1, establish the case for my version of compatibilism, and #2, to take on incompatibilism, as (for the most part) the only view I will challenge directly.

9.6. What it Means to be “Motivated”

Earlier, I used the term “motivated”, so I will have to describe it. 

Motivated is the opposite of incidental. Something is motivated if it’s aligned with a goal. Normally, this would open the can of worms where I would have to define what a “goal” is. However, since virtually everyone recognizes that goals exist, I don’t think that’s necessary for convincing people. 

Two scenarios:

Scenario 1: You are a knight. You are chasing your enemy through the woods. You trip on a tree branch. Was the tree branch “motivated”? No, it was only incidentally in a position to cause the tripping. 

Scenario 2: Suppose that you are a knight, and swing your sword. Your blade is dull, so it doesn’t cut through your opponent’s armor. A different knight swings a sword that is identical in all ways except that it is sharper, so it pierces the opponent’s light armor. In this example, the differences in the swords made a difference. Were the swords “motivated”? 

The most intuitive answer is no. The swords had no motivations of their own, they are just metal instruments. 

However, the swords were made by a sword maker. The blacksmith wanted to make sharp swords, because it was his source of income, and therefore his method of feeding himself. By feeding himself, he can survive. So although the sharp sword did not have motivation in itself, it was motivated instrumentally; it was caused by a motivated “first mover” of the story.

Further upstream, the reason the sword maker wanted to survive was because of features of his psychology determined by genes. I am cautiously comfortable calling genes motivated, even though genes do not have minds, and can therefore pass no opinion. I am employing the term motivated to mean operational motivation (driving complex adaption to the environment), rather than psychological motivation (the ability to actually make plans).

With evolution by natural selection, the fact that a gene exists is a result of how it is. This is in contrast to most systems. The fact that a nebula exists does not result from the properties of the nebula, but rather the conditions prior to the nebula which led to its formation. 

You may accuse me of constructing this definition so as to unfairly privilege life. However, my use of the term is loose. Note that according to this theory, one can also label downstream, instrumental objects as “motivated”, so long as they were aligned with the motivations of the creator. 

10. Examples: Coerced vs Non-Coerced Situations

10.1. The Brain Tumor

A man starts behaving strangely. He goes on to kill 3 people. He dies 3 weeks later, either from suicide, or execution, or any other reason; it doesn’t really matter why. Scientists dissect his brain and discover a tumor in a location that could plausibly have caused homicidal tendencies. They conclude that the tumor arose at around the same time when he started to behave strangely.

Did the man kill people out of his own “free will”? I would assert that he killed people out of his will. However, I would hesitate to call it free will. His decision to kill had a single, “outside” cause: the abnormality in his brain. The usual decision-making apparatus capable of complex cost-benefit analysis was not allowed to be at play, because it was intercepted/destroyed by the tumor.

Interestingly, some incompatibilists use the above thought experiment (based on a real case) to make an argument that free will does not exist. They say, “Every single person is just a less extreme version of the above case. It’s tumors all the way down. No one truly has any more free will than the tumor man, because we are all the victims of outside forces, just in more distributed ways”.

I use the above thought experiment to conclude just the opposite. There is a true, categorical, typological distinction between a man with a brain tumor and everyone else. Notwithstanding the possible subtle equivalences one could draw between the tumor and properties of regular brains, the imposing and centralized nature of the tumor simply screams out to you.

In a typical circumstance, a person deciding whether to commit a crime would have a lot of considerations operative in their head. Will I get caught? If so, will I go to jail? How will this hurt my reputation in front of my friends and family? How will it hurt my reputation in front of society? Will I feel shame? What will I gain materially from the crime? What will I gain in terms of satisfaction? Will it help me get out of my current situation? Etc. Etc.

In the case of the brain tumor, whatever such considerations there were, the normal complexity of the decision was thrown out of whack. A single consideration, the tumor, crowded out all else, to the point where any one other consideration (such reputation in front of one’s family, for instance) could not possibly have been considered, along with any of the other “pros” and “cons”.

10.2. Other Situations

1. Suppose you rob a store. You take the cash and use it to buy liquor.

2. Suppose you rob a store. You are being forced to do it by a gang member who is using you to avoid risk for himself. Someone is in a car right outside the store, holding a handgun pointed at you. If the robbery doesn’t take place, or you try to flee, he gets shot.

3. Suppose you are homeless. You steal some food.

4. Suppose you are a retired property-owner. You do a street robbery to get more money for drugs.

5. Suppose you are choosing between two job offers. The jobs are very different; perks of one job are disadvantages of the other, and vice versa. However, all together, they seem almost exactly equally as good. Objectively, it does not seem like one job is any better than the other. Finally, you decide on job #2 for gut-level reasons.

6. Suppose you read a math proof. The proof is as airtight as 2+2=4. It is definitive, and you possess the level of mathematics knowledge necessary to fully understand it. You then accept the accuracy of the proof.

7. Suppose you glance at a sign and automatically read the text on it. You read it, but not “intentionally” because you understood it the moment you looked at it.

8. Suppose you pass a beggar. You feel a bit awkward, because you feel bad for not giving a dollar. You are trying to be generous. Finally, you turn around and hand the man a $5.

9. In a drunken rant, you say something you didn’t intend to.

10. You touch a hot stove and move your hand away immediately.

11. You are a severely mentally disabled person. You hit someone and bruise them.

12. You are a dog. You decide to help a boy.

In which of these situations do you have free will, and which do you not? I will tell you my answer. I think you have free will in situations # 1, 4, 5, 8, and to some extent 3, 9, and 12.

In scenario # 6, you had some choice in whether to read the math proof, but once read, your agency is reduced. You have gained knowledge essentially forcing you to accept the proof.

10.3. Opaqueness of Decisions

Consider situation #6, the case of the math proof. In that case, the decider knows exactly the reasons for their decision to accept the math proof. At the same time, they did not have a choice in the matter. Knowing exactly why you made (you had to make) a decision goes hand in hand with not having a choice. There are some people who fight me on this. They say that you can know something logically and yet believe the opposite. I disagree with them on that, but I don’t think it’s that important. If you do have free will in this situation, all the better for my argument. This is a hotly debated topic in philosophy; I just want to acknowledge free will might be lacking here. 

Consider situation #5, the case of choosing between job offers. In that case, you begin by making a list of “pros” and “cons”. But no matter how much you stare at the list, they find that one of the offers simply resonates more for some unknown reason. The pros-and-cons list encapsulates the analytic reasons, such as the number and so on, but mere analytics does not persuade you by itself. When you finally decide to accept job #2, it is a “gut-level” decision. 

This does not mean that your decision is random or not based on evidence. Rather, it means that the weighing, processing, and extrapolation from disparate data points is so complicated that it is not simply summarizable. Your motivations are opaque even to you. It is these intuitive decisions which bubble up from the subconscious which I would characterize as most “free-will-like.” The mostly-but-not-total opaqueness to your conscious mind is a quality.

This can create a sort of paradox: the cases in which we have the most free will are also the cases in which we are least conscious for the reasons for our own decisions. The decisions arise mysteriously, seemingly out of the ether. Oddly enough, some incompatibilists use this point to build their case, pointing out the mystery as evidence for lack of control. I retort that this is what is meant by the phenomenon of free will – this is exactly what it is. When that is not true, when your reasons (and reasons for caring about those reasons, continued recursively) are fully known to you – in that opposite case you would most evidently lack free will.

With that said, that is a necessary but not sufficient condition. Scenario 9 and 10 portray instances of “instinct”, but in a different sense than the kind of instinct that intersects with free will. The drunkenness of scenario 9 represents reduced agency, and scenario 10 is emblematic of unconscious decision making. By contrast, high-agency decisions employ the thoughtfulness of higher-order brain functions.

An alien species 1,000 times smarter than us might theoretically see our brains as simple, straightforward, and predictable as we see ants. But if we ever encountered them, I would simply concede that free will must come in degrees, and they have more expansive free will than us. We may need a different word for what they have, but in any case, it wouldn’t faze me if free will is operative as a series of sliding scales.

11. Criminal Justice: Applications of Compatibilist Free Will

11.1. Before the Law

So far, I have been treating this topic as merely a semantic contention, and so far, that has been germane. However, I would like to speak to the utility of the compatibilist free will distinction. Assuming that compatibilist free will is a legitimate concept, then whether or not compatibilist free will exists in specific scenarios can have normative implications. If you do not believe compatibilist free will is something possible to distinguish, then you will not be able to make certain morally relevant distinctions in scenarios where it is possible to do so.

Let’s suppose you are a prosecuting attorney speaking before a judge. The man you are prosecuting, Mr. James is a fully sane individual and otherwise upstanding college student who was recently caught robbing a liquor store. The judge asks you, “just to be clear, your suspect Mr. James, he was acting of his own free will, correct?”

How do you respond? Do you simply say “yes”, if that answer applies? Or do you launch into a lengthy diatribe, “Well your honor, technically you see, we live in a deterministic universe. Nobody actually has any free will, because everything in the universe is part of a causal chain of atomic motions that inevitably determine exactly the thoughts and actions …. etc etc.” No! It would be crazy to respond that way. But by answering “yes”, you are conceding that the question is not meaningless. There is something that the judge is asking, there is some information he would like to know. When the judge says “free will”, he is referring to some real identifiable property.

11.2: Deterrence

When I say “deterrence”, a lot of left-leaning people may instinctively respond, “deterrence doesn’t work!” That, in my opinion a knee-jerk response, misses the larger picture of deterrence. To say that deterrence is ineffective functions as a narrow political point. People who claim that typically mean something akin to, “increasing the prison sentence for murder by 30% has no statistical impact on the murder rate.” I can even concede that; most would-be murderers are probably not thinking about the exact prison-length punishment. That fact is also not pertinent to what I will argue. When I appeal to deterrence, I mean something more akin to, “if we had no laws against shoplifting, then shoplifting would be much more common.” Or, in actuality, “if we had no laws against shoplifting, then privately-funded security would pick up the slack, accomplishing the same deterrent effect.” In a cleaner example, what I mean is, “if there were no fines, or other repercussions, for dumping toxic sludge into the lake, then more toxic sludge would accumulate in the lake.” Few people, outside of hardcore anarchists would disagree with that: deterrence is effective in that instance.

However, I will make quite a stronger claim: deterrence is the basis for the entire justice system. Every other modus operandi for the justice system is merely a non-core externality.

11.3. Models of the Justice system

I think most people are implicitly operating with an upside-down model of justice. The model is the following.

Criminal justice system, model 1:

1. Someone does something wrong

2. We must hold that person accountable

3. So we put them in jail

And then people have arguments about whether the purpose of accountability is to punish or rehabilitate or something else. This is not strictly wrong, but I think it is a bit misleading. Taking a step back, I prefer the following model.

Criminal justice system, model 2:

1. As a society, we decide to prevent certain actions

2. The best way to prevent certain actions (crimes) is to make rules (laws) forbidding those actions

3. In order for a rule to operate as a rule, it must be enforced; otherwise it is ineffectual

4. We enforce rules by holding people accountable (putting them in jail) when they break the rules

This model de-centers the individual. The reason we put a person in jail really has little to do with the impact of doing so on that individual criminal. Rather, we must hold each criminal accountable, in order to uphold the integrity of the rule system. If we did not, rules would not be effectual (they would have no deterrent effect). 

Ironically, this “make an example out of the convict” model is (it feels to me) much less vengeance-based than alternatives. Each criminal we sentence is themselves a victim: a sacrificial lamb in a sense. We cannot let them go free because of the effect on potential future criminals if our institutions failed to behave in that way. Questions of emotions and fairness take on a new meaning when sentencing a criminal is really about upholding a law-based order, and less about judging that particular criminal’s “worth” in the visceral sense. 

This is not an endorsement of punishments to “make an example” out of people, nor cruel and unusual punishments. Care should be taken to not frame it like that.. We don’t want to conjure images of public torture displays. If you break the law, we are not going to parade you around, but we are going to catch you, because that is a necessary component of having laws. It is simply: if you can’t enforce a law, why have it? Good laws are enforced consistently.

I maintain my conclusions about model #2: The true unspoken purpose of punishments is prevention. Punishments do this because they convey the right information about our institutions, their activities, and how they respond to specific behaviors. The fact that people must suffer by going to jail is a necessary evil, not the desired end-in-itself. 

If the criminal’s victims take pleasure in the fact of his suffering, that is an externality of the system, but punishments should not be determined based on that. We have moved past that. (On the other hand, this does not contradict what was said earlier about the theoretical causes of said pleasure. The desire-for-vengeance mentally likely evolved for some of the same deterrence mechanics that play into the modern justice system.)

(For many people, it just “feels” like “justice is served” to hold people accountable. That feeling is arguably at odds with viewing convicts as poor sacrificial lambs. It’s just motivated for the wrong reasons. But in the end, it does cash out to wanting the same result: it wants the innocent let free, and the guilty not. So who am I to say it’s a “wrong” feeling?)

11.4. Susceptibility to Deterrence

Consider the case of the severely mentally disabled person who hits someone and bruises them. For a more extreme case, there have been several incidents, in wartime, of disabled people killing beloved soldiers on their own side. For instance, during World War II, a soldier driven to madness from PTSD killed another man in his company, and during the Vietnam war, an extremely low-IQ person (on the level of ~60 IQ) shot and killed a well-liked officer in confusion.

What should happen to these men? If you really did not know what you were doing, then you cannot be blamed for the motive in the crime. Further, if you are so disabled that you could not have taken better precautions, then negligence too is not quite the right charge. If you are a continued danger to those around you, perhaps you should be institutionalized and cared for somewhere away from others, and prevented from leaving. However, you should not be held accountable in the traditional sense. There is no purpose in “justice” in the sense of the word whereby it is a euphemism for punishment.

It is a similar thing in the case of animals. Suppose you are the victim of an elephant attack. Does it make sense to put the elephant on trial, and punish it? They actually used to do that. But that is clearly useless. No, there is no use putting the elephant on trial, because it does not understand the concept of accountability, let alone was the elephant acting out of malice. If there is a takeaway, it is to be more careful around elephants, or to keep them out of urban areas, etc. But that is the type of emotional reaction you have to bad weather.

Accountability, in the sense I mean it, requires the capacity of the persons we hold accountable to comprehend and respond to the concept itself. In theory, we hold people to the extent that they “could have known better”. Meaning, they should have been susceptible to deterrence, but calculated (wrongly, and crudely) a higher expected utility from lawbreaking given the possible punishments and enforcement apparatus existing. In other words, they wrongly calculated that in their case they could “get away with it”.

Here we find the link between free will and deterrence. Within the context of criminal justice, the relevance of free will is as an indicator that someone should theoretically have had the mental wherewithal to respond to deterrence schemes.

There are many things that society can do to harshen the deterrence for an action, namely pass laws and increase punishments. In doing that, an update will propagate to the brains of however many million people live in your country: they will, in theory, be less likely to do that action. If an action was done with free will, then that process is relevant; the enforcement of the punishment will help the aforementioned update to process.

If an action was not done with free will, then that process is not relevant. For that person, for that situation, a consideration of the socially-enforced consequences could not possibly have come into play. Recall section 10.2., scenarios #2 and #11. Justice as a form of deterrence is not relevant there. We can allow for those exceptions, and not enforce punishments in those cases, because nobody will fail to be deterred upon learning about those exceptions. No one will think, “I will become mentally disabled, so that I can get a free pass on robbing stores”. Etc.

Would-be criminals need to believe, in the moment they have the free will to make the decision, that the category of people/scenarios susceptible to punishment could potentially include them if they got caught. I believe my terming of free will is coherent in the context, having spent a good amount of time establishing the model in which it operates.

So why does any of this matter? The relevance of this section is simply to point out that 1) A compatibilist terming of free will is important insofar as it allows us to communicate about important things,  2) a compatibilist terming of free will describes a real phenomenon, albeit emergent from existing psycho-social mechanisms. Regardless of whether you like the use of the term, there is real meaning it denotes. 

Anecdotally, I have noticed, among many incompatibilists, a discounting of and failure to grasp deterrence as an important variable generally, even in otherwise unrelated philosophical discussions, and I don’t think that’s a coincidence. Deterrence is a large reason why we care about free will as a culture, why it be so unsurprising that a prominent term pertinent to it exists.

11.5. The False Project of No-Blame Justice Systems.

I have heard the following from free will objectors:

When you realize that there is no free will, it has implications on how you view morality. You realize that the temptation for retaliation is not just ethically wrong, but actually irrational. It is so because the bad actor did not even truly have a choice. They were simply behaving like a machine, the most sophisticated machine known to man, but a machine nonetheless. There is no use, therefore, to the concept of “moral culpability”. No one is morally culpable because they are doing what they had no choice to do.

The way that you feel, in clear cases where a criminal obviously did not act out of free will, I feel like that for all criminals and wrongdoers, because it is the same situation deep-down. If we could create a pill, the “does not want to murder” pill, then we could just force everyone to take it, and eliminate murder. In such a world, what would be the point of hating murderers? But such a pill could hypothetically exist, so we might as well view murderers like that already. This philosophical realization has the potential to transform one’s view of the justice system, and indeed the entire concept of blame.

This view sounds appealing for several reasons. Hatred, retaliation, and blame: it renders these things not just bad, but irrational superstitions, which sounds intuitively to be the right ethical approach. But I don’t agree with it; in our current world, projects for a zero-blame justice framework strike me as provably misguided.

To critique a traditional mode of thinking, without considering the potential benefits of it, always strikes me as myopic pseudo-clever bloviation. Someone with intellectual curiosity should be able to mount a steelman defense. This is the only time in this paper when I’ll invoke evolutionary psychology. Surely, a psychological trait so absolutely prominent as the desire for retaliation must have had some utility in our ancestry. Otherwise, we would not have it, due to the costs it asks of us.

There is a reputational effect of retaliation. If a shop owner allows theft from their shop, the shop is more likely to be robbed in the future. The same logic holds true on different scales. If a town has the reputation for driving thieves out of town, then thievery will dwindle within the borders of the town. This is the utility of retaliation.

One possible response, that I have heard from some but not all free will objectors:

Even if we concede that perhaps retaliation had some passing utility in ancient history, we surely have no need for that now as a species. Armed with our new philosophical knowledge, we can build liberal institutions which obfuscate the need for primitive instincts. Institutions should not feed base-level instincts.

Where I can be sympathetic to parts of this, I still disagree. Although we as a society can build institutions to defeat the requirement to be fueled by base instinct of individual members, it is much harder to defeat the underlying mechanics that rewarded those base instincts evolutionarily in the first place. The same logic by which the desire for retaliation was theoretically rewarded in an evolutionary sense applies also to institutional criminal justice. Accordingly, a justice system that behaves on the mechanism of accountability affords benefits. We adopted it over time because we discovered that it worked. As for why it worked, and works, for which I provided my theory in section 11.3, we needed not know on a logical, mechanistic level, in the process of adoption; we merely needed know that it did work.

As liberal institutions take over, retaliation becomes accountability. There is a meaningful distinction between the concepts. Retaliation can beget more retaliation; accountability enforced by a third party is a definite improvement. But they share the initial deterrence factor. 

I have no complaint against rehabilitative justice, which aims to reduce recidivism and foster reconciliation. I have no complaint against rhetoric preaching the virtues of forgiveness, which aims to reduce hatred and resentment between people. And likewise for any similar projects. My only complaint is with the added premise that we, therefore, can carelessly discard what we have learnt of accountability.

Let me be clear about something I am not saying. I am not saying: “free will does not exist, but it’s a useful fiction of society, so we’ll pretend that it does.” No. Free will, as I define it, literally exists. And we can observe its existence in the fact that it is not incoherent to ask if a criminal committed an act out of free will. Moreover, one  reason (among others) it is worth talking about, and defining as such, is due to anthropocentric facts about its relevance to social functions and institutions.

12. Reducibility and Cognitive Science

12.1 Reducibility and Criticality

In this section, I am using somewhat proprietary definitions of “criticality” and “reducibility”. To be “critical” is the opposite of “reducible”. If a node is critical, you cannot be removed without causing a meaningfully different result. If a node is reducible, then it can be replaced by a much simpler one without meaningfully changing the behavior of the system across a range of states. 

Criticality is a quality that comes in degrees, where a system with higher criticality has a larger number of critical nodes. The equivalent is true with reducibility as to reducible nodes.

The word “meaningful” is another term I am using in a somewhat proprietary way. I will discuss the distinction later. For now, that distinction is doing some legwork, in order to disambiguate the use of the word as a free-will-relevant concept from other, more pure-math concepts.

The term “reducible” is borrowed from computer science. Algorithm A is said to be reducible to algorithm B if A can use B to accomplish A’s task, with only the added complexity below a certain threshold (relative to the input) needed to tailor B’s results to A’s task. This computer science conception of complexity has inspired my general understanding of complexity, although I do not pretend that the term “reducible” as applied in this section is necessarily the same as the computer science definition I have just given.

12.2. Degrees of Freedom

Many compatibilists have advocated for a view of free will based on “degrees of freedom”. I remember Daniel Dennett once giving the example of playing a game of chess with some chess-playing agent. After the agent makes a move, you say, “oh, it could have castled.” What does it mean “it could have castled”? Perhaps you are playing against a deterministic chess-playing AI, to which the deterministic character is transparent to us. But even in that case, it seems to make sense to say “it could have castled”. In other words, its programming realistically allowed for that alternative option if the RNG had gone a different way, as opposed to very bad moves which that incarnation of the chess bot would never have allowed.

I don’t actually believe that chess bots have free will, because I would deny RNG rises to that level. Even so, the example is illustrative. What does degrees-of-freedom mean, according to this view? I take it to mean that the action was highly dependent. If even one of many internal actions had changed, it would have delivered a different but still-valid result. In chess there is also such a thing as a forced move, which is the opposite case. For forced moves, there is nothing in your decision-making apparatus you can vary to deliver a different yet legal move.

12.3. Natural Examples

Consider the following situation. Inclement weather causes a lightning strike which hits a power cable. This cuts power to the neighborhood house, which, without a backup source, shuts off the router. Due to no internet, and bad cellular coverage, a businessman currently at his home misses an important email. Because of that, he runs into trouble at work.

Weather, and lightning striking the power cable, was an extremely chaotic system. Additionally, Internet communication is an extremely complicated system. Finally, the social dynamics which caused the man to lose his job were very complex. 

What was not complex in the scenario? Although the router is typically a very complex piece of electrical engineering, none of that complexity was able to be expressed in the above situation. The functioning of the router was reduced. Looking at the chain of cause-and-effect as a whole, the actions involving the router, although perhaps deserving a mention, were entirely reducible to the power loss. It simply was not possible for the router to do anything in the absence of power other than default to the “disabled” position.

In this first example, I say that the router is “reducible”. With that said, there are many other variables in the system that did not even deserve a mention because they had even more of a reduced impact on the series of events (despite being involved in some way). Those unmentioned variables are the truly reducible components.

Now consider an acorn floating in the water of a stream. As the acorn gets to a waterfall, the acorn falls down, hitting rocks 5 times, before continuing in the stream. Now imagine that the cliff of the waterfall was steeper, and the same acorn only hits 1 rock before continuing at the bottom. This may not be the perfect analogy, because fluid dynamics are complex, and it is possible to find many interesting differences, however, for the purpose of this example, we will restrict our analysis to “does the acorn make it through to the bottom or not?” With respect to that question, the rocks that the first acorn hits can be safely ignored as having no impact. Those 5 rocks are, for all intents and purposes, “reducible” to 1 rock. 

12.4. Criticality of Brains

Consider a situation in which you find yourself cornered. You are a sailor on a ship which is on fire, and you have nowhere to run. Of course you know how to swim. In response to your predicament, you jump overboard, aiming to swim to the nearby ship. What was the cause of your jumping overboard? It was “reducible” to the fire, and your location with respect to the fire. Whatever other thoughts you might have had in the moment, an outside observer could ignore almost everything except for those causes. Because of that, the level of free will you had in that situation was reduced. 

Another way of putting it is: there wasn’t much to think about, you just had to act. This is a “reduction” in the respect that the outside observer is simplifying their model of the cause-and-effect network, (in this case by stripping out any stray thoughts more complicated than “fire, must jump”), without losing much fidelity in their grip of the situation.

With that said, the decision to jump was not entirely without mental consideration. So when I say the jump decision was reducible to the fire, I mean within the limits of some threshold. In fact, it would be more accurate to say its degree of criticality was very low.

If we restrict our analysis of a mind to a single situation, then we can run into a problem. Suppose a mind is faced with a complex decision, like the decision of where to go for college. It is choosing between two options. However complex its inner workings, in the end the choice will resolve to a binary decision. It is tempting to say, “Aha, the mind can be reduced to a decision process that chooses option 1 instead of option 2”. That is fallacious. If you reduced the mind to said decision process, then it would not work when faced with other difficult choices. The criticality of a mind, therefore, is not evaluated with respect to a single decision. Rather, it is evaluated with respect to a large range of decisions. There are hypothetically near-infinite varieties of decisions that a given mind could be faced with, and the mind should be considered in totality.

12.5. Brains are Motivated: Why “Random” Systems Don’t Count

I promised that I would return to what the word “meaningful” means as in “to create a meaningfully different result.” Without that distinction, you will end up labeling systems as having free will that clearly do not. 

Earlier, I gave the example of the weather generating a storm as an extraordinarily complex system that cannot be straightforwardly reduced. But obviously, I do not believe that a storm has free will. 

I called the storm “complex”, in the sense that follows from chaos theory: small differences in starting inputs can cause large differences in the result. Indeed, this is where the famous “butterfly effect” comes from: that a butterfly flapping its wings can cause a storm (almost certainly not true, but a good illustration of the concept.)

However, complexity is not the same as reducibility in the full sense, although they are related. An illustration of why is in the example of a highly “random” function, like a die throw. A die throw is a highly “chaotic” action in the respect that it’s highly sensitive to starting inputs. However, those sensitivities are not meaningful. 

A die throw can be replaced by any appropriately-random but much-more-simple procedure (such as a modulo arithmetic calculation). That much-more-simple procedure would technically generate different results in any given situation, but those differences are not meaningful when you only care about “is this generally random?”

Suppose we grade a system merely by the ability to generate unique or near-unique outputs over a wide range of inputs, then hashing function would fare well. However, a hashing function actually has low meaningful criticality. Why? Because a key component is the internal complexity, not just the variance in external results. A hashing function, and pseudorandom functions for that matter, operate on relatively simple modular arithmetic. The workings of a hashing/pseudorandom function can be reduced to a short series of equations. They may require insight to derive, but they are not necessarily complex in themselves.

Just to reiterate. If we mathematically analyzed dice rolls, the results that dice produce would not be any different in kind to what you can do with pseudorandom number generators arranged to produce similar-seeming results. Again, technically the results would be different in any given throw/run, but those differences are not meaningful, so we treat it as the same. Hence, a die is reducible to a pseudorandom number generator, and a pseudorandom number generator is reducible to a few arithmetic calculations (computationally speaking), and is therefore low in criticality.

Now we can come back to the storm. I contend that, when it comes down to the unpredictable and extremely-chaotic confluence of factors that determine whether or not a storm will be formed, it is for all intents and purposes akin to a dice roll, in the following sense. Although an indeterminate number of influences could have contributed (billions of butterflies flapping their wings!), none of that is “meaningful”. Contrast this against a choice that a brain takes: the decision is clearly connected to the inputs that a brain receives. The atmosphere just scrambles those inputs into noise, to the point where the granular-because information is effectively lost as noise, and it’s a fool’s errand to even try to model/reconstruct what they were.

Brains are different from storms, as they do not “scramble” the information, but rather utilize it in a purposeful, directed way. They are not so easily reducible to chaos or noise. 

12.6. Brain Complexity

Some components may exhibit doubled criticality when they come to be important once, and then come to be important later within the same process. This is simple enough; some elements may make decisions multiple times. But in the case of brain functions, this is kicked up an order of magnitude.

A single neuron firing can potentially kick off a chain reaction which leads to one decision over another. If that last sentence may turn out to not be technically scientifically accurate, then for the rest of this section, just swap out “neuron” for “the smallest structure in the brain that is potentially responsible, by action or inaction, for a change in final decision.” The argument is essentially the same. The point is, in the relevant respect, the brain is not stochastic, meaning that lower-order actions do not cancel each other out to become irrelevant to the final decision. Rather, even very small components are there because they can potentially have an impact on one’s decisions.

But single neurons increase the criticality of the brain by even more than that. This is the absolutely most important thing for understanding what I consider special about the brain. Within the process of making a decision, single neurons can be repeatedly used. Let us model thoughts as waves that propagate throughout the brain. In the process of making a decision or determination, these “thought-waves” can double-back, reusing the same neural pathways multiple times. In this way, individual neurons can be critical multiple times in the process of making a decision. If we deconstruct the behavior of the brain that lead to any given determination, we can find numerous “key points” in which a neural pathway was operative.

There is another dynamic: the more neurons are present in the brain, the more “sockets” that a given synapse from a different neuron can theoretically connect to. In this way, the introduction of a given neuron makes the brain more complicated not just in itself, but as a factor of the existing mental infrastructure. For each neuron you add, a benefit is multiplied across every existing neuron which now has a new place to which it can connect.

This is reminiscent of “Metcalfe’s Law”. Metcalfe’s Law is about social networks, and so may not hold in the same way for neural networks, but at the very least, the laws that govern the benefits to networks differ from “ordinary” systems. I came across a fascinating neuroscience paper titled: “Brain Computation Is Organized via Power-of-Two-Based Permutation Logic”. Fundamental to how brains work is an exponential growth relationship.

This leads to a discussion of “asymptotic growth”: the increase in complexity as a function of the number of added units. Meaning, the increase in complexity can be linear, logarithmic, exponential, polynomial, and so on. A high-order asymptotic growth results in extremely high criticality as the numbers get large. Then take into account that the brain is composed of ~86 billion neurons. 

12.7. Isn’t it Arbitrary What’s “Meaningful”?

Take two differently-constructed but otherwise-identical number generators. Say, on one hand a computer program that randomly outputs 1 or 0, and on the other hand a coin flip. Do these systems produce meaningfully different results? I don’t think so. But even in saying that, isn’t it arbitrary? 

I think it’s not arbitrary. What I mean by “meaningful” is actually the same as what I mean by “motivated”, defined in section 9.6. The difference between two random number generators is incidental.

Technically, my definition of free will contains a duplication. The definition specifies “motivated yet non-coerced [actions]”, but if you unpack what I mean by “coerced”, the same concept of motivation crops back up again in the form of “meaningful.” Though redundant, I have written it that way because I thought it was more clear. 

I think that brains have motivation, but storms don’t. I think there is a meaningful difference. I don’t think that’s unreasonable, and if you feel the same way, we are in agreement on this topic.

But isn’t “meaning” contingent upon what kinds of questions one is asking? I can only say that almost all concepts are constructed this way: with implicit dependence on relevance to our knowledge-objectives (what kind of information you’re trying to get, depending on your local purposes). My discussion in 12.8 picks up on this thread.

12.8. Isn’t it Arbitrary What can be “Reduced”?

This is a critique that I take very seriously. In the earlier example of the storm causing a power outage shutting off the Internet which prevents the man from sending an email on time to keep his job – I claim that the loss of Internet can be reduced, whereas the weather dynamics cannot be immediately reduced, and the social dynamics (of the job) can’t be reduced at all. Isn’t that arbitrary?

Intuitively, the power loss causing an Internet outage is a more direct cause-and-effect relationship, whereas the weather is a chaotic system, and the social dynamics are high-level and cannot be trivialized. Both are systems with many inputs and outputs that cannot be easily predicted. But isn’t that putting undue importance on my own human subjective ability to “predict” something, when I should be able to point to some hard physical difference? 

There is a criticism to be made here, but it is not unique to this issue. Because reduction involves reframing multiple objects as one, it connects to a deeper epistemological question in philosophy about decomposition. How does one categorize subcomponents into objects? However you go about it will be arbitrary. Whenever you say that some information about one system is more relevant than information about another system, you are smuggling a debatable schema for determining relevance. A wide variety of philosophical issues will run into this at some point in their articulation. 

Another possible critique of my views concerns the earlier thought experiment of the man who was forced to jump out of the ship, instead of being consumed by the fire. I claim that he had no choice, because regardless of any minor varying in the starting inputs, his jumping was inevitable. However, one could vary the starting conditions so that the man was nearly suicidal, and therefore on the fence about whether to walk into the flame. So perhaps there’s some bias in my choice of construction of counterfactuals? 

I think that critique is limited. There are edge cases in any situation, and that construction was useful for the example. We merely need to specify that he is unlikely to be suicidal as one of the conditions of the thought experiment, and I think almost anyone would accept that specification. Moreover, if an edge case is very unlikely, then the knowledge of that fact can be factored in. 

In the situation with the electricity and the router, we know based on our experiences that routers don’t work without electricity. But we can imagine an alien world where there is a much more complicated interaction between routers and electricity, according to which the relationship cannot be reduced. So, making a claim “X can be reduced” is implicitly a claim about what situations are likely in our world.

13. Neuroscience and Consciousness

13.1. The Chinese Room

One possible objection to section 12 is the following.

Suppose you replaced the brain with a simple set of enumerations listing how the brain would behave in an enormous corpus of possible situations. If this list is detailed and good enough to do a decent job of approximating the brain, as far as predicting what it would do in most situations, you are forced to concede the brain is reducible to that. However, such an entity would intuitively lack free will. How do you square this position? 

If you do not have an objection similar to the above, then you can skip this section.

Now to address the objection. The description reminds me of the Chinese room thought experiment. For reference, this Chinese Room thought experiment goes like this.

Searle is a man who only speaks english. He is locked in a closed room with papers, pencils, erasers, filing cabinets, and has a book of instructions. Searle receives Chinese characters through a slot in the door, then processes them according to the book of instructions. Via this process, he produces Chinese characters as output, without himself understanding the content of the Chinese writing. By these actions, he can pass the Turing test, as judged by the Chinese-speaking examiner outside the room.

The Chinese Room thought experiment is usually used to discuss consciousness. However, I want to make it clear that, although consciousness and free will may be related, I do not at all consider them the same thing. I merely think that some arguments that can be made about consciousness are relevant to, or carry over to, free will when the framing is changed.

A “philosophical zombie” is a being that behaves in such a way that it appears to an outside observer to be a conscious being, but in fact lacks any conscious experience. 

Ok. If a consciousness-philosophical-zombie is a being like that described, then I can make an alteration. A free-will-philosophical-zombie is a being which behaves sufficiently like humans (which I think to have free will), while consisting in fact of a simple set of enumerations (a list of every possible situation and how to answer each of them).

What do philosophers think about consciousness-philosophical-zombies? I will reference the same PhilPapers poll of philosophers from section 1.2. Here are the results.

Are zombies…

Conceivable but not metaphysically possible   331 / 931 (35.6%)

Other    234 / 931 (25.1%)

Metaphysically possible    217 / 931 (23.3%)

Inconceivable    149 / 931 (16.0%)

A plurality say that philosophical zombies are “conceivable but not metaphysically possible”. I agree with that judgment. Consciousness researcher Anil Seth gives the example of a jet aircraft flying backwards. It is possible to imagine a jet aircraft flying backwards in the sky. You can probably picture that image. However, if one understands the machinery of a jet aircraft, they know that’s not possible.

The Chinese Room story references the Turing Test. If the Turing Test lacks any back-and-forth, passing it is trivial. For the Turing Test to work, it relies on there being a lot of back-and-forth. The longer the time span of the back-and-forth, the harder it is to pass.

In a long back-and-forth, the machine has to remember what you said a long time ago. If you tell the machine some piece of information on hour 1, then it should respond differently to you in hour 3. Of course it can forget information, but within reasonable bounds of remembering as much as a human would. 

In order to accomplish that, the prompts would have to be extremely long. Any enumeration of possible circumstances would have to be truly impossibly long. At a certain point, it would be simpler to store the Chinese-room-instructions in a compressed state, such as a tree structure so as to model branching conversations. 

But once we allow the instructions to store information in a compressed state, then the ideal representation of the instructions is not anything resembling an enumeration, or even a branching tree structure for that matter. This improved model would take on the form of a deeply integrated reasoning model. 

And closer it gets to convincingly-human, the more the model starts to near the complexities of a brain. At that point, it can be said to exhibit, without irony, free will/consciousness/whatever we’re reasoning about.

So the choices are:

1- Allow the super-inefficient explicit numeration that is not compressed in any way. Those instructions would be impossibly long. I don’t lose much sleep over the notion, “This list of instructions larger than the solar system has free will.” We’ll worry about that when we build it. I don’t know if such an entity could even physically exist. If it did, we’re talking about something that could respond to social pressures in the same way that humans can, except that that contingency is technically pre-programmed. 

2- Allow compressed representations, which are then effectively AI models, and which soon begin to resemble brains in ways relevant to this discussion.

I started to write this paper before the popularity of Chat-GPT Chat-GPT challenges notions of philosophical zombies. Is it conscious? Does it have free will? Is it a zombie? Those questions are too big to address within the scope of this paper. 

13.2. Graph Theory

I believe the field of graph theory has great insights which are relevant to understanding which conditions give rise to free will as I understand it. A graph (read: network) that is too centralized and interconnected may fail to give rise to adequate parallelization (many distinct mental tasks happening simultaneously) which is necessary for some complex forms of thought. Some theories of consciousness imply delineation/parallelization of thoughts, whereby thousands of semi-independently-working yet simultaneous idea-strains gradually accumulate/aggregate into a single decision. On the other hand, a graph (read: network) that is too decentralized may lack the necessary interconnectedness to adequately integrate seemingly-disconnected concepts.

There is a lot that I am desperate to say about consciousness as it relates to free will. They are not the same thing, but they tend to co-exist. I do not take a position on whether or not they are related. A way they could be related is if the conditions which give rise to free will (at least to my level of satisfaction) necessarily or as a requirement also give rise to consciousness, or if one is required for the other, and so on. To make a claim about that would require either advanced neuroscience knowledge or immense arrogance. I have the latter, and as such there are many things I want to say on this topic. (In particular, researcher Anil Seth has strongly informed my views on consciousness.) but I will suppress the urge to talk much more about it, as it’s not directly related to the argument of this paper, and because this paper is already too long.

Once we define free will as some feature of the brain, as I believe it is, then it opens up a huge world of research and discovery, giving way to philosophical discussions of how they relate to many prescient aspects of the human experience. And one way I am motivated to rebuff the incompatibilists: I view their role in this discussion to walk in and shut all of that down; to say, “nope, let’s not talk about free will, it’s fake!” That would be fine if it was a disagreement about facts, but for many of the philosophers most vocal on this topic, it’s only a semantic dispute. So I look down upon their impact.

13.3. Neuroscience Experiments that Purport to Disprove Free Will

There are several neuroscience experiments which purport: “When participants make decisions in the lab, we find that by the time a person reports being conscious of what decision they will make, our experiment shows that their brain has markers of actually having decided what to do several increments of time previously.” Papers written based on this experiment tend to then claim, “Therefore, there is no free will!”

I can poke some holes in that line of reasoning. For one, the effect operates in a lab setting. There is no chance that someone is going to make a truly important life decision, such as, “Which house should I buy?” in a lab setting. Inevitably, the choices the participants make are going to be toy problems, of the “should I choose the blue back or black ball?” variety. You do not even measure a question like, “Should I rob that store?” which, although is more spur-of-the-moment, is fundamentally more life-altering (and therefore of a different, impossible to study quality) than toy questions.

However, even if I take these experiments on their own merits, there is still a flaw in the logic. “Brains show signs of making a choice a second before the person is conscious of making the choice. This shows, you have no free will.” No it doesn’t. It shows exactly what the experiment tests, no more, no less. It may have deep implications for our understanding of consciousness. But, as I’ve stated previously, free will is not the same as consciousness.

Intuitively, the extreme version of this, a philosophical-consciousness-zombie who has free will, strikes us as impossible. Even I don’t go that far. But I can live with some theoretical disconnect, in the following way. Some operations in the brain give rise to conscious experience, while others are responsible for acts of free will. This is the result of the arrangement of various parts of the brain. When free-will decisions are made, you need not be conscious of that behavior in that instant. Only slightly offset, the operations which give rise to consciousness “come online”. 

Those parts/operations, I contend, must do something important – perhaps reintegrate the information in order to prepare the brain for future decisions, perhaps some sort of interpolating – but something.We are conscious within dreams, and dreams are probably not useless. Whatever gives rise to consciousness almost certainly possesses some function. As it relates to free will, it could be that those functions create the conditions which are necessary to give space for free will. But in any case, it need not be that the two things perfectly coincide. Even if consciousness does not correspond to anything useful, it is a distinct concept.