The Brain Has A Special Faculty of Reason

We Do Not Just Compute

Introduction

A premise taken to be an axiom among many naturalist atheists is that what the brain does is essentially just computation. Many explicitly adopt the computational theory of mind, while others hold different theories, but all in all, they largely seem to agree—the brain isn’t doing anything special, it’s just a complicated machine. We have no special faculty that enables us to grasp truths or apply reason or anything of the sort; we’re just a fancy meat computer following preprogrammed rules, given to us by evolution.

I think that this is clearly false, and if true would put us in a skeptical scenario. Humans have the ability to reason in a way that enables us to be in some way acquainted with the truth in a way that no algorithm designed purely by natural selection could ever be. There are, in fact, virtually no arguments ever given for the conclusion I’m arguing against—it’s generally just assumed, occasionally with brief references to the success of evolutionary psychology. But the nonmechanist has a perfectly good explanation of that—of course evolution explains how we got to have the various capabilities that we have! However, there was survival advantage to having a general capacity for reasoning, and so natural selection selected for this general capacity. Thus, defenders of the view I endorse here, which we might call Nonmechanistic Rationalism (NR), have a perfectly good explanation of the success of evolutionary psychology—we don’t deny any of the things that reductive materialists claim exist, but simply claim that there is something more that they do not admit in their ontology.

The falsity of NR is often assumed but rarely argued for. This is part of a wider trend of people dismissing anything that is not included in our best scientific theories; falsely believing that science is all you need for a complete model of the world, leaving out robust mathematical facts, modal facts, moral facts, logical facts, and sometimes even consciousness (of course, those doing the leaving out tend to claim that they’re not really leaving out consciousness, before describing something they call consciousness when it’s nothing of the sort). When I first realized that some of my views committed me to accepting NR, I was shaken—it felt like flying in the face of science. There is something jarring about accepting it, some unscientific feeling that one gets.

But the more I thought about it, the more I realized that it is almost impossible to deny. It only feels unscientific because lots of scientifically minded people tend to take it for granted, not because there is a powerful argument for it from science. So I encourage readers who find my arguments shocking or are troubled by their unscientific feel to really introspect and see if their rejection is based on just a visceral feeling of unease, or is based on reasons. I suspect they will find it to be the former.

One brief clarification that should be obvious: I believe in evolution by natural selection. The model I put here is the following:

1. There are psychophysical laws that cause particular physical states to give rise to consciousness.

2. The psychophysical laws also enable the consciousness to have a faculty of intuition, by which they can grasp truths by thinking about them. This depends on various features of the physical system and is far from infallible, but it enables us to grasp important truths.

3. Evolution selects for the physical systems that give rise to rationality because they are adaptive. Being able to reason well and grasp significant truths is adaptive, so it’s selected for.

Note that when I talk about reason, I mean this in a broad sense that includes being able to grasp what’s true. Reasoning is not just about deriving accurate conclusions from premises, it’s also about figuring out correct premises in the first place. If you’re uncomfortable with this notion in the first place, just scrap replace all mention of rationality with the more nebulous “good thinking.”

Two reasons to be more open to NR

Lots of people are non-physicalists about consciousness. I’ve argued for non-physicalism elsewhere at some length, Chalmers’ book The Conscious Mind provides quite a solid case for it; if you have not yet read Chalmers, you really ought to. I won’t rehash the arguments here, but they are great in number and very convincing.

But once we’ve admitted that the mind is not just physical, we should be more open to guessing that it can do things that purely physical things can’t do. An algorithm following preprogrammed rules has no special faculty of intuition, but the mind might. Once we reject physicalism, we should be much more open to such explanations.

In addition, many people think that physicalism can’t account for mental states being about other things—a feature misleadingly dubbed intentionality. There are various arguments for why a purely physical system can’t be about another physical system. I’m quite sympathetic to this view; it just seems obviously impossible for a clump of atoms to be about another clump of atoms. Beyond this core intuition that clumps of matter can’t be about other clumps of matter, there are various other arguments.

Suppose we grant one or both of these assumptions. These should make us much more willing to accept that the mind has some non-mechanistic rational faculty. Once we accept that the brain is not just a machine, not just a meat computer, and is in fact doing things that no meat computer ever could do, then why attribute it only the purely computational or mechanistic properties of a meat computer? If we grant that the mind has the ability to think about things in a way that no mechanistic system ever could, because no mechanistic system can enable thinking about anything, then we should be much more sympathetic to NR.

An analogy might be helpful here. Suppose we thought that everything in the universe might be explicable in terms of the higher-level properties described by Newton. We then discovered that there are some things that Newtonian laws can’t adequately describe—some weird things at the quantum level. Well then we should be more willing to accept other weird things at the quantum level. Similarly, once we see that the brain is doing things that no matter could ever do, specifically involved in thinking about morality, modality, and mathematics, we should be much more sympathetic to the notion that it is doing other things that matter could never do; things that are also involved in thinking about morality, modality, and mathematics.

The types of knowledge we want to maintain

Humans have intuitions about a lot of domains. In this section, I’ll argue that we should believe that our intuitions about many domains are veridical. Then, in the next section, I’ll argue that unless NR is true, our intuitions about such domains wouldn’t be reliable.

1. Moral intuitions: moral intuitions are, as the name suggests, intuitions about morality. We have lots of moral intuitions including that if A is better than B and B is better than C, then A is better than C; that pain is bad and worse than pleasure; that it’s wrong to torture babies for fun; that children should be treated well rather than abused; that Hamas’s terror attacks against Israeli civilians are evil; that it’s wrong to stab people when you come across them in the streets; that it’s wrong to cause lots of very bad things for the sake of small benefits; that the holocaust was bad; and we have many more. If NR is the only way to maintain such beliefs then NR is clearly true—any argument against NR will appeal to premises that are much less obvious than that the holocaust was bad.

2. Modal intuitions: these are intuitions about what’s possible and necessary. Paradigm examples include the following: there can’t be true contradictions, married bachelors are impossible, something can’t be neither red nor not red, in all possible worlds the moral and mathematical facts are both the same (e.g. if you teleport to a different world you won’t discover that there are only 52 prime numbers rather than infinity or that torturing babies for fun when it makes no one better off is fine), and all things are themselves. These are similarly obvious—you can’t just see with your mind that some states of affairs can’t obtain. There aren’t very many people who are skeptical about these modal claims—they’re just so obvious. If one is skeptical of them, then it’s hard to take their view seriously, but we can even argue against their view empirically—if they were right that contradictions were possible in principle in an alternative world, it would be very surprising that we haven’t found any (they might dispute that we haven’t found any, pointing to the liars paradox, but it would still be surprising how few we’ve found).

3. Mathematical intuitions: These are what they sound like, intuitions about mathematics. Examples include 1+1=2, each step in the proof that there are infinite prime numbers is true, multiplication is commutative (a x b = b x a), division isn’t commutative (a/b doesn’t equal b/a), and that the greater than relation is transitive. These claims, like the ones before, are extremely obvious and play a role in explaining the world; if they were merely useful fictions, then it would be hard to explain the consistent regularities described by the laws of mathematics and their integral role in physics.

4. Broad metaphysical intuitions: These are intuitions that don’t fit neatly into one of the other categories. Examples include: the prime minister is not a prime number, consciousness can’t be reduced to facts about structure and function, the number 2 is not explainable in terms of cats, something can’t have a color without a shape, if A is inside of B which is inside of C then A is inside of C, if A is only true because of B and B is only true because of C then A is only true because of C, simpler theories are more intrinsically probable, no one is taller than himself, and more. If we deny these, we’re obviously in crazyville, being unable to know, absent empirical investigation, that people aren’t numbers. In fact, this would lead to induction undermining skepticism—if we can’t trust that simpler theories are more virtuous, then we have no reason to suspect that there isn’t an extra law of nature that will cause the universe to self-destruct one second from now.

5. Logical intuitions: these are intuitions about logic. Examples include: if A implies B and A is true then B is true, if P is not not true then P is true, if A is true if and only if B is true and B is true then A is true, and everything else you learn in a standard logic book. These are similarly obvious; any argument against us being able to know them will be less intuitive than the notion that, for instance, modus ponens is valid.

6. Epistemic intuitions: these are intuitions like the following: if proposition A has more evidence than proposition B and A is incompatible with B, you should believe A instead of B; it’s irrational to believe that the earth is flat; if you know something is true, you should believe it; it’s irrational to believe things for which there is no supporting evidence if there is lots of evidence against them; and much more. If one gives up epistemic normativity then their view is self-defeating, because they grant that you have no reason to believe their arguments.

Reflecting on these areas of knowledge should demonstrate that absent the ability to gain knowledge that is not directly about the physical world, we are in essentially a skeptical scenario, unable to trust induction or the reliability of our senses. Thus, if denying NR requires we give up this knowledge, or leaves it unexplained how we come to know these things, then it is a deeply unattractive option. If a theory has to leave dozens of strange coincidences unexplained, then it is a terrible theory, and if it gives one no reason to believe they have hands, bachelors can’t be married, and that torturing babies is wrong, then it is nuts!

Why we can’t maintain knowledge about these domains unless we accept NR

If we deny NR, then we think that the only way we come to know facts about these domains is by observing the physical world. The brain, on such an account, is a physical organ that takes in physical information—perhaps while also producing a few nonphysical qualia and thoughts about things. But if that’s true, then there’s no way that the truth of these facts can explain why we believe them. On such an account, natural selection is solely responsible for imbuing us with our beliefs, wholly unrelated to their truth. Nonphysical facts cannot physically influence the brain, and thus it would be a coincidence that our knowledge of these nonphysical facts is accurate. In contrast, believers in NR have a solid explanation of our knowledge of these facts—we have a basic capacity for reasoning which allows us to figure out that these things are the case. Our nonphysical mind grasps these truths—we believe them because they’re true.

Suppose you had lots of beliefs about things. You weren’t sure exactly why you believed those things. Eventually, you learn that the reason that you believe them is that your brain is reading off the results of an 8 ball that spits out random sentences and then taking them to be true. This would give you a solid reason to abandon your beliefs; if the reason you believe them has nothing to do with their truth, then you should give them up.

If you get a bit of evidence for the magic 8-ball theory you should still be very hesitant to accept it, because if it were true, it puts you in a skeptical scenario and means that you have no reason to trust any of your beliefs. Similarly, deniers of NR hold that natural selection has given us our beliefs in a way that is, in most cases of the nonphysical facts, not at all influenced by the nonphysical facts. If you found out that the reason you see a particular chair is not caused or explained by the existence of the chair, then that would give you reason to give up belief in the chair—similarly, if our beliefs about non-natural facts aren’t explained by the non-natural facts, then we have no reason to believe them. We can make this argument more precise:

1. For A to give you a reason to believe B, then A must be more likely if B is the case than if B is not the case.

2. If NR is false then the non-natural facts being the case does not make it any more likely that it would appear that there are non-natural facts than if they weren’t the case.

3. Therefore, if NR is false then the fact that there appear to be non-natural facts gives us no reason to believe that there are non-natural facts.

4. If the fact that there appear to be non-natural facts gives us no reason to believe that there are non-natural facts, then we have no reason to believe that there are non-natural facts.

5. But we do have reason to believe that there are non-natural facts.

6. So NR is true.

Here I use non-natural facts broadly to include the list of facts described before. One might doubt that each of them are non-natural—though I think they all are non-natural—but if so, feel free to substitute in your own preferred term.

1 is a trivial implication of bayes theorem. 4 is hard to deny—what else could give us a reason to believe in the non-natural facts other than the fact that they seem to exist. 2 is similarly plausible—if the non-natural facts don’t explain our beliefs about the non-natural facts then they don’t make them more likely.

Even if one doubts the specific premises, the core intuition is clear; if our beliefs in the non-natural facts are not explained by the non-natural facts, then we are much like the person who believes in a chair based on seeing it, despite knowing that the reason they see the chair has nothing to do with its existence, or who believes in pink elephants because an 8-ball told them that they existed, even though the 8-ball is just a cheap toy that doesn’t track the truth.

Tomas Bogardus has a nice way of pressing the point in his precisely titled paper Only All Naturalists Should Worry About Only One Evolutionary Debunking Argument. Bogardus’s argument is aimed at providing a debunking of morality for those who deny NR (which he refers to only as Rationalism), but the basic point can be made for any of the 6 categories of fact that I describe.

Bogardus notes that deniers of NR think that there is a purely mechanical explanation for our moral beliefs. For example, maybe the reason we believe that pleasure is good and pain is bad is that there was some selection pressure for that belief—maybe in the ancestral environment, lots of people found talking about the badness of pain to be an aphrodisiac, and so sexual selection favored talking about the badness of pain. But there’s a nearby possible world where people found talking about the badness of scorpions to be similarly arousing, and thus the only moral belief that people affirmed was that scorpions are bad. But if this is true, why should we trust our belief that pain is bad over the belief that scorpions are the sole bad? We seem to be on an epistemic par—both believing that morality is as it is because natural selection favored it. If these guys are our epistemic peers, as are the infinite other beings in possibility space with different moral views, we should suspend judgment in our moral beliefs. But then we should just be error theorists on parsimony grounds.

The same point can be made with reference to any of our other beliefs. For instance, take our belief that the prime minister is not a prime number. You could imagine weird selection pressures that favor the belief that the prime minister is a prime number. And yet surely we shouldn’t defer to those who believe that. NR has a good explanation of that fact—through our ability to reason, we can simply see that the prime minister isn’t a prime number. But other views don’t.

The common response here from deniers of NR is to claim that we have something like direct access to the moral, modal, and mathematical facts. Thus, we don’t need to justify them through inference to the best explanation of our intuitions, and as a result, explanatory irrelevance is irrelevant. This broad line of response is defended by Richard Chappell and described well by Korman and Locke. The basic idea here is that a response to someone who tries to provide an evolutionary debunking of, for instance, the goodness of friendship would be the following:

1. It’s no accident that we think that friendship is good. There’s an obvious evolutionary explanation of this fact.

2. It’s no accident that friendship is good. That’s a necessary moral fact.

3. So there’s an easy explanation of how we come to know the moral fact that friendship is good.

I notice this line of response is especially popular among epiphenomenalists. Richard Chappell defends it, and one of my best friends who is also an epiphenomenalist also defends this line of reasoning. This is no coincidence; it’s basically epiphenomenalism but for the moral, mathematical, modal, metaphysical, and more facts. It says that those are things that we have justified beliefs about even though the facts neither cause nor explain our beliefs. But I think that it has similar problems to epiphenomenalism, in that it has trouble explaining how we come to know about consciousness. Epiphenomenalists can say that our direct acquaintance with consciousness explains how we know we’re conscious, but no similar strategy is available here—that we’re conscious makes a difference to our direct acquaintance, but the non-physical facts I describe don’t make a difference to our brain states. Specific problems for this account:

1. It requires rejecting a premise of the argument I gave above and also of Bogardus’s argument. But I think all the premises are hard to argue against.

2. I’m sympathetic to the idea that we can be non-inferentially justified in believing certain claims about morality, modality, and mathematics. But if you think the brain is just following some algorithm, then it’s hard to see how that is the case. It would be like saying “the only reason I believe that God exists is because the non-truth-tracking 8-ball tells me, but this confers non-inferential justification.” No! For a process to infer non-inferential justification it has to be reliable! You do not get to stipulate that the process that leads to our beliefs is wildly unreliable, but then also that you get non-inferential justification! Once you learn that the reason you believe X has nothing to do with the truth of X, then you should cease to believe you have good justification for believing X. To have non-inferential justification to believe something it seems like it has to make a difference to your belief, but on this account, non-natural facts don’t make any difference.

3. This is a broader concern that I haven’t thought about much, but it’s hard to see how a mere algorithm could confer non-inferential justification. A computer can’t have direct acquaintance with any facts! If we are just meat computers then we couldn’t have direct acquaintance with any facts either.

4. Even if we grant that we have justification for believing that we know the non-natural facts, this seems to still call out for explanation. Suppose you had an 8-ball that guessed whether people were going to date each other in the next year. You use it to guess whether two people will date and it’s right over and over again. This would seem to call out for explanation. Why is it that it’s so right? Positing brute unexplained coincidences over and over again is a terrible explanation. But that seems to be the denier of NR’s explanation—that the non-natural facts line up with our beliefs is just a happy accident. Additionally, Huemer argues that once we grant a basic capacity for rationality, we get a much better explanation of our specific moral beliefs—it’s easier to explain why we believe in transitivity if you think we believe it because it’s true.

Conclusion

Those who reject Nonmechanistic Rationalism have a hard time explaining how we come to know various non-natural facts. There are various facts that do not plausibly impinge on the output of the laws of physics but that we know nonetheless. Only NR explains that fact. NR is, of course, controversial, probably being inconsistent with many views of consciousness including physicalism and epiphenomenalism (you could probably make those views consistent with it though, if you adopt a view like Penrose’s, but it would come at some cost). Interactionist dualism is the best fit for such a view, but that’s okay, because interactionist dualism is the best view, and so is NR. Finding that the two best views require each other supports both of them.

This also gives some reason to be a theist or an axiarchist. Theism has a natural explanation of why we’d have a basic capacity for reason in a way that can’t be accounted for in terms of simple mechanisms. Of course, it has about a million problems, including explaining why our reasoning is so fallible, but this still might be some evidence.

NR seems unappealing at first. But once you realize that it’s a prerequisite for our having knowledge of lots of things we obviously know, it becomes clear that it is right. The brain is not just a meat computer; we have a unique non-mechanical ability to reason.

I both started and finished this post today! Amazing how productive one can be when putting off studying for midterms (don’t worry family members reading this, I still studied a lot!)