Conviction without self-deception

"Believe in yourself" is perhaps the most common trope to be found in self-help books and motivational texts. It appears in fiction (especially children's books), film (especially sports films), and motivational posters. Coaches of sports teams labor to inspire. Low morale is the bane of teams.

The model, I think, goes something like this. Imagine two people who are about to swim in a 10-person race. They're both standing on the diving platform. The first one knows how their performance compares to the performance of their opponents, and knows that they have (statistically) a 17% chance of winning. The second one ignores all the odds, and has psyched themselves up, and is certain that they're going to win. They can feel the conviction in their body; their mind shuts down any hint of a thought to the contrary; they are pumped, excited, and filled with that sense of certainty.

Common wisdom says that the second person is going to push themselves harder (and, thus, increase their relative chances of winning). The first person seems less able to tap into their reserves, and more liable to despair when they fall behind. The second person seems more likely to give the race everything they've got.

What gives? Is epistemic virtue (in the form of acknowledging a mere 17% chance of victory) harming the first swimmer? Is their respect for truth causing them to be worse at swimming?

The fledgeling rationalist might say "Yes, and this is why rationality is not for everyone: Rationality is only for people who care more about knowing true things than about succeeding at physical tasks like swim races."

The intermediate rationalist might say "No, that's wrong; rationality is not about having true beliefs, it is about winning. Having true beliefs is regularly useful, but when it stops being useful, stop doing it. If you race better by becoming certain you're going to win, then throw epistemics out the window, and become certain you're going to win the race."

This is a better sentiment, but still, I think, misguided.

What I say is: stop conflating feelings with beliefs.

The conviction that the second swimmer feels is not a belief. It's a feeling. It's a mental stance; a way of thinking. Maybe they're feeling excited. Maybe their heart rate is higher. Maybe they have a tingling feeling throughout their body. Maybe their thoughts are more focused and singular. But none of those things are beliefs; they are not statements about the world. I say, learn to detach the conviction from the statistics: You can still enter the mental state we refer to as "certainty that you're going to win", without in fact predicting victory with high credence.

Our language and our culture and our poorly designed brains make it very easy to conflate feelings with beliefs. For instance, there's this feeling that correlates with extreme confidence that we call "certainty," and thus, it's easy to imagine that that state is only accessible to people with extremely high confidence in some relevant proposition. But the feeling and the credence don't have to come in lockstep — the two can be disentangled.

How? I find that the answer is very different for different feelings (that common wisdom says are linked to strange epistemic states), and also very different for different people. I personally get pretty far by simply "getting out of my own way": Where a straw rationalist might feel the beginning glimmers of conviction or excitement or hope, and then squash it with a thought such as "but statistically this is very likely to fail", I simply... don't squash the glimmer. Because "statistically this is likely to fail" just doesn't bear on the feeling, from my perspective. I'm allowed to feel hopeful about a thing even while being well-calibrated on its chances of success. I'm allowed to feel conviction before a race, even if I'm well-calibrated about my odds. I don't need to lose the useful mental stances, simply because I'm better-calibrated.

When I was a kid, I got into a number of arguments with my brother. I remember various distinct feelings that I had in different types of arguments. Sometimes, I'd be uncertain, but pretty sure he was reasoning poorly. In those situations I'd feel a sense of caution, an impulse to deflect, and that impulse that has you wanting to raise your hands and say "look, I'm not sure myself, but...". Other times, I'd know that I was wrong, but I'd be unwilling to lose face. In those situations, I'd feel defiant or trapped, and I'd have impulses to escape or lash out. Other times, I'd have very high confidence in my own beliefs, and I'd feel a strong sense of certainty and conviction, which gave rise to feelings of frustration, or righteousness, or solidity.

Nowadays, through various methods, I've done some rewiring on which feelings correspond to which epistemic states. Throughout that rewiring, I've endeavored not to throw the baby out with the bathwater.

I don't know if you associate "certainty" with a feeling, but I still can — I associate it quite strongly with that feeling that I'd have in particular arguments as a kid, and with parts of what I imagine the second swimmer feeling when they're "certain they're going to win."

I suspect that many rationalists, upon learning that one can never be certain, simultaneously lose access to both the epistemic state of certainty, and to the feeling. They say, "well, I can never be certain of anything," and they start managing their beliefs differently, and they become less prone to overconfidence, and they become more amenable to evidence, which is all great. But simultaneously, I suspect many start finding the feeling that we label "certainty" to be repulsive.

The feeling is not the belief! For me, that feeling was strongly correlated with cognitive flaws in my youth (let's just say that I was not a very well-calibrated 8-year-old), and therefore I definitely treat it with some suspicion in similar contexts. But the feeling can still be useful in other contexts.

For example, it is useful to the swimmer.

I think it's important to tease apart feelings from beliefs. If you're standing on that diving platform, I think it's important to simultaneously know you have a 17% chance of victory, and fill yourself with the excitement, focus, and confidence of the second swimmer. Become able to tap into conviction, without any need for the self-deception.

One of the most common objections to truth-seeking that I have found is "if I believed the truth, then I wouldn't be able to feel [X] anymore, and my life would get worse," for values of [X] including "hope," "happiness," and "conviction." So I say: disentangle the feelings from the beliefs. Detatch the grim-o-meter. Be a little reckless. Just because we call the feeling "certainty" doesn't mean that you're only allowed to feel it when your confidence is unreasonably high.