Birth of a Stereotype
I imagine that many ‘enlightened’ people disbelieve in stereotypes, thinking them useless, and something below them; only the ignorant masses rely on stereotypical thinking and generalisations. Stereotypes are improbable, and making such generalisations from anecdotal evidence at best is faulty. I used to think like that—and I probably still do—nevertheless, someone raised an argument that stereotypes aren’t useless; they exist for a reason. There was a reason why the stereotype was born. As such, you would expect that a stereotype conveys at least some information and is better than nothing.
I think a few factors contribute to the formation of a stereotype. I will take a minute to explain a stereotype. A stereotype is a relation of the form X ⇒ Y. It maps a class of people/individuals/what have you to a property X. For example, people who wear glasses are smart. Occasionally, some individuals may conceive the relation as Y ⇔ X. E.g. Smart people wear glasses. I suspect this is due to reasons unrelated to the stereotype (e.g. inability to distinguish between ’=>’ and ’<=>’). I hope this is not common among the general population—the average human can’t be that irrational, right? I shall give a charitable interpretation of the masses, and discuss only the relation ‘X ⇒ Y’. I would stick to two particularly conspicuous stereotypes and try to hypothesise how they originated. For one, anecdotal evidence, combined with the availability heuristic make people overrate the occurrence of certain relations. I shall pick the “people who wear glasses are smart” example: Some smart people use glasses. Due to the availability of smart people who use glasses the relation ‘glasses’ ⇒ ‘smart’ gets reinforced. However, this hides an implicit assumption; that the average IQ/intelligence (perceived or otherwise)/‘smartness’ of people who wear glasses. Is this true? Normally, my common sense tells me that this is false, however there is correlation between height and intelligence: NCBI, Wikipedia, medicalxpress. I have not yet read any of them in sufficient detail—or any detail for that matter—as at the time of writing this, but I have bookmarked the links for future study. I cannot ascertain whether glasses actually correlate to intelligence and the effort required for that, is more than I’m willing to commit to an article I am writing out of boredom. I would appreciate if the more medically knowledgeable readers fixed my ignorance.
That said, I shall offer a hypothesis for the scenario in which there is no correlation between glasses wearing and intelligence. I would try not to violate Occam’s razor, but there is no way I’m going to go through the rigours of Solomonoff induction—I’ll probably have to learn that in detail first—for this. Nor am I going to investigate the Kolmogorov complexity (something I more or less understand) either. I have reason to believe, that events which are strange or different from the norm leave a stronger mark in our memory. Events that are emotionally charged are often associated with the emotion and are retained more in episodic memory Psychology Today. On first encountering the glasses users that were smart the fact that they wore glasses might have left a deep impression on the people who encountered them and may have been associated with their perceived intelligence. The brain is also an obsessive pattern matching engine drawing connections even when they are not there: Scientific American, Wikipedia Psych Central, and so, the relation ‘wears glasses’ ⇒ ‘smart’ may have started to take root. Furthermore, categorising smart people and glasses together is easy. Kahneman 2011 “Thinking Fast and Slow”[1] brought forth the theory of ‘cognitive ease’; people’s brains work along the line of least strain or greatest cognitive ease. If there are two decisions/decision making procedures, we tend to go with the one that is associated with more cognitive ease. I suggest that this (meta) heuristic of maximising cognitive ease, would make us more readily associate two characteristics with each other when there may be no logical reason for such an association. This may have led to the association of ‘glasses’ and ‘intelligence’.
The above is the charitable hypothesis. I decline—at this juncture—to mention the less charitable one.
After the formation of the stereotype, it was reinforced in a feedback loop. The presence of the stereotype in the media reinforced its availability in our brains, and primed us to notice it when it does occur. Furthermore confirmation (positive) bias, may make us selectively notice the manifestations of the stereotype, and ignore the many cases when the stereotype was wrong. I do remember feeling disappointed when a kid I met in primary school was nowhere near as smart as I had expected. I wonder if that was when I started to dislike stereotypes. I wonder if there might be some groupthink involved in reinforcing the stereotype? I certainly suspect it (probability greater than 0.75 based solely on anecdotal evidence) in my society (Nigeria) but does such thinking abound in the Western World as well? I would appreciate feedback from my users on this issue.
I suspect that the representativeness heuristic combined with base rate neglect, cause people to overrate the proportion of a certain group of people who fit a certain characteristic. In the glasses example, people neglect the rate of people with glasses compared to the general population. Combine this with glasses taking to be representative of smart people and we have them overrating the proportion of smart people who wear glasses.
Borrowing again from Kahneman 2011, I posit that stereotypes from an easy to implement and convenient heuristic (cognitive ease), and such are applied widely. Rather than dealing with each s in X as a separate individual that needs to be considered separately and handled as such, we pull up any stereotypes we know about X, and by deductive reasoning apply it to s. We can now deal with s on a better footing than when we started. Such reasoning is not in fact wrong, and would in fact be advisable—if the stereotypes were actually accurate. If for example 80% of glasses wearers had IQs north of 100, then relying on the stereotype is better than going in with zero information. Alas, the stereotypes seem to be unfounded. The perceived utility of stereotypes may contribute to the feedback loop, and make them much harder to kill.
The non-charitable hypothesis.
One stereotype that frustrated me—not just due to its inanity—but due to its sheer intractability; try as I could, I could not decipher the origin of the stereotype: “blondes are dumb”. I eventually realised/was made to realise the (a plausible) origin of that stereotype. I shall describe the conversation below for your benefit referring to myself as ‘DG’ and my conversation partner as ‘Alpha’.
DG: “It was a rant; I was venting my frustration about Nigeria. People do not read rants to gain an unbiased opinion. I even put ‘warning rant ahead in the post’.”
Alpha: “I know, but you are contributing to the stereotype they have about Nigeria. Anyone that reads this would now start thinking that all Nigerians are like this.”
DG: “Stereotypes are so stupid. I mean they’re completely baseless and unfounded. Look at the ‘blondes are dumb’ stereotype.”
Alpha: “Stereotypes exist for a reason!”
DG: “What could possibly be the cause for the ‘blondes are dumb’ stereotype? Like how? How could it possibly exist?”
Alpha: “….”
DG: ….”
Alpha: “Well you know blondes are generally considered more attractive…” DG: “That’s it! Blondes are aesthetically more pleasant, and as such are more likely to work in jobs that are less intellectual, and considered the domain of bimbos.”
Alpha: “Blondes would be more likely to be hosts, waitresses…”
DG: “And strippers and porn stars. People are jealous; others can’t have everything, so they tend to bring them down and pin other qualities on them to make up for any advantages they have dumb blondes compensate for their beauty with lowered intelligence.”
I am not Yudkowsky, and so I would not proffer an evolutionary psychology hypothesis for why blondes are more attractive among Caucasians (plus I don’t know shit about evolutionary biology, much less evolutionary psychology, and so I’ll be talking out of my ass—which would be fine if I were trying to guess the teacher’s password—however, seeing as that is not my aim here, I’ll refrain). The fact is that blondes are considered more attractive among Caucasians (relative to other Caucasians and—I suspect—to other races as well). People do not like that the Universe isn’t fair; they prefer to believe in a just world. I mean wouldn’t it be more convenient if the world were just and fair; only your hardwork mattered, no one was unfairly gifted in the genetic lottery? I suspect that the just world hypothesis is why there is a certain class of people (significant and maybe the majority, but I lack the relevant statistics) who try to downplay IQ and argue that it is irrelevant, and not a true measure of intelligence. Einstein’s ‘quote’ (alleged if the “ were not clear enough) about how “everyone is a genius but if you judge a fish by its ability to climb a tree it would live its whole life believing it’s stupid”, is often quoted among that class as a holy maxim, and used to defend the convenient status quo of equality in the genetic lottery. I find the entire quote and mindset bogus, but I’ve digressed enough already, and shall now steer this article back on track. “Blondes can’t both be beautiful and have the same smarts as everyone else”: this jealousy helped reinforce the stereotype of the blonde bimbo. The fact that the proportion of blondes among bimbo professions may have in fact been higher, would have helped legitimise the stereotype. People see what they want to see, and as such they would ignore the proportion of blondes among the intellectual elite as well. I have learned not to underestimate the human mind’s capability of cognitive dissonance—it was 17 years before I finally chose Science myself after all.
Unlike the ‘glasses’ ⇒ ‘smart’ stereotype, I suspect the ‘blonde’ ⇒ ‘stupid’ stereotype is motivated almost entirely by the jealousy of the masses. Even as education may eradicate the former stereotype, I suspect the latter would last for a while longer due to the pervasiveness of the “just world hypothesis” and the desire for fairness of the masses.
References
[1] Kahneman Daniel “Thinking Fast and Slow” 2011 Ch 5. pp 65-76. Random House of Canada Limited.
I think even besides the just world hypothesis, there’s a statistical thing at work in the blondes example, which has a name that I am totally unable to find right now.
If you imagine (just as a simple example) that most people you interact with regularly are people providing services to you commercially, who therefore have jobs; and that further, getting a job requires one to be a strong applicant on one of various axes (e.g. a more attractive person can get a job with less intelligence, and vice versa), then you will find those things inversely correlated (spuriously) in the population of job-havers, due to a selection effect with a name I can’t remember.
EDIT: https://en.wikipedia.org/wiki/Berkson%27s_paradox
I’ve bookmarked the link, and I’ll read it at my leisure. I may update the article to include the information from the link at that point in time.
As a reader, this gives me pause. If you didn’t have any more compelling reason to write than that, you shouldn’t expect anyone to have a compelling reason to read. Maybe give yourself more credit: you weren’t merely bored, the fact that you may have felt bored is incidental to the fact that you had something to say!
Solomonoff induction is uncomputable; it’s great to be aware that the theoretical foundations exist, but it’s also important to be aware of what the theoretical foundations are and aren’t good for. (Imagine saying “there’s no way I’m going to go through the rigors of predicting the future state of all air molecules here given their current state” when what you actually want is a thermometer.)
But “On first encountering _X_ that were _Y_, the fact that they were _X_ might have left a deep impression on the people who encountered them” works for any _X_ and _Y_; it can’t explain why the stereotype links glasses and intelligence in particular.
A more specific hypothesis: people are more likely to need glasses while reading, and reading is mentally associated with intelligence because it is in fact the case that P(likes to read | intelligent) > P(likes to read | not intelligent).
Don’t leave your readers in suspense like that; it’s cruel! (Also, what makes a hypothesis “charitable”, exactly?)
Are they?
Eliezer doesn’t have a magic license authorizing him in particular to tell just-so stories: if he can do it, anyone can! (Some argue that we shouldn’t, but I don’t think I agree.)
It’s possible that the “blondes are dumb” stereotype arose first from random noise and then was reinforced in other ways. For example, I have seen blondes who were actually not very intelligent deliberately acting as stupid as they could. They were purposely playing to the stereotype and apparently believed that it looked cute. But other people who did not notice that they were doing it deliberately might have had the stereotype reinforced by that behavior.
It would be better to think of it as X correlates with Y, or X is evidence for Y. And unlike your ⇒ relation, which you never adequately specified, these two relations are symmetric.
Correlations are symmetric, but is evidence for may not be (depending on how you interpret the phrase): P(A|B) ≠ P(B|A) (unless P(A) == P(B)).
However, the likelihood ratio (P(B|A)/P(B|~A)), a.k.a., the quantity you actually care about when updating on new evidence, is symmetric.
A kind of causation. X implies Y. X may be strong evidence for Y, but Y may be extremely weak evidence for X, so the two relations are not really symmetric.
You seem to be confusing causation and “evidence for” implication. DON’T. Wet streets are evidence for rain, but when streets do not cause rain.
Bad eyesight gene may heighten IQ, or the genes are intercorrelated somehow. So the arrow can go both ways. I would think that humans think in clusters rather that A → B.
Re the correlation between nearsightedness and IQ. Someone on Scott’s blog post re Ashkenazi IQ mentioned that the blindness gene has a correlation with High-IQ.
https://www.hindawi.com/journals/joph/2015/271746/
No real politics behind the correlation (which is why OP used it as an example) so my prior would be high; though I would have to check the trail to see if the findings stand up.
Also, we have a good evolutionary/social principle of a costly signal. Bad eyesight and still survive natural selection.
I thought the standard explanation for why smart people wear glasses is that smart people are likely to spend more time reading books, which isn’t that good for one’s eyes. Conversely, if one is nearsighted one will have an easier time reading books than playing sports, and so is likely to at least become more knowledgeable.
I do not want theories for how eyesight and IQ may be correlated—I want statistics showing bad eyesight and IQ are significantly positively correlated. The theories can come after. Do we have statistics that show that?
Dude. https://scholar.google.com/scholar?q=myopia+intelligence+OR+IQ https://en.wikipedia.org/wiki/Near-sightedness#Correlations