Overcoming Our Vs. Others’ Biases

Many groups over the last century or so have defined themselves in terms of trying to overcome other people’s biases, while our group here is defined more in terms of trying to overcome our own biases.  What is the relation between these two types of activities?

Over the last century many groups have self-identified as repressed "minorities" and sought to organize themselves to act to shame and threaten so others will "repress" them less in various ways.  More recently many (but hardly all) of these groups have formed a coalition to together support all their demands, under the banner of supporting "diversity."  The argument seems to be that we all tend to be biased to treat poorly those who deviate from a certain ideal, e.g., rich middle-aged middle-height white male right-handed hearing Christian heterosexuals.   

We gather here instead to try to overcome our own biases, whatever those may be.  So we must confront our relation to this history.  If we accepted this coalition’s main claim, we would try to overcome our own bias favoring this standard ideal.  And we would have to judge how much of this bias remains in our culture, after the many successes of this coalition in remaking our culture.  If we accept a more general claim that we tend to be biased against all minorities of any sort, we would try to overcome our own bias against minority features or activities.  (Is there a better way to phrase a general claim here?)

Of course we must also consider the hypothesis that we are not biased against minorities in general, or this coalition in particular, either because we never were so biased or because we have already overcome such biases.  In this case we might view the continued lobbying of self-identified repressed minorities as just a selfish grab for more attention and deference than they deserve [added: or more charitably, just an honest mistake on their part.]

What say ye all?  And what evidence can help us decide?

GD Star Rating
loading...
Tagged as:
Trackback URL:
  • Caio

    Is it possible that certain biases, even if they were eradicated, are part of our cognitive makeup and would be relearned within a generation or two? I mean specifically in-group/out-group biases which can be based on any number of identifying factors, notably (but hardly limited to) race, ethnicity and religion. Even if they’re not expressed as outright bigotry, it might be a good thing that minority groups continue to assert themselves so society won’t slip into deeper bias and eventual bigotry.

  • http://apperceptual.wordpress.com/ Peter Turney

    I think there is a tendency in some of the discussion here, even in this post itself, to view rationality as a single, unique thing. I don’t think that overcoming bias necessarily means converging on a single unique ideal of rationality. In fact, I think it is rational to encourage diversity. From an evolutionary perspective, a diverse population is more robust to environmental changes. A wide range of ideas and perspectives on the world is more likely to evolve better models and better ways of understanding the world. I believe that striving for rationality will make us more diverse, not more homogeneous.

  • michael vassar

    It seems to me that another difference is that our project is aimed primarily at overcoming deviations from logic in our formal verbal utterances while diversity supporters are more concerned with overcoming deviations from fairness in our actions and in our beliefs, even if our utility functions value unfairness or if the relevant unfair beliefs correspond to (usually exaggerated versions of) statistical realities.

    I think that there is value in distinguishing between the lobbying of disenfranchised majorities and minorities, on the one hand, and that of enfranchised majorities on the other. Feminism is inextricably entangled with postmodernism because on the one hand we have, in at least some respects, an oppressed enfranchised majority, while on the other hand there is no way to make sense of an enfranchised oppressed majority within an enlightenment framework. Trouble is, the enlightenment was basically a hack for avoiding certain pathological patterns of discourse. We don’t yet know how to build institutions within which non-enlightenment assumptions are allowed but which still avoid the pathological patterns that those patterns. Figuring out how to build such institutions is, for me, the ultimate goal of this Overcoming Bias project. As such, hopefully we will continue, though so far continue sounds self-aggrandizing, we have barely started, to develop norms of deliberation which, among other things, enable coherent discourse about feminism without serious risk of degeneration into pseudo-intellectual free-for-all.

  • http://hanson.gmu.edu Robin Hanson

    Caio, continued vigilance is required for almost all our identified biases. For example we must continue to beware of assuming Earth is the center of the universe. But this effort seems quite different from the efforts to reach a point where, with such continued vigilance, the bias is basically fixed.

    Peter, I don’t see how I assumed anything singular about rationality.

    Michael, to the extent we think some levels of fairness are correct and others in error, overcoming bias in fairness is a subclass of the more general class of overcoming biases in all beliefs. So then, are there ways we should treat this subclass differently from others?

  • michael vassar

    Robin: Fairness is about actions for the most part, not about beliefs. Unfortunately, sometimes it is also held to be unfair to hold certain beliefs, either because those beliefs would make unfair actions personally or collectively utility maximizing, undermining fairness, or because known cognitive biases make it likely that holding those beliefs will, at least in unwary humans and possibly in all humans, implement a causal dynamic which produces other unfair beliefs, unfair actions, and incorrect beliefs.

    Concretely, we generally hold it to be unfair to make even statistically valid generalizations about a person based on features of that person which they are unable to change. There is a much weaker consensus for the attitude that it is unfair to make statistically valid generalizations about a person based on features that would be very personally costly to change. We don’t do this consistently. I’m unable to change where I attended college or the fact that I am short, and it would be costly for me to change my lack of an elite university PhD or of a standard American accent, but people are not held to act unfairly in making true or false generalizations from these facts the way they are in making generalizations from homosexual behavior (but see “Queer Eye for the Strait Guy” for an example of an exception) or religious affiliation (but see preferential treatment in governmental contracting for minority owned businesses utilizing a definition of minority which includes certain religious groups for another exception). The principles being followed, like so much else, seem to be roughly those of public choice theory constrained by broad vague cultural norms themselves constrained by human psychology.

  • http://yudkowsky.net/ Eliezer Yudkowsky

    Peter, probability theory and decision theory are pinned by a number of optimality and uniqueness proofs, in theory. So who gets to use the best-achievable rational beliefs and strategies, and who has to use suboptimal ones for the sake of “diversity”?

    Robin, I think that it’s important to distinguish between social discrimination and cognitive biases a la the conjunction fallacy. Just because one word “bias” is used to refer to both of these does not make them the same thing at all. There are cognitive biases underlying social discrimination but there is more to social discrimination than that, and in turn, social discrimination is only one small part of the universe of cognitive biases.

    We are gathered to overcome our own cogntive biases, not necessarily paying any special attention to social discrimination. Frankly, when I see an issue that a majority of the damn planet is worked up about, that cannot actually kill off the human species outright, I mentally label it “receiving proportionally too much attention” and work on more drastic problems that are receiving proportionally too little attention.

    I worry about women being told not to be rational, and I worry about myself as a male author not knowing when I am annoying women in my writings about rationality. These have the potential to get in the way of my working with half the planet, and incidentally, if there are existing anti-discrimination efforts in this space I haven’t encountered them. Other forms of discrimination are problems but they are not my problems.

  • http://occludedsun.wordpress.com Caledonian

    Do not fall into the error of thinking that people who use the same words mean the same things.

    People do not always have the same concept in mind when they discuss things like ‘fairness’ and ‘equality’, most especially when talking across a perceived political divide.

    The more positive the term is, and the more generally it is perceived and socialized as a self-evident good, the more likely it will be used as a smokescreen to hide the real assertions being made in the argument.

  • http://yudkowsky.net/ Eliezer Yudkowsky

    Actually, on rereading the main post I find that I didn’t respond to the primary question:

    Of course we must also consider the hypothesis that we are not biased against minorities in general, or this coalition in particular, either because we never were so biased or because we have already overcome such biases. In this case we might view the continued lobbying of self-identified repressed minorities as just a selfish grab for more attention and deference than they deserve.

    There are numerous powerful studies suggesting that, while we are not “biased against all minorities of any sort”, the application of negative social stereotypes has the same sort of human-universal status as all the other biases we discuss here. Can we trust this research? It seems very plausible to me from an a priori standpoint, even if it does happen to be research in an academically faddish area. I don’t see any particular reason why I would be immune to the non-conscious influences thus discovered, even though I carry out the prophylactics that any sensible person uses against the conscious sort.

    On reflection, my previous post was too harsh; many people feel I should restrict myself to talking only about AI, but I don’t restrict myself thusly. I might discuss general antidiscrimination efforts if I found something to say about them. What I dislike is the notion that I could be obligated to drop a line of thought about AI theory or rationality to think about feminism. I suppose I shouldn’t attack this, though, until someone actually defends it.

    The idea that we don’t need to worry about discrimination biases because we have already overcome them does not strike me as plausible.

  • josh

    Regarding the general claim that we are biased against minorities of any sort:

    Are we a priori biased against minority activities, or do we select our own activities first, and feel disdain for those of others; after all, we will prefer the activities we choose. In the second case, by definition, the majority will seem to disfavor the behavior of the minority. Minorities, likewise, can feel disdain for the behavior of majorities as they are often observed to do. This bias may not have anything to do with minorities or “ideals” per se unless we accept that there isn’t much consensus on what the ideal is.

    Of course, an individual’s understanding of the “ideal” is probably affected by the people they socialize with, so this could lead to people being biased toward too much ideal convergence (or divergence) among groups.

  • http://hanson.gmu.edu Robin Hanson

    Eliezer, I’ve heard that the lit on stereotypes mostly finds them to be accurate – some err in being too negative, while others err in being too positive. Happy to see a more thorough review though.

    When considering whether you will socially discriminate, you consider your beliefs about the characteristics of the people involved, beliefs about your preferences, and beliefs about who will censure you or retaliate how. Arguments directed to persuade people not to discriminate seem to largely argue some of these beliefs are in error.

  • Vladimir Slepnev

    Even more dangerous question. Even if we’re biased against others, are we the losers for it? Rationality, according to this blog, is all about winning. Why are we better off treating minorities fairly? Black Panthers-style threats and emotional blackmail discounted by default.

    Also, lovers of “diversity”, you can’t have it without jingoism and insular group values. Such measures as ensuring proportional representation of women in certain activities, or making housing discrimination illegal, don’t promote diversity – they promote a homogeneous society where all groups are alike in all aspects. Not saying this is good or bad, let’s just get the terms straight, and not say diversity when we mean homogeneity.

  • http://www.xuenay.net/ Kaj Sotala

    Note that feminist groups start out from the idea that one first has to realize the hidden biases and injustices in themselves, before they can begin correcting them in others. So the correct characterization would be that we try to overcome our own biases, while they attempt to overcome their own biases and those of others.

    Separating cognitive and social biases also feels somewhat artificial to me – we’ve had disagreement case studies (for instance) here before, so why not examples of how, say, confirmation bias leads to rigid gender roles being upheld? Many, though not all, social biases are still caused by cognitive biases. See, for instance, this male privilege checklist – there’s a lot of non-cognitive stuff in there, but also a lot of stuff that seems to be a result of the availability heuristic and a form of status quo bias. Gender roles are also a good example of how biases can be self-fulfilling, with a popular conception of individuals of a certain gender being of a certain kind leading to just that. (Also, most of those points in the linked article seem very valid, so the accusation of feminism and such being “just a selfish grab for more attention and deference than they deserve” seems unwarranted.)

    I feel that this community and the feminist communities could have a lot to offer each other – since it’s obvious the people here don’t know much about the topic, could the editors consider inviting some feminist writers over to make guest posts?

  • http://profile.typekey.com/halfinney/ Hal Finney

    There is a classical argument that employers who overcome discrimination bias will have an economic advantage, as they can employ members of repressed groups for lower wages. Some have claimed that this means that there would be no discrimination bias in employment. However it seems more likely that this bias, like others, cannot be overcome just by intending to do so, and that we would therefore see a range of degrees of discrimination bias among employers.

  • http://brokensymmetry.typepad.com Michael F. Martin

    Whatever evidence would tend to falsify our hypotheses about our own bias or lack of bias.

    J.R.R. Tolkein told his son that when he went to confession, he should confess to the priest whom he despised.

  • http://grendelkhan.livejournal.com grendelkhan

    For someone so interested in overcoming his own biases, it seems never to have occurred to Robin that the word “minority” means something concrete, that minority groups aren’t simply self-defined, or that putting the word “repress” in scare quotes implies a conclusion already reached. When his bias and privilege were pointed out, he got defensive. How depressingly unsurprising.

    Yes, treating women like mushy-headed creatures from beyond the tenth planet betrays bias on your part, Robin. The problem isn’t other people. The problem is you. No, a series of guest posts aren’t needed. There are places to learn about this stuff, and it’s the least you can do to try and learn a little rather than trying to shield yourself from the apparently terrifying prospect of admitting you were ignorant. There’s nothing wrong with being ignorant; it’s the easiest thing in the world to fix, if you actually want to.

  • http://hanson.gmu.edu Robin Hanson

    Grendel, the quotes are intended flag a neutral questioning uncertainty regarding whether the labels actually apply, not to claim that they do not apply.

  • Unnamed

    [W]e might view the continued lobbying of self-identified repressed minorities as just a selfish grab for more attention and deference than they deserve.

    Robin, why is this the only alternative you listed to the hypothesis that minorities are actually correct about our biases? I would’ve thought that you have enough experience with disagreements to avoid jumping to such a hostile accusation. Why couldn’t they just be mistaken, and perhaps a bit behind the times?

    Take the case of African-Americans. If African-Americans do tend to overestimate the extent and importance of social discrimination against blacks, then I think that’s pretty understandable given the widespread and severe discrimination against blacks in the not-too-distant past, along with the fact that, besides reducing discrimination, the Civil Rights movement also made the remaining discrimination much less visible. That means it can be hard to tell how much discrimination remains. Since African-Americans learned habits of thought when discrimination against them was an enormous problem (and the struggle against it was very important), and since there continue to be fairly large disparities between blacks and whites in America, it wouldn’t be surprising if many blacks would continue to see discrimination as a main reason for their troubles even if it has actually been reduced to a relatively minor problem. And of course some of the standard biases (like confirmation bias in interpreting ambiguous actions as cases of discrimination) could contribute to this mistake.

    There’s no need for group-selfishness to enter into the picture. There is also a pretty direct argument against group-selfishness playing that big of a role: plenty of middle class white males have similar pro-diversity views.

  • http://grendelkhan.livejournal.com grendelkhan

    Robin, how does that differ from scare-quoting? (I’ve have a few questions about the “Holocaust”. And I’m looking to learn more about the “prejudice” that some people talk about.)

    If you’re wondering why some of the reactions you’ve gotten range from bewildered to flat-out pissed, it’s because the things that you’re wondering about the existence of are as obvious as a punch in the mouth to the people they affect. Jonathan Schwarz has spoken more eloquently on this point than I can, but it gets to the heart of the matter; in your position, you don’t see these things unless you make an effort to educate yourself. That doesn’t mean they don’t exist.

  • http://hanson.gmu.edu Robin Hanson

    Unnamed, fair enough, I’ve added that possibility to the text.

    Kaj, yes such groups also often try to overcome their perceived lack of self-esteem, but most of their efforts are elsewhere.

  • http://apperceptual.wordpress.com/ Peter Turney

    Peter, probability theory and decision theory are pinned by a number of optimality and uniqueness proofs, in theory. So who gets to use the best-achievable rational beliefs and strategies, and who has to use suboptimal ones for the sake of “diversity”?

    Eliezer, aren’t you contradicting what you said in The Design Space of Minds-In-General and No Universally Compelling Arguments? Are you saying that Bipping AIs, Gloopy AIs, and Freepy AIs cannot all be rational? There is only one point in the design space that is rational? This seems to be a rather narrow view to me.

    I think we can agree that perfect rationality is not possible. Since any mind is finite, at best we have bounded rationality. Different minds may have different bounds, and hence different beliefs and different strategies. You could take a “God’s eye” point of view, and say that one of those bounded minds comes closest to the “truth”, which can only be known to an unbounded mind, but the “God’s eye” point of view is unattainable in reality.

    Briefly, only God could know what are “the best-achievable rational beliefs and strategies” (and I’m an atheist).

  • http://apperceptual.wordpress.com/ Peter Turney

    Peter, I don’t see how I assumed anything singular about rationality.

    Robin, your use of the term “we” in your post above seems, to me, to assume a lot about “us”. I think “we” are more diverse than you suppose.

  • Anonymous

    Peter, if we didn’t have bounded rationality, is there any reason to think that our utility functions would eventually converge? If they did, would that be an indication that ‘perfect rationality’ had been achieved?

  • Feminist

    The last century “or so”? I think it’s been longer.

    http://mailer.fsu.edu/~cupchurc/worldhist/readinglist.html

    And wow, Oxford University Press! Publishes scholarship on…..”race”!!! Whoda thunk!

    http://www.us.oup.com/us/catalog/general/subject/Sociology/RaceEthnicity/?view=usa
    126 titles in “Race and Ethnicity”

    19 titles in “Race and American Culture”
    http://130.88.203.72/us/catalog/general/series/RaceandAmericanCulture/?view=usa

    A doctoral reading list in feminist theory!
    http://www.sfu.ca/womens-studies/Programs/phdreading.html

    And a quick “race and ethnicity” search at Oxford university reveals that yeah, they’re doing some thinking and teaching and writing about it too:

    http://googlesearch.oucs.ox.ac.uk/search?submit.y=7&search_type=pages&searchreferer_id=1&client=oxford&output=xml_no_dtd&access=p&proxystylesheet=admin-dev&submit.x=15&ie=UTF-8&site=default-minus&q=race%20and%20ethnicity

    I wonder what the academics at Oxford would think of the level of thought and discourse on race here at Coming Over Biassed!

    Signed: Feminist

  • http://apperceptual.wordpress.com/ Peter Turney

    Peter, if we didn’t have bounded rationality, is there any reason to think that our utility functions would eventually converge? If they did, would that be an indication that ‘perfect rationality’ had been achieved?

    Anonymous, as far as I know, none of the optimality and uniqueness proofs to which Eliezer alludes deal with bounded rationality. They all take the God’s eye point of view. I do not know of any “convergence” proof for bounded rationality.

  • rfriel

    I think I know what you mean when you contrast “overcoming our own biases” with “overcoming other people’s biases,” but that dichotomy sounds really arrogant to my ears. Even if the intent is good, that framing sounds a bit like “we OB readers are bettering ourselves, while other people just sit around whining!” Which is not a statement I want to stand behind.

    I think a better sound-byte version of the issue is that OB deals with what bias is, while many other people deal with where bias is. OB tries to catalogue and study the ways that people can be biased. The hope is that, armed with this knowledge, readers can determine for themselves which ideas and institutions around them are products of bias. On the other hand, many other people are in the business of actually discussing which ideas and institutions are biased. A feminist blog, for example, will probably devote a lot of posts to talking about whether certain ideas, laws, etc. are biased against women. This relies on a notion of what it would mean to be “biased against women”–and that, in turn, on what it means to be “biased.” This is what OB investigates.

    I’m not sure how much I buy the argument that we are dealing with two definitions of the word “bias” here. Of course, bias against minorities can take several forms. It can be positive (in the philosophical sense), as when someone thinks that members of a minority are worse in some way (when they aren’t). Or it can be normative, as when someone thinks that members of a minority should not receive good things because it is morally bad for them to be happier. (Or a variant of the latter, where someone acts out of emotional distaste for a minority, but where the emotion is not a moral emotion.) The positive kind of bias is definitely the sort of thing OB talks about. The other kinds are not so clear, because they involve moral claims beyond “we should believe the truth.” But now that Eliezer is getting into ethics, OB might be on the road to discussing the idea of normative bias. That would be interesting.

  • Unknown

    Since this blog is about overcoming our own biases, I wonder why no one has ever set down a post or a comment saying, “What I said previously was wrong; I was subject to such and such a bias.” It’s true that people have modified their positions, but I don’t remember anyone here accusing himself of having been subject to a bias, ever.

  • Mir

    I would like to hear Eliezer’s opinion on the Duplicate Paradoxon. It would be nice to hear argumentation beyond ‘both duplicates are REALLY me’, since in my mind it does not solve the Duplicate Paradoxon.
    As Ban Best says:

    “Specifically, it is a very serious problem for me to imagine two Ben Bests standing side-by-side, one facing North, the other facing South. After duplication, would I see the view to the North or the view to the South? I cannot answer this question with the counter-question “which ‘I’?” There can only be one I to me, I can only look in one direction at once. If the duplicate facing North is destroyed, I will be either dead or alive. I cannot be in two minds, two bodies and two locations at the same time. I cannot speak of one “I” facing North and one “I” facing South. That is an objective description of “I” — and is utterly inappropriate. The experience of being me is unitary, solitary and subjective.”

    I am rather new in the blog, and I’m sorry if the discussion has already taken place.