Robin, that's what I've been objecting to this whole time. I accept that my mind is a big system, that the pressures in it don't all work the way "I" (that is, the deliberative battleground) wish they would. It may even be that the causes of my actions are sometimes not what the little voice in my head thinks they are. None of that means my ideals are not what I chose them to be. It means that, having made my decision, I now need to enforce it.

Expand full comment

Eliezer, I expect many future posts on this topic, that go beyond my initial advice, which was to accept that your motives are less ideal than you wish.

Expand full comment

Robin, could you say what you think we should *do about* this bias, to overcome it? A confusing belief should be cashed out as an anticipated experience, and oftimes it is more revealing yet to speak of actions.

Expand full comment

Eliezer, I think we are having a serious problem finding language to express our points of view. For example, I don't see how you could think I claimed anything like that there have been zero successful altruists. I less concerned about whether I am pessimistic or optimistic, and more concerned that we understand what we are saying.

Carl, unless I say otherwise, all my claims are intended as tendencies, and I expect there to be exceptions.

Expand full comment

Robin, you're being way too pessimistic here. Yes, conscious thoughts can be twisted, warped, rearranged, subverted, retold in revisionist histories, but they are not epiphenomenal. The number of successful altruists in human history is not zero.

I point out that being extremely pessimistic is (a) a reason not to try (b) a social defense which argues that no one else could have done better. Surely, one ought to be a bit suspicious here...?

Expand full comment


My example was pointing to the deadweight losses of hypocritical behaviour: the distortion of idealistic motives by selfish unconscious drives is a Rube Goldbergian process that efficiently realizes neither selfish nor altruistic aims, nor a weighted average of the two according to their strength. If the two motives are considered and balanced explicitly by an honest agent, then there are Pareto-improving gains to be had by modifying action.

You wrote:"we can resolve our hypocrisy two ways: we can start really living up to our high ideals, or we can admit we don't care as much as we thought about those ideals," but both options will require concrete change.

Second, as I mentioned above, I specifically doubt that the "more honest will tend to have lower expectations about achieving ideal ends" is true across domains because the dishonest will frequently 1) not have made any rigorous estimates about success, or 2) not have thought about efficiency at all, only a 'fuzzy feeling.' Honesty reduces self-serving bias in expectations about achievement, but improvements in efficiency make the overall effect ambiguous.

Third, the example incorporated an actual residual desire (10%) for idealism. You seem to be alternating between two characterizations of self-deception: in one, idealistic desires are truly epiphenomenal (#1 in my earlier comment) as in your corporate PRD model, while in the other the desires actually have motivational force, but less than we imagine. The second seems more accurate: the 'PRD employees' are like environmental activists hired by oil companies to provide credible defenders, and fed a misleading line by management. If increased honesty/transparency makes it clear that they are wasting their efforts they may leave for Greenpeace, even if they can't change the company.

Expand full comment

I added a section to the post above, offering a useful analogy. Here is another analogy: Your conscious mind is like a boat on the ocean, taking credit for the way it rises and falls as ocean waves pass under it.

Expand full comment

Ahh, I see that you're using learn in the context of the battleground, whereas I thought you were using it more generally. I'm not aware of any evidence that the subconscious learns anything to try to defeat conscious attempts to better follow high-minded ideals.

Geez, every time I use the word "subconscious" I want to put a disclaimer by it. Hopefully you will forgive me.

"Which pressure wins out can quite reasonably be determined by which side *you* choose to be on."

I think that the ability to do this varies very widely amongst people, due to differences in intelligence (because proper introspection is hard) and disposition (ADD, depression, and the like make it very hard to win).

Expand full comment

Robin, deliberation is a battleground but the battleground itself can *choose sides*. You *are* your deliberation. You are the little voice in your own head. So if you, the battleground, pick the high-minded pressure, it's not going to be an even fight.

You, the battleground, are an ancestrally anomalous case; the pressuring circuitry did not evolve to strategically defeat an aware opponent, but only to influence an unaware one.

For the background theory of mind, it's a bit old, but see http://www.singinst.org/LOG....

Pdf, the "low motive" behavioral reinforcement / strategy-pressuring architecture certainly influences how all sorts of skills are learned, but it's questionable whether the system itself can be said to "learn", especially in a sense that contributes to defeating "high motives" when the two poles of the adaptation are brought into nonancestral conflict by our reflective knowledge of evolutionary psychology. Of course the answer could be, "Yes, it learns" but I'm not sure what evidence there is for that, or what sort of learning would take place. Obviously the system influences what is learned by us, the battlegrounds. Does the system learn how to influence us better? How? What parameters are being adjusted?

Expand full comment

Eliezer, yes of course, "the vast majority of the mind serves neither purpose specifically." You keep saying such things as if you think I said the opposite. You apparently have a particular complex theory of mind which I am so far finding it hard to interpret from your writings. So at the moment I can't think of what more to say that what I've already said. "Your" "deliberation" is not some autonomous process, it is in part a battle ground for these two tendencies we have been discussing. So the existence of deliberation per se doesn't say which side has what advantages.

Carl, the scenario you paint is one of thousands we could paint. I don't understand its relevance.

Expand full comment

"The "low motive" *circuitry* - not mind, but hardwired pressure on the mind - *cannot* deliberate, it *cannot* think, it probably cannot even learn"

There is evidence that several kinds of trauma, especially during childhood, can shape people's reactions. There's evidence for implicit memory for pictures that lasts 17 years.


There's evidence that cultural upbringing shapes some basic attitudes that may determine one's political predisposition. So I would not say that it cannot learn, depending on what "it" we're talking about.

On the other hand, I definitely agree that it can't deliberate.

Expand full comment

'We can usefully distinguish the parts of our minds that are recruited for the purpose of presenting ourselves as following high motives, and the parts of our minds that makes sure that in practice we achieve low motives.'

At most, we could single out a few pieces of circuitry as having been adapted specifically for those two purposes. But the vast majority of the mind serves neither purpose specifically, and is recruitable by both circuits. Circuitry does not control directly; it has its effect in the thoughts it shapes. By the time you actually think "I want to save the world..." or "...and becoming tribal chief and crushing my opponents seems like a good way to do it", you're thinking thoughts that were shaped by thoughts that were shaped by circuitry. And, as the above example suggests, in the ancestral environment, these pieces of circuitry were not opposed at all; they worked in smooth harmony. Human beings are not fought over by God and Satan; we were created by a single optimizer, natural selection.

'You describe these parts as "deliberate thought and learned skills" versus "emotion and hardwired reinforcers" and claim that the former can have more complex "knowledge and skills" and think faster than the later.'

*You* exist on a level of organization above the idealism circuitry and the behavior warps. Deliberation, "thinking", takes place via unique, non-repeating information flows, which we experience as being the little voice in our own heads. Knowledge and skills are individualized, learned, and persistent over time. Neural circuitry is species-universal and encoded into a comparatively small amount of genetic information.

It is reasonable for "you" to do things that run counter to the pressures of neural circuitry - even subtle warps - if you have more knowledge and skill at the art of rationally idealistic deliberation, than the neural circuitry originally evolved to warp.

You have a bipolar adaptation of idealistic circuitry and behavior-warping circuitry, which works in harmony in the ancestral environment, but tends to be set at odds once human beings achieve some skill at reflection, knowledge of history, and awareness of evolutionary psychology. Now the pressures diverge - not because of changes in the pressuring circuitry, but because of new knowledge in the rest of your mind. Which pressure wins out can quite reasonably be determined by which side *you* choose to be on. Because you, the deliberative process, are aware that you have a choice, which you can make explicitly, and then deliberate on how to shape your further deliberations so that the "high" pressures win out.

Human beings sure are complex creatures, aren't we?

'It is true that you are conscious mostly of the high motive mind activity. But this does not at all mean that the low motive mind activity cannot deliberate, learn, be complex, and think fast. And both parts are "you."'

Neither part is you. The "low motive" *circuitry* - not mind, but hardwired pressure on the mind - *cannot* deliberate, it *cannot* think, it probably cannot even learn. The low motive is a pressure on your deliberation. So are the high motives. But the pressure is applied at different points - remember, in the ancestral environment, these were not *opposed* pressures. They only come into opposition within you because you possess reflective knowledge that your ancestors didn't. If you realize this and deliberately choose a side, then your deliberate mind activity will be on the side of high motives; and while the "low motive" will continue to pressure your deliberation, it cannot actually deliberate apart from your primary stream of consciousness.

Expand full comment


I am skeptical that in your scenario one could suddenly gain an honest self-assessment (by your stipulation, without changing what we really care about) without affecting actions, even if motivations remained unchanged.

Take our inefficient charitable giver, and specify that he gives $5,000 to the opera he attends. He self-deceives himself into thinking that this is because he is an altruistic person, concerned with the welfare of everyone, not paying attention to the social status it brings among his opera-loving associates.

Then, one day on the road to Damascus our opera buff realizes that he 'really cared' about the social status, to the tune of 90%, while only 10% of his motivation was altruistic. It seems implausible that behaviour could be unchanged, even if the desires are constant: why not explicitly separate the two activities given this new knowledge? He would reallocate $500 to the optimal charity, almost certainly getting more than 10 times the altruistic ideal-attainment, beating any nebulous prior estimates of opera value (keeping in mind people who are not honest about the efficiency of their charities are likely not to have any quantitative estimates at all). The remaining $4,500 could then be allocated purely on the basis of status maximization (which might lead to greater net status via non-opera conspicuous consumption, without the altruistic/group-loyalty charade).

Expand full comment

Calca, this web forum is organized into posts and comments on those posts. Please don't take the comment section of any post as a place to ask any random question; stay on topic.

Expand full comment

I have a few more questions. Has anybody ever been able to change someone else's mind through honest debate? In my experience even when I succeed in getting partial agreement with my positions, invariably when I meet the person again some time later the person has reverted to his/her original position, in other words, my arguments didn't stick.

Secondly, can there be situations where even though bias has been spotted it is still better to keep it instead of overcoming it? Suppose for instance, that science "proves" that there are indeed differences between the sexes as far as the intellectual process is concerned. Isn't the idea that "man are created equal" a good bias to have?

You have probably already considered such questions.

Expand full comment


I'm just trying to understand, don't get me wrong. I'm just very skeptical at the moment. What worries me are situations like, say Stalinist Soviet Union. Take for instance this sentence of ChrisA: "the aim of this blog that it would be better to do so through rational analysis or at least using others rational analysis. Doing so must avoid wasted time/resources in pursuing mistaken goals or undertaking activities I shouldn't have. To think otherwise would be nihilistic to me..."Well 'nihilistic' was a common accusation thrown at dissidents in USSR. Who is to decide who is wasting time/resources, in the end I don't think we will ever know.

(But I like ChrisA's example of the with justice system.)

Expand full comment