I dislike overstatement

Jan 10 JDN 2459225

I was originally planning on titling this post “I hate overstatement”, but I thought that might be itself an overstatement; then I considered leaning into the irony with something like “Overstatement is the worst thing ever”. But no, I think my point best comes across if I exemplify it, rather than present it ironically.

It’s a familiar formula: “[Widespread belief] is wrong! [Extreme alternative view] is true! [Obvious exception]. [Further qualifications]. [Revised, nuanced view that is only slightly different from the widespread belief].”

Here are some examples of the formula (these are not direct quotes but paraphrases of their general views). Note that these are all people I basically agree with, and yet I still find their overstatement annoying:

Bernie Sanders: “Capitalism is wrong! Socialism is better! Well, not authoritarian socialism like the Soviet Union. And some industries clearly function better when privatized. Scandinavian social democracy seems to be the best system.”

Richard Dawkins: “Religion is a delusion! Only atheists are rational! Well, some atheists are also pretty irrational. And most religious people are rational about most things most of the time, and don’t let their religious beliefs interfere too greatly with their overall behavior. Really, what I mean to say that is that God doesn’t exist and organized religion is often harmful.”

Black Lives Matter: “Abolish the police! All cops are bastards! Well, we obviously still need some kind of law enforcement system for dealing with major crimes; we can’t just let serial killers go free. In fact, while there are deep-seated flaws in police culture, we could solve a lot of the most serious problems with a few simple reforms like changing the rules of engagement.”

Sam Harris is particularly fond of this formula, so here is a direct quote that follows the pattern precisely:

“The link between belief and behavior raises the stakes considerably. Some propositions are so dangerous that it may even be ethical to kill people for believing them. This may seem an extraordinary claim, but it merely enunciates an ordinary fact about the world in which we live. Certain beliefs place their adherents beyond the reach of every peaceful means of persuasion, while inspiring them to commit acts of extraordinary violence against others. There is, in fact, no talking to some people. If they cannot be captured, and they often cannot, otherwise tolerant people may be justified in killing them in self-defense. This is what the United States attempted in Afghanistan, and it is what we and other Western powers are bound to attempt, at an even greater cost to ourselves and to innocents abroad, elsewhere in the Muslim world. We will continue to spill blood in what is, at bottom, a war of ideas.”

Somehow in a single paragraph he started with the assertion “It is permissible to punish thoughtcrime with death” and managed to qualify it down to “The Afghanistan War was largely justified”. This is literally the difference between a proposition fundamentally antithetical to everything America stands for, and an utterly uncontroversial statement most Americans agree with. Harris often complains that people misrepresent his views, and to some extent this is true, but honestly I think he does this on purpose because he knows that controversy sells. There’s taking things out of context—and then there’s intentionally writing in a style that will maximize opportunities to take you out of context.

I think the idea behind overstating your case is that you can then “compromise” toward your actual view, and thereby seem more reasonable.

If there is some variable X that we want to know the true value of, and I currently believe that it is some value x1 while you believe that it is some larger value x2, and I ask you what you think, you may not want to tell me x2. Intead you might want to give some number even larger than x2 that you choose to try to make me adjust all the way into adopting your new belief.

For instance, suppose I think the probability of your view being right is p and the probability of my view being right is 1-p. But you think that the probability of your view being right is q > p and the probability of my view being right is 1-q < 1-p.

I tell you that my view is x1. Then I ask you what your view is. What answer should you give?

Well, you can expect that I’ll revise my belief to a new value px + (1-p)x1, where x is whatever answer you give me. The belief you want me to hold is qx2 + (1-q)x1. So your optimal choice is as follows:

qx2 + (1-q)x1 = px + (1-p)x1

x = x1 + q/p(x2-x1)

Since q > p, q/p > 1 and the x you report to me will be larger than your true value x2. You will overstate your case to try to get me to adjust my beliefs more. (Interestingly, if you were less confident in your own beliefs, you’d report a smaller difference. But this seems like a rare case.)

In a simple negotiation over dividing some resource (e.g. over a raise or a price), this is quite reasonable. When you’re a buyer and I’m a seller, our intentions are obvious enough: I want to sell high and you want to buy low. Indeed, the Nash Equilibrium of this game seems to be that we both make extreme offers then compromise on a reasonable offer, all the while knowing that this is exactly what we’re doing.

But when it comes to beliefs about the world, things aren’t quite so simple.

In particular, we have reasons for our beliefs. (Or at least, we’re supposed to!) And evidence isn’t linear. Even when propositions can be placed on a one-dimensional continuum in this way (and quite frankly we shoehorn far too many complex issues onto a simple “left/right” continuum!), evidence that X = x isn’t partial evidence that X = 2x. A strong argument that the speed of light is 3*108 m/s isn’t a weak argument that the speed of light is 3*109 m/s. A compelling reason to think that taxes should be over 30% isn’t even a slight reason to think that taxes should be over 90%.

To return to my specific examples: Seeing that Norway is a very prosperous country doesn’t give us reasons to like the Soviet Union. Recognizing that religion is empirically false doesn’t justify calling all religious people delusional. Reforming the police is obviously necessary, and diverting funds to other social services is surely a worthwhile goal; but law enforcement is necessary and cannot simply be abolished. And defending against the real threat of Islamist terrorism in no way requires us to institute the death penalty for thoughtcrime.

I don’t know how most people response to overstatement. Maybe it really does cause them to over-adjust their beliefs. Hyperbole is a very common rhetorical tactic, and for all I know perhaps it is effective on many people.

But personally, here is my reaction: At the very start, you stated something implausible. That has reduced your overall credibility.

If I continue reading and you then deal with various exceptions and qualifications, resulting in a more reasonable view, I do give you some credit for that; but now I am faced with a dilemma: Either (1) you were misrepresenting your view initially, or (2) you are engaging in a motte-and-bailey doctrine, trying to get me to believe the strong statement while you can only defend the weak statement. Either way I feel like you are being dishonest and manipulative. I trust you less. I am less interested in hearing whatever else you have to say. I am in fact less likely to adopt your nuanced view than I would have been if you’d simply presented it in the first place.

And that’s assuming I have the opportunity to hear your full nuanced version. If all I hear is the sound-byte overstatement, I will come away with an inaccurate assessment of your beliefs. I will have been presented with an implausible claim and evidence that doesn’t support that claim. I will reject your view out of hand, without ever actually knowing what your view truly was.

Furthermore, I know that many others who are listening are not as thoughtful as I am about seeking out detailed context, so even if I know the nuanced version I know—and I think you know—that some people are going to only hear the extreme version.

Maybe what it really comes down to is a moral question: Is this a good-faith discussion where we are trying to reach the truth together? Or is this a psychological manipulation to try to get me to believe what you believe? Am I a fellow rational agent seeking knowledge with you? Or am I a behavior machine that you want to control by pushing the right buttons?

I won’t say that overstatement is always wrong—because that would be an overstatement. But please, make an effort to avoid it whenever you can.

There is no problem of free will, just a lot of really confused people

Jan 15, JDN 2457769

I was hoping for some sort of news item to use as a segue, but none in particular emerged, so I decided to go on with it anyway. I haven’t done any cognitive science posts in awhile, and this is one I’ve been meaning to write for a long time—actually it’s the sort of thing that even a remarkable number of cognitive scientists frequently get wrong, perhaps because the structure of human personality makes cognitive science inherently difficult.

Do we have free will?

The question has been asked so many times by so many people it is now a whole topic in philosophy. The Stanford Encyclopedia of Philosophy has an entire article on free will. The Information Philosopher has a gateway page “The Problem of Free Will” linking to a variety of subpages. There are even YouTube videos about “the problem of free will”.

The constant arguing back and forth about this would be problematic enough, but what really grates me are the many, many people who write “bold” articles and books about how “free will does not exist”. Examples include Sam Harris and Jerry Coyne, and have been published in everything from Psychology Today to the Chronicle of Higher Education. There’s even a TED talk.

The worst ones are those that follow with “but you should believe in it anyway”. In The Atlantic we have “Free will does not exist. But we’re better off believing in it anyway.” Scientific American offers a similar view, “Scientists say free will probably doesn’t exist, but urge: “Don’t stop believing!””

This is a mind-bogglingly stupid approach. First of all, if you want someone to believe in something, you don’t tell them it doesn’t exist. Second, if something doesn’t exist, that is generally considered a pretty compelling reason not to believe in it. You’d need a really compelling counter-argument, and frankly I’m not even sure the whole idea is logically coherent. How can I believe in something if I know it doesn’t exist? Am I supposed to delude myself somehow?

But the really sad part is that it’s totally unnecessary. There is no problem of free will. There are just an awful lot of really, really confused people. (Fortunately not everyone is confused; there are those, such as Daniel Dennett, who actually understand what’s going on.)

The most important confusion is over what you mean by the phrase “free will”. There are really two core meanings here, and the conflation of them is about 90% of the problem.

1. Moral responsibility: We have “free will” if and only if we are morally responsible for our actions.

2. Noncausality: We have “free will” if and only if our actions are not caused by the laws of nature.

Basically, every debate over “free will” boils down to someone pointing out that noncausality doesn’t exist, and then arguing that this means that moral responsibility doesn’t exist. Then someone comes back and says that moral responsibility does exist, and then infers that this means noncausality must exist. Or someone points out that noncausality doesn’t exist, and then they realize how horrible it would be if moral responsibility didn’t exist, and then tells people they should go on believing in noncausality so that they don’t have to give up moral responsibility.

Let me be absolutely clear here: Noncausality could not possibly exist.

Noncausality isn’t even a coherent concept. Actions, insofar as they are actions, must, necessarily, by definition, be caused by the laws of nature.

I can sort of imagine an event not being caused; perhaps virtual electron-positron pairs can really pop into existence without ever being caused. (Even then I’m not entirely convinced; I think quantum mechanics might actually be deterministic at the most fundamental level.)

But an action isn’t just a particle popping into existence. It requires the coordinated behavior of some 10^26 or more particles, all in a precisely organized, unified way, structured so as to move some other similarly large quantity of particles through space in a precise way so as to change the universe from one state to another state according to some system of objectives. Typically, it involves human muscles intervening on human beings or inanimate objects. (Recently it has come to mean specifically human fingers on computer keyboards a rather large segment of the time!) If what you do is an action—not a muscle spasm, not a seizure, not a slip or a trip, but something you did on purpose—then it must be caused. And if something is caused, it must be caused according to the laws of nature, because the laws of nature are the laws underlying all causality in the universe!

And once you realize that, the “problem of free will” should strike you as one of the stupidest “problems” ever proposed. Of course our actions are caused by the laws of nature! Why in the world would you think otherwise?

If you think that noncausality is necessary—or even useful—for free will, what kind of universe do you think you live in? What kind of universe could someone live in, that would fit your idea of what free will is supposed to be?

It’s like I said in that much earlier post about The Basic Fact of Cognitive Science (we are our brains): If you don’t think a mind can be made of matter, what do you think minds are made of? What sort of magical invisible fairy dust would satisfy you? If you can’t even imagine something that would satisfy the constraints you’ve imposed, did it maybe occur to you that your constraints are too strong?

Noncausality isn’t worth fretting over for the same reason that you shouldn’t fret over the fact that pi is irrational and you can’t make a square circle. There is no possible universe in which that isn’t true. So if it bothers you, it’s not that there’s something wrong with the universe—it’s clearly that there’s something wrong with you. Your thinking on the matter must be too confused, too dependent on unquestioned intuitions, if you think that murder can’t be wrong unless 2+2=5.

In philosophical jargon I am called a “compatibilist” because I maintain that free will and determinism are “compatible”. But this is much too weak a term. I much prefer Eleizer Yudkowsky’s “requiredism”, which he explains in one of the greatest blog posts of all time (seriously, read it immediately if you haven’t before—I’m okay with you cutting off my blog post here and reading his instead, because it truly is that brilliant), entitled simply “Thou Art Physics”. This quote sums it up briefly:

My position might perhaps be called “Requiredism.” When agency, choice, control, and moral responsibility are cashed out in a sensible way, they require determinism—at least some patches of determinism within the universe. If you choose, and plan, and act, and bring some future into being, in accordance with your desire, then all this requires a lawful sort of reality; you cannot do it amid utter chaos. There must be order over at least over those parts of reality that are being controlled by you. You are within physics, and so you/physics have determined the future. If it were not determined by physics, it could not be determined by you.

Free will requires a certain minimum level of determinism in the universe, because the universe must be orderly enough that actions make sense and there isn’t simply an endless succession of random events. Call me a “requiredist” if you need to call me something. I’d prefer you just realize the whole debate is silly because moral responsibility exists and noncausality couldn’t possibly.

We could of course use different terms besides “free will”. “Moral responsibility” is certainly a good one, but it is missing one key piece, which is the issue of why we can assign moral responsibility to human beings and a few other entities (animals, perhaps robots) and not to the vast majority of entities (trees, rocks, planets, tables), and why we are sometimes willing to say that even a human being does not have moral responsibility (infancy, duress, impairment).

This is why my favored term is actually “rational volition”. The characteristic that human beings have (at least most of us, most of the time), which also many animals and possibly some robots share (if not now, then soon enough), which justifies our moral responsibility is precisely our capacity to reason. Things don’t just happen to us the way they do to some 99.999,999,999% of the universe; we do things. We experience the world through our senses, have goals we want to achieve, and act in ways that are planned to make the world move closer to achieving those goals. We have causes, sure enough; but not just any causes. We have a specific class of causes, which are related to our desires and intentions—we call these causes reasons.

So if you want to say that we don’t have “free will” because that implies some mysterious nonsensical noncausality, sure; that’s fine. But then don’t go telling us that this means we don’t have moral responsibility, or that we should somehow try to delude ourselves into believing otherwise in order to preserve moral responsibility. Just recognize that we do have rational volition.

How do I know we have rational volition? That’s the best part, really: Experiments. While you’re off in la-la land imagining fanciful universes where somehow causes aren’t really causes even though they are, I can point to not only centuries of human experience but decades of direct, controlled experiments in operant conditioning. Human beings and most other animals behave quite differently in behavioral experiments than, say, plants or coffee tables. Indeed, it is precisely because of this radical difference that it seems foolish to even speak of a “behavioral experiment” about coffee tables—because coffee tables don’t behave, they just are. Coffee tables don’t learn. They don’t decide. They don’t plan or consider or hope or seek.

Japanese, as it turns out, may be a uniquely good language for cognitive science, because it has two fundamentally different verbs for “to be” depending on whether an entity is sentient. Humans and animals imasu, while inanimate objects merely arimasu. We have free will because and insofar as we imasu.

Once you get past that most basic confusion of moral responsibility with noncausality, there are a few other confusions you might run into as well. Another one is two senses of “reductionism”, which Dennett refers to as “ordinary” and “greedy”:

1. Ordinary reductionism: All systems in the universe are ultimately made up of components that always and everywhere obey the laws of nature.

2. Greedy reductionism: All systems in the universe just are their components, and have no existence, structure, or meaning aside from those components.

I actually had trouble formulating greedy reductionism as a coherent statement, because it’s such a nonsensical notion. Does anyone really think that a pile of two-by-fours is the same thing as a house? But people do speak as though they think this about human brains, when they say that “love is just dopamine” or “happiness is just serotonin”. But dopamine in a petri dish isn’t love, any more than a pile of two-by-fours is a house; and what I really can’t quite grok is why anyone would think otherwise.

Maybe they’re simply too baffled by the fact that love is made of dopamine (among other things)? They can’t quite visualize how that would work (nor can I, nor, I think, can anyone in the world at this level of scientific knowledge). You can see how the two-by-fours get nailed together and assembled into the house, but you can’t see how dopamine and action potentials would somehow combine into love.

But isn’t that a reason to say that love isn’t the same thing as dopamine, rather than that it is? I can understand why some people are still dualists who think that consciousness is somehow separate from the functioning of the brain. That’s wrong—totally, utterly, ridiculously wrong—but I can at least appreciate the intuition that underlies it. What I can’t quite grasp is why someone would go so far the other way and say that the consciousness they are currently experiencing does not exist.

Another thing that might confuse people is the fact that minds, as far as we know, are platform independentthat is, your mind could most likely be created out of a variety of different materials, from the gelatinous brain it currently is to some sort of silicon supercomputer, to perhaps something even more exotic. This independence follows from the widely-believed Church-Turing thesis, which essentially says that all computation is computation, regardless of how it is done. This may not actually be right, but I see many reasons to think that it is, and if so, this means that minds aren’t really what they are made of at all—they could be made of lots of things. What makes a mind a mind is how it is structured and above all what it does.

If this is baffling to you, let me show you how platform-independence works on a much simpler concept: Tables. Tables are also in fact platform-independent. You can make a table out of wood, or steel, or plastic, or ice, or bone. You could take out literally every single atom of a table and replace it will a completely different atom of a completely different element—carbon for iron, for example—and still end up with a table. You could conceivably even do so without changing the table’s weight, strength, size, etc., though that would be considerably more difficult.
Does this mean that tables somehow exist “beyond” their constituent matter? In some very basic sense, I suppose so—they are, again, platform-independent. But not in any deep, mysterious sense. Start with a wooden table, take away all the wood, and you no longer have a table. Take apart the table and you have a bunch of wood, which you could use to build something else. There is no “essence” comprising the table. There is no “table soul” that would persist when the table is deconstructed.

And—now for the hard part—so it is with minds. Your mind is your brain. The constituent atoms of your brain are gradually being replaced, day by day, but your mind is the same, because it exists in the arrangement and behavior, not the atoms themselves. Yet there is nothing “extra” or “beyond” that makes up your mind. You have no “soul” that lies beyond your brain. If your brain is destroyed, your mind will also be destroyed. If your brain could be copied, your mind would also be copied. And one day it may even be possible to construct your mind in some other medium—some complex computer made of silicon and tantalum, most likely—and it would still be a mind, and in all its thoughts, feelings and behaviors your mind, if not numerically identical to you.

Thus, when we engage in rational volition—when we use our “free will” if you like that term—there is no special “extra” process beyond what’s going on in our brains, but there doesn’t have to be. Those particular configurations of action potentials and neurotransmitters are our thoughts, desires, plans, intentions, hopes, fears, goals, beliefs. These mental concepts are not in addition to the physical material; they are made of that physical material. Your soul is made of gelatin.

Again, this is not some deep mystery. There is no “paradox” here. We don’t actually know the details of how it works, but that makes this no different from a Homo erectus who doesn’t know how fire works. Maybe he thinks there needs to be some extra “fire soul” that makes it burn, but we know better; and in far fewer centuries than separate that Homo erectus from us, our descendants will know precisely how the brain creates the mind.

Until then, simply remember that any mystery here lies in us—in our ignorance—and not in the universe. And take heart that the kind of “free will” that matters—moral responsibility—has absolutely no need for the kind of “free will” that doesn’t exist—noncausality. They’re totally different things.