Often, people think about their self-worth/self-confidence/self-esteem/self-efficacy/self-worth in ways which seem really strange from a simplistic decision-theoretic perspective. (I’m going to treat all those terms as synonyms, but, feel free to differentiate between them as you see fit!) Why might you “need confidence” in order to try something, even when it is obviously your best bet? Why might you constantly worry that you’re “not good enough” (seemingly no matter how good you become)? Why do people especially suffer from this when they see others who are (in some way) much better than them, even when there is clearly no threat to their personal livelihood? Why might you think about killing yourself due to feeling worthless? (Is there an evo-psych explanation that makes sense, given how contrary it seems to survival of the fittest?)
There might be a lot of diverse explanations for the diverse phenomena. I think providing more examples of puzzling phenomena is an equally valuable way to answer (though maybe those should be a comment rather than an answer?).
This seems connected to the puzzling way people constantly seem to want to believe good things (even contrary to evidence) in order to feel good, and fear failure even when the alternative is not trying & essentially failing automatically.
Some sketchy partial explanations to start with:
Maybe there is a sense in which we manage the news constantly. It could be that we have a mental architecture which looks a lot like a model-free RL agent connected up to a world model, being rewarded for taking actions which increase expected value according to the world-model. The model-free RL will fool the world-model where it can, but this will be ineffective in any case where the world-model understands such manipulation. So things basically even out to rational behavior, but there’s always some self-delusion going on at the fringes. (This only has to do with the observation that people sometimes try to make themselves feel better by finding arguments/activities which boost self-esteem, not with other weird aspects of self-esteem.)
There’s a theory that, in order to be trustworthy bargaining partners, people evolved to feel guilty/shameful when they violate trust. You can tell who feels more guilt/shame after some interaction with them, and you can expect these people to violate trust less often since it is more costly for them. Therefore feelings of guilt/shame can be an advantage. Self-worth may be connected to how this is implemented internally. So, according to this theory, low self-worth is all about self-punishment.
Previously, I thought that self-worth was like an estimate of how valuable you are to your peers, which serves as an estimate of what resources you can bargain for (or, how strong of a bid can you successfully make for the group to do what you want) and how likely you are to be thrown out of the coalition.
Now I think there’s an extra dimension which has to do with simpler dominance-hierarchy behavior. Many animals have dominance hierarchies; humans have more complicated coordination strategies which involve a lot of other factors, but still display very classic dominance-hierarchy behavior sometimes. In a dominance-hierarchy system, it just makes sense to carry around a little number in your head which says how great (/terrible) a person you are, and engage in a lot of varying behaviors depending on your place in the hierarchy. Someone who is low in the hierarchy has to walk with their tail between their legs, metaphorically, which means displaying caution and deference. Maybe you have trouble talking to people because you need to show fear to your superiors.
From the literature on self esteem
is sociometer theory and
is hierometer theory.
Hierometer theory is relatively new (2016) and could be though of as a subset of sociometer theory if sociometer theory is interpreted more broadly. Accordingly it has less research backing it up and that which is there is mostly by the original proponents of the theory.
This paper gives an introduction to both and a summary of evidence (I found this diagram a useful explanation of the difference). The paper suggests that both are true to some extent and complement each other.
I’ve included some quotes below.
Sociometer theory
Hierometer theory
Findings
Cool, thanks!
My current theory is that self-esteem isn’t about yourself at all!
Self-esteem is your estimate of how much help/support/contribution/love you can get from others.
This explains why a person needs to feel a certain amount of “confidence” before trying something that is obviously their best bet. By “confidence” we basically just mean “support from other people or the expectation of same.” The kinds of things that people usually need “confidence” to do are difficult and involve the risk of public failure and blame, even if they’re clearly the best option from an individual perspective.
This makes a lot of sense to me. It fits in with my sense that:
people with low self-esteem don’t expect others to help/support them long-term, as they feel they’ll be ‘found out’ as being worthless/terrible
whenever anyone is helpful to them they respond disproportionately effusively and say how kind the other person is.
In future I will model surprising low self-esteem as failing to accurately read signals about their level of respect/power. And that people with appropriately low levels of low self-esteem should focus on being useful to the people and communities around them.
Self-esteem existing seems to be a consequence of self-identities existing in the first place.
Having a self-identity in turn makes it easier for an agent to reason about how the agent should interact with it’s environment because the self-identity is where the directly accessible information the agent has about itself is stored.
It seems to me straightforward that an agent optimizes for a future state where it’s self-identity or the information it stores about itself is positive.
The book Transform Your Self by Steve Andreas provides a good layout how self-identity works for practical purposes and how one can rewrite it.
So let me take my current working detailed model of how human minds work and see what it says about self-esteem, a phenomenon I’m fairly familiar with as I’ve done some work on it in the past and was successful such that I used to have self-esteem “problems” and now I don’t, and the problems I do have that look sort of like issues of self-esteem have deeper roots than what is normally addressed in self-esteem self-help and positive psychology literature from what I’ve read.
To the extent that we can model human minds as hierarchies of control systems that aim to minimize prediction error and maintain various set points (presumably for adaptive reasons), we’d expect it to look like humans, using our ability to observe the behavior of others and introspect, are both trying to make correct predictions about how much esteem (seemingly a metric that tracts an important generator of status, prestige, and generalize capability to get things done) they have and maintain an amount of esteem necessary to fulfilling other preferences (i.e. minimizing the error in other predictions and maintaining other set points). This will result in a few different behaviors depending on conditions, remembering that in this model humans do thing by predicting that they will to cause the necessary neurons to fire to cause the predicted behaviors to happen:
Evidence suggests esteem is lower than thought, all else equal: person feels down, sad, dejected, etc. as they make an update away from the desired set point of current or higher esteem. This is the person who has just been let down in a bad way, like losing their job, a friend, or being told they have lost esteem.
Person feels bad that esteem is not higher, all else equal: they feel bad because they are regularly suffering prediction error while trying to aim for the amount of esteem their esteem set point suggests they should have. This is the person who is depressed or despondent that they can’t seem to do anything to get as much esteem as they think they deserve.
Person fells better after accepting they are a “loser”, all else equal: people with low self esteem sometimes identify with and revel in their low esteem, and engage in behaviors that keep their esteem low because it keeps prediction error low; in this case the set point gets revised downwards. This is the person who accepts low status as one of society’s losers and acts to reinforce that position.
Evidence suggests esteem is higher than thought, all else equal: person feels elated, excited, happy, etc. as they make an update upward that they have more esteem than their desired set point predicted they would have. This is the person who thought they were not going to get a job, make a friend, go on a date, etc. and then it happens and they are happy surprised.
Person feels happy anxiousness (manic?) that esteem is not lower, all else equal: they feel good because they are over their esteem set point but also worried that it might be prediction error and so operate out of fear of prediction error coming true and esteem being as low as the set point. This is the person who has a high paying job, a great romantic partner, or otherwise is vested with lots of esteem who also thinks it is underserved and is attached to holding on to that esteem and preventing the mistake from being realized.
Person feels better after accepting they are a “winner”, all else equal: people with high self esteem sometimes identify with and revel in their high esteem, and engage in behavior that keep their esteem high because it keeps prediction error low; this feels to many like a natural state of good, skilled, qualified, etc. people getting the rewards they feel they deserve. This is the person who accepts they are a winner and comfortably acts to keep their esteem high.
Other outcomes seem possible, those are just the situations that were salient to me.
This also suggests a way to fix self-esteem, which I think matches most of what I’ve read about it: find a way to bootstrap esteem generation such that esteem error prediction is reduced and esteem starts moving up towards the set point, and specifically adopt a belief seeing you are more esteemed than you thought is part of why you are worthing of esteem such that it causes a positive feedback loop that takes you up to and maybe even past your set point.
On the other side, it also matches what we hear about imposter syndrome: find evidence that you are not in error so that you stop being surprised and maybe also move your set point up to its new value.
Finally, it suggests a way to avoid all this suffering over esteem: learn not to have a fixed esteem set point such that you can only make prediction errors, not feel bad that your prediction is also not predicting the set point.
I should note that my answer does not address the question “what is esteem/worth/etc.?”. I think that’s a trickier question because it’s actually a whole bunch of stuff we cluster together for a wide variety of reasons as a result of dependencies in how we assess meaning in ways that we know how to make legible, so I ignore it here and just assume esteem is some kind of metric, specifically meaning that it can be measured and that the measure looks enough like a real number that we can pretend it is. Reality is likely more complicated than that, but I’m not interested in exploring a full reduction of esteem right now, so this seems a reasonable enough practical choice to at least say something as it relates to how esteem-like processes operate in the mind under this theory.
Thanks for this answer. Describing how prediction error minimization theory applies to self-esteem in such a clear way helped me understand both that theory as well as self-esteem more clearly.
I’m confused by the “way to fix self-esteem” you describe. I do not understand how “esteem generation” might reduce “esteem error prediction”. What would be a concrete example for such a process? I haven’t done any research on that topic though, except reading maybe 2 or 3 articles on LW, so I might well be missing some crucial parts of how the theory works. Can you recommend any ressources which might serve as a good starting point?
Maybe it will help if I make the model more formal, since there are multiple variables and it can be a little hard to see what’s going on if you don’t already have the intuitions to track them from normal words.
Let m be a measure of self esteem as a result of seeing some evidence e about the world (an observation). The prediction error is the extent to which E:E→M (the expectation relation that goes from evidence to a measure of self esteem) diverges from A:E→M (the relation that calculates the actual update in measure of self esteem). So prediction error looks like when E(e)=m≠A(e), and the larger the difference between E(e) and A(e) the larger the error.
A self-esteem set point, s∈M, is a measure of self esteem you are targeting such that if A(e)≠s then you want (set an expectation that E(e′)=s for some as yet unobserved evidence e′) to increase or lower your observed self-esteem A(e) such that it matches s.
Warning: This is an off-the-cuff model of the theory I just made up right now, so it’s probably non-standard and I’d have to think/read about the formalism more to fully endorse it (it’s also a little slopping in a couple places because I’m out of practice). I mean to use it only as a pedagogical tool here.
When I suggest we can fix self-esteem, I mean we can work to adjust A(e) and E(e) so that they better match, and work to alter A such that the actual esteem you observe yourself to have also matches the set point s. What that looks like in the case of wanting more esteem than you currently have and repeatedly expecting to have more esteem than you observe yourself having (the case where E(e)<A(e)) and it being true that A(e)<s. The fix in this case is to take actions that cause E(e) to rise to match A(e) and take actions that cause A(e) to rise to match s, and even better if this can be done in concert by making A conditional on |E(e)−A(e)| such that A(e) increases as A(e)−E(e) decreases towards 0.
Oh, to your other question, I have no special resources to recommend. I mostly arrived at a version of this theory on my own, then found out about Friston and perceptual control theory and was like “ah, great, someone already worked this out, one less thing for me to worry about!”.
In my recent post on value drift I had a paragraph with a bunch of links, though I think they are the things you may have already seen. For completeness and in case you missed any of the ones I liked, here it is copied into this comment:
The pursuit of high self-esteem is a cultural mistake. There are touted benefits I don’t think I need to repeat, but there are costs as well: distorted or narcissistic self-perception, emotional instability due to contingent self-worth, and hostility towards those who threaten one’s ego. These are traits unbecoming of a rationalist, to say the least. More recent research points to self-compassion as the superior strategy. The attitude produces the same benefits, but without the drawbacks. [Epistemic status: there is published research on this topic, but I’m generally suspicious of findings in psychology due to the replication crisis.]
There’s another model of self-esteem that I’ve found usefully predictive, in addition to models about status and hierarchy, which doesn’t seem to be mentioned here.
The model is that we have an image in our heads of our “ideal self.” That is, the person we think we “should be.” This model comes from a number of places, including other peoples expectations of who we should be, who we think is the person that can accomplish the things we want to accomplish, etc.
In addition, we have a model of our heads at any given moment of “who we are”, our self-concept and identity. This also comes from a number of places, including what other people say about us, the way we interpret memories and evidence from our own life, etc.
The level of self-esteem is based on the overlap between our ideal self and our self-concept.
Image: https://nursekey.com/wp-content/uploads/2017/02/B9780323091145000177_f17-02ab-9780323091145.jpg
Yeah, you got it right. You wanna take as much as possible from others without getting slaughtered, so you keep track of your status. Not much to it.
You get a whole lot of pathological anxiety and suicide these days because the environment has shifted somewhat, Instagram and billionaires and precarious labor and whatnot. I would like to see the numbers for suicide pre-industrial revolution; I wouldn’t expect a lot of them.
The “environment” also shifted dramatically with the agricultural revolution. I think there was plenty of angst to go around in the pre-industrial era. I would not expect the suicide rate to have increased on average.