The Halo Effect

[Eliezer Yudkowsky]

The affect heuristic is how an overall feeling of goodness or badness contributes to many other judgments, whether it’s logical or not, whether you’re aware of it or not. Subjects told about the benefits of nuclear power are likely to rate it as having fewer risks; stock analysts rating unfamiliar stocks judge them as generally good or generally bad—low risk and high returns, or high risk and low returns—in defiance of ordinary economic theory, which says that risk and return should correlate positively.

The halo effect is the manifestation of the affect heuristic in social psychology. Robert Cialdini summarizes:¹

Research has shown that we automatically assign to good-looking individuals such favorable traits as talent, kindness, honesty, and intelligence (for a review of this evidence, see Eagly, Ashmore, Makhijani, and Longo, 1991). Furthermore, we make these judgments without being aware that physical attractiveness plays a role in the process. Some consequences of this unconscious assumption that “good-looking equals good” scare me. For example, a study of the 1974 Canadian federal elections found that attractive candidates received more than two and a half times as many votes as unattractive candidates (Efran and Patterson, 1976). Despite such evidence of favoritism toward handsome politicians, follow-up research demonstrated that voters did not realize their bias. In fact, 73 percent of Canadian voters surveyed denied in the strongest possible terms that their votes had been influenced by physical appearance; only 14 percent even allowed for the possibility of such influence (Efran and Patterson, 1976). Voters can deny the impact of attractiveness on electability all they want, but evidence has continued to confirm its troubling presence (Budesheim and DePaola, 1994).

A similar effect has been found in hiring situations. In one study, good grooming of applicants in a simulated employment interview accounted for more favorable hiring decisions than did job qualifications—this, even though the interviewers claimed that appearance played a small role in their choices (Mack and Rainey, 1990). The advantage given to attractive workers extends past hiring day to payday. Economists examining US and Canadian samples have found that attractive individuals get paid an average of 12–14 percent more than their unattractive coworkers (Hamermesh and Biddle, 1994).

Equally unsettling research indicates that our judicial process is similarly susceptible to the influences of body dimensions and bone structure. It now appears that good-looking people are likely to receive highly favorable treatment in the legal system (see Castellow, Wuensch, and Moore, 1991; and Downs and Lyons, 1990, for reviews). For example, in a Pennsylvania study (Stewart, 1980), researchers rated the physical attractiveness of 74 separate male defendants at the start of their criminal trials. When, much later, the researchers checked court records for the results of these cases, they found that the handsome men had received significantly lighter sentences. In fact, attractive defendants were twice as likely to avoid jail as unattractive defendants. In another study—this one on the damages awarded in a staged negligence trial—a defendant who was better looking than his victim was assessed an average amount of $5,623; but when the victim was the more attractive of the two, the average compensation was $10,051. What’s more, both male and female jurors exhibited the attractiveness-based favoritism (Kulka and Kessler, 1978).

Other experiments have demonstrated that attractive people are more likely to obtain help when in need (Benson, Karabenic, and Lerner, 1976) and are more persuasive in changing the opinions of an audience (Chaiken, 1979) . . .

The influence of attractiveness on ratings of intelligence, honesty, or kindness is a clear example of bias—especially when you judge these other qualities based on fixed text—because we wouldn’t expect judgments of honesty and attractiveness to conflate for any legitimate reason. On the other hand, how much of my perceived intelligence is due to my honesty? How much of my perceived honesty is due to my intelligence? Finding the truth, and saying the truth, are not as widely separated in nature as looking pretty and looking smart . . .

But these studies on the halo effect of attractiveness should make us suspicious that there may be a similar halo effect for kindness, or intelligence. Let’s say that you know someone who not only seems very intelligent, but also honest, altruistic, kindly, and serene. You should be suspicious that some of these perceived characteristics are influencing your perception of the others. Maybe the person is genuinely intelligent, honest, and altruistic, but not all that kindly or serene. You should be suspicious if the people you know seem to separate too cleanly into devils and angels.

And—I know you don’t think you have to do it, but maybe you should—be just a little more skeptical of the more attractive political candidates.

¹Robert B. Cialdini, Influence: Science and Practice (Boston: Allyn & Bacon, 2001).

[James_D._Miller]

Perhaps firms should conduct “blind” interviews of potential employees in which the potential employee is interviewed while behind a screen.

[Different_Jeff2]

I believe it’s been shown that intelligence does correlate positively with height. Is it possible that it correlates with symmetry or other physical traits that make people attractive? I have no doubt that a bias is at work here, but is it possible that bias has some foundation in fact?

[avichapman]

If it’s true that intelligence correlates with height, I wonder if it is because childhood nutrition affects height? Perhaps childhood nutrition also affects brain development. Interesting.

[Girl B]

Whoa…You could likely correlate this to a string of societal factors going back through time as well such as: attractive child receives more attention & assistance in areas of difficulty, which leads to understanding & an ability to pull ahead, snowballing over time (& combining with further effects of attraction bias) throughout their education. Until eventually they’re easily able to retain more knowledge, therefore becoming more intelligent than the less attractive classmate.

[anon_reader]

Jeff may be on too something here. I remember reading that symmetrical facial features is somewhat dependent on the level of oxygenation the fetus gets in the later stages of pregnacy. Of course I may have mixed up my facts but if this is true it seems to provide something of an explaination for the possible corrolation Jeff talks about.

[Eliezer Yudkowsky]

Not if you’re basing your judgment of intelligence on a description that’s held constant with just the pictures swapped.

[frontier64]

Swapping pictures leading to predicting higher intelligence is the point Jeff makes. A symmetrical face is a piece of evidence that positively correlates with higher intelligence just as a description of someone’s accomplishments is evidence of intelligence. The description is much better evidence, but the attractiveness remains somewhat important.

[lesswronguser123]

The description is much better evidence, but the attractiveness remains somewhat important.

I would like to disagree on this one, if all of the major accomplishments of the candidate is written on that description then the attractiveness doesn’t really matter as an evidence for intelligence, it’s already been taken into account in that description.

[Elliot]

For example, a study of the 1974 Canadian federal elections found that attractive candidates received more than two and a half times as many votes as unattractive candidates (Efran & Patterson, 1976).

This does not mean what it sounds like it means. Well, it could, but it doesn’t have to. Specifically, this result is consistent with the voters’ claims that they don’t vote for candidates because of physical attractiveness.

This is a case of “correlation does not imply causation”. Just because good looks were correlated with votes doesn’t mean they caused votes. There could be another effect causing both.

Such effects are easy to imagine. For example, perhaps people with good looks receive more encouragement in school and from their parents, and thus turn out smarter. Then they could have received all those votes because they were genuinely better candidates. This particular possibility may have been looked for and ruled out, but there are infinitely many others.

The important thing is that you can’t find the truth purely by finding correlations. What you need are explanations. Specifically, there needs to be a detailed explanation of how being more attractive causes favoritism (and also of what causes people to be blind to their own favoritism). And when we have that explanation, then we can compare it to rival theories that explain the observed data, including the correlation, in other ways.

[Roland2]

I wonder if this also depends a bit on context:

I remember when studying mathematics, that almost all professors and most of the good students where weird in several ways: the way they dressed, the way they spoke, the way they behaved etc… It seemed that the greater the genius the weirder his behavior.

Now I wonder if the bias in these academic settings could be the other way round: if you look like a normal person you will stand out and everyone will think: this is probably not one of the brighter guys. On the other hand if your last haircut was 2 years ago and you are always wearing the same jacket everyone will say: wow this guy must be one of the geniuses here.

What do you think?

[pnrjulius]

I think it’s more this effect: The people doing the hiring WANT you to be pretty, but they also want you to be brilliant.

If you are pretty enough, you can get in without being too brilliant; or if you are brilliant enough, you can get in without being too pretty.

Ideally, you would be both (i.e. you would be Richard Feynman); but you’ve got to have one or the other, or nobody will hire you.

Hence, our sample will consist of A: Pretty people who aren’t that brilliant P(P&~B) B: Brilliant people who are not pretty P(~P&B) C: Pretty people who are also brilliant P(P&B)

If we presume merely that prettiness isn’t strongly correlated with brilliance and both are relatively rare, then we will find exactly what we see: The brilliant ones who get hired are also mostly not pretty.

[Caledonian2]

Paying attention to the description is probably less useful than looking at the features—the features are generally harder to misrepresent, and at the very least haven’t passed through another person’s mind before they got to you.

[Adirian]

As Different says, but in regard to connections between perceived intelligence, and perceived honesty, to pick two particularly useful examples—the usefulness of either quality, in regard to your interactions with the individual, are dependent upon both. I/​e, it isn’t a great idea to trust what a not-too-bright individual tells you, even if he or she has never told a lie in their life, for the simple reason that they may not have the faculty to evaluate their statements. And the reverse might be true—particularly bright individuals may not be good candidates for trust, particularly on important issues, because they have great faculty for making value judgments about when it is most profitable to lie. Individuals at either extremes may not be good candidates for trust.

I had a long discussion with my brother on precisely the issue of attractiveness—in regard to banks. Banks are great examples because they spend immense quantities of money making themselves look respectable. So—would you trust your money to a bank that was going to spend some of it making itself look good? Or would you trust your money to a bank that doesn’t care how it looks? A bank that cares about its reputation enough to spend massive amounts of money maintaining it isn’t going to sacrifice that by stealing your relatively small sum—it is the better choice, presuming on the rationality of the bank. An individual who goes to great lengths to APPEAR competent is going to try to BE competent—someone who doesn’t care whether or not they appear competent do not care whether you think they are competent, and hence, may not make an effort to be more competent as relates to you and your business. The better-groomed candidate, other things being equal, is the better choice, presuming upon that individual’s rationality. (Particularly since the issue is reinforcing—clients and customers, after all, are making the same judgments.)

An individual who makes an effort to appear more attractive likely has a reason for doing so—they may care what other people think of them (which suggests they’ll be nicer, when somebody is watching, at least), they may want to appear more competent (which suggests they’ll be better at other things they do, presuming they follow similar levels of investment), and, presuming they DO have rational reasons for taking care of their appearance, they may simply be smarter than your average person. Naturally good looking people may be getting the benefits of biases we develop based on those who acquire appearances by effort—and may get points for honesty, as well, because they aren’t attempting to “lie” about their appearance. (Which would be interesting, because it would mean naturally good-looking individuals gain more benefits than those who provided much of the bias incentive to begin with.)

Just some thoughts.

[Girl B]

This goes to the saying “dress for the job you want—not the one you have” which many seem to misunderstand as a show for others when in fact the origin is based on the hypothesis that we’re constantly influenced by what we do & can actually project ourselves into a better situation by believing it to be possible.

The bank analogy brings to mind Apple—the brand: your privacy the most important thing to them. Therefore, although we have no way of knowing if thats actually true (as we’re all VERY aware of PR-based-branding ploys these days) we still buy into it because they’re much more likely to avoid tarnishing that brand by faking privacy concern than another brand who doesn’t pretend to care.

[TGGP4]

This seems relevant, though perhaps the causality is uncertain.

[Lake]

Eliezer—wasn’t Jeff’s comment intended to suggest, not that there isn’t a bias, but that the bias may be adaptive? Offhand I can’t imagine quite what edge it might supply, but perhaps some story could be told.

[pnrjulius]

Well, it encourages you to mate with people who are both pretty and intelligent, which seems like it would be good for your genes.

[gutzperson]

@roland It helps to have lots of grease marks on your jacket. No dry cleaner or washing machine should ever touch these. The mathematical mastermind is usually dressed in brown and green cords with lots of spots, while an artistic genius looks either like a colourful parrot or a dark-suited banker. On the other hand, bankers like to wear their designer-made bicycle helmets all day long. If you are a female long legged blonde with blue eyes you are more likely to get the job if your employer is a short asymmetric male. The halo effect promising a good shag and many little mini blondes. A long legged blonde tends to employ a short legged dark haired obese female and would like to marry her asymmetric male boss. Boss is going to replace the blonde model if she is older than thirty. All of them are honest and kind as well as greedy, only interested in the best possible outcome for all parties. No irony intended!

[Sam2]

Eliezer

Just for interest—Joel Wapnick, a music education scholar at McGill University (and also, I discover from Googling him just now, an international Scrabble champ) has shown that people rate the quality of musical performances from more attractive players more highly than those from less attractive players. No surprise there. However, the effect persists even in an auditory-only condition, i.e. when the raters cannot see the performers. Wapnick has replicated the finding in different situations over a series of papers.

Sam

[Eric3]

Do you have a reference for “stock analysts rating unfamiliar stocks judge them as generally good or generally bad—low risk and high returns, or high risk and low returns—in defiance of ordinary economic theory, which says that risk and return should correlate positively”?

[Wuschel Schulz]

I am also still looking for a reference on that one...

[ME3]

I am also struck by the correlation-vs.-causation issue in the canadian voters study. Moreover, how do we know that the attractiveness rating isn’t actually a reflection of the qualities the voters claim to be looking for? I.e. a more confident, intelligent, eloquent candidate would probably appear more attractive than one who isn’t, all other things being equal.

[Bob3]

I am sympathetic to the counter-comments but need to point out that most of us (those who are not perfect 10s) want to believe that there is something else underlying the evidence that looks matter. Who wants to accept that their talents are dominated by appearance?

[pnrjulius]

On the other hand, if you’ve done reasonably well for yourself, it probably means you are good-looking, which I guess is a good thing?

[Doug_S.]

I’ve noticed this in myself… I find it hard to think poorly of attractive females, even when the evidence indicates that I ought to.

[pnrjulius]

I do the same for both males and females (and I’m bisexual, which may be part of it).

And now the question becomes: What can we do about this? Should we train ourselves to automatically distrust good-looking people (that seems… problematic at best)?

[Nate2]

it would seem that a good strategy would be to always vote for the ugliest candidate and hire the ugliest job candidate—on the presumption that their (difficult to judge) talents must be tremendous in order for them to make it this far in the process despite their looks.

[Adirian]

Nate—that strategy can only work insofar as other groups aren’t utilizing it, and to that extent, you will be punished through those groups hesitating to employ you in their capacities as clients and customers.

[sasserstyl]

I’d be interested to know the “curve” drawn when physical attractiveness is plotted against level of bias—whether there is a linear relationship, or whether the effect (indeed affect) tails off (or even turns negative?!) at some point. i.e. whether things just keep getting “better” the more attractive you are.

I think that the quoted Cialdini text blurs the distinction between physical attractiveness and grooming.

I can well imagine that people react positively to a well-groomed person because of the values such care over appearance indirectly demonstrates (social intelligence/​wealth/​hygeine etc.).

A halo effect based on pure physical attractiveness probably has more to do with a net positive bias resulting from a complex set of sexual dynamics created by different combinations of male/​female and attractive/​unattractive.

[Ian3]

http://​​www.economist.com/​​displaystory.cfm?story_id=10311266

″...the accumulating evidence suggests that physical characteristics do give clues about intelligence, that such clues are picked up by other people, and that these clues are also associated with beauty.”

[jedermann]

“You should be suspicious if the people you know seem to separate too cleanly into devils and angels.”

The halo effect perhaps provides a better explanation for the apparent insistence on Hitler’s weird sex life:

http://​​www.slate.com/​​id/​​2205359/​​pagenum/​​all/​​

(from post on NYT ideas blog)

[pnrjulius]

Also the belief that he has only one testicle, which appears to be completely made up by British sailors.

[JJ10DMAN]

The opposite is also true: a “negative halo effect” can be easily observed, wherein “bad” traits are also similarly grouped and feed on each other.

An interesting part of halo effects is that people seem to understand them on an instinctual level—not enough to get rid of them, but enough to exploit them...

I’ve drawn an extremely strong correlation in a particular online game between having a marijuana reference in one’s handle and being bad at the game; being bad is not strongly linked with marijuana references, but that’s only because they’re in the extreme minority of the population; if you’re sporting a “420”, you’re almost definitely underperforming. I’ve never bothered forming a hypothesis as to why this is, but it is.

So one day I decide to try a little experiment—for funsies, nothing rigorously scientific, just a “see what happens” thing—and predict aloud that an ally with a name referencing marijuana would perform poorly compared to the other players in the game. I turned out to be right (he was even worse than expected), but the interesting part was his response:

He implied my prediction was wrong, evidenced by that he was writing his college thesis on the effects of THC on the body.

The only way this statement makes sense is if we trace it through an expectation on his part that he can rebut the argument using the halo effect. “I am extremely accomplished academically,” I could almost read on the screen, “Therefore, I am not a poor performer in this online video game.”

[A1987dM]

I suspect the intended implication was more like “I know what the effects of marijuana on humans are, and they don’t include making people worse at MMORPGs, so just because I smoke marijuana doesn’t mean you can conclude I must be bad at this game”.

[pnrjulius]

Right! I mean, who would expect video game performance to correlate positively with academic performance? It could well be the other way around: Perhaps people who play video games for hours don’t do their homework!

[fr00t]

I for one would openly admit that a political candidates physical appearance would factor into my decision on whether or not to vote for them. For reasons such as, but not limited to: because I’d rather look at JFK giving the state of the union address than Nixon.

I’ve arrived at the conclusion that there is a decent correlation between attractiveness and intelligence, and more weakly with other positive characteristics. One might argue that such an effect stems from genetics, i.e., phenotype implies something about genotype: I see no reason why having a symmetrical face should directly improve ones fitness. Or perhaps attractive individuals have better opportunities in general, and for that reason acquire the characteristics associated with having a charmed life. That is, the halo effect validates and thusly annihilates itself.

[FrancesH]

Ive noticed something interesting thats sort of this but backwards. This is how it goes: when I first meet someone, I might find them aesthetically displeasing. But as I get to know them, if they become my friend or just someone I find nice and intelligent, I find them prettier than I did to begin with. Possibly it`s just that as I now know more about them, their beauty or lack thereof becomes less important in my judgement about them.

[Desrtopa]

This probably has more to do with the exposure effect than a change of priorities.

[FrancesH]

Quite possibly; I am not certain. I have a question—not a challenge, just a question: I have noticed that it is much harder to call up the faces of people I am intimately familiar with than people I barely know. I thought that maybe it was because I knew more about the latter than the former, and that it was related to the above issue. If it is exposure effect, than maybe not. Any ideas?

Also, I may have accidentally reported your comment. If so, my huge, overwhelming, absolute apologies. (I thought I was hitting the reply button.) Consider it a newbie mistake and feel free to castigate me all you want. I will stand and accept your metaphorical rotten fruit.

[Alicorn]

Apparently mods can make reports evaporate. I have so evaporated; worry not.

[FrancesH]

Thank you. Thank you, thank you, thank you. Words cannot express the depths of my gratitude, and I am not being sarcastic.

[Alicorn]

You are quite welcome.

[RHollerith]

You used three backquotes in writing your comment that should be forward quotes. (Forward quote is to the right of the semicolon on almost every keyboard.) Probably good for you to know for the future.

[Lakehouse]

Grading on appearance is unconscionable and should be a federal crime but in my experience it’s very wide spread. Phd’s who do it say they’re right, but they can’t offer any justification. “It just is” is not an argument, it’s the ABSENCE of an argument. How is grading children on their appearance (and this happens) not child abuse? It causes emotional harm and may very well cause permanent cognitive damage. Grading on looks is an abomination born of eugenics, and it smacks palpably of Hitler and Nazi Germany. As someone personally abused by this, I consider it to be nothing short of evil.

I see mention here of some correlation between looks and intelligence. It is frequently true, but not an absolute. And it’s very possible that the cognitive advantages that come with better looks are to some degree the result of the enhanced nurturing, positive attention and social interaction they receive/​partake of during their developmental years.

[ekimekim]

While you are understandably upset, please NEVER compare something to the actions of Hitler/​Nazis. It in itself is such a famous and extreme case of negative affect bias (Hitler is bad, Hitler did X, therefore X is bad) that trying to use it only makes your other arguments seem less valid (this, too, is a bias, but its easier to avoid the trigger than try to eliminate the effect). A particularly tongue-in-cheek discussion of that particular example can be found here.

[wedrifid]

While you are understandably upset, please NEVER compare something to the actions of Hitler/​Nazis.

I endorse comparisons of Genghis Khan and Hitler whenever the topic happens to come up in conversation among people who aren’t oversensitive. Comparing various leaders on the impressiveness of their achievement or the magnitude of their moral transgressions is entertaining. Also acceptable is speculating about what would in a fight between, say, a tiger and a bear. Or 3 wolves vs a bear. Or a Hippopotamus vs a tiger, three wolves and a bear.

[mat33]

You are right. Most certainly so.

Nevertheless, it feels just fine to know, that democracy would most probably put something ratlike from the KGB ranks and dungeons into high security cell, and not in the White House.

[jeronimo196]

Nevertheless, it feels just fine to know, that democracy would most probably put something ratlike from the KGB ranks and dungeons into high security cell, and not in the White House.

Necroing, since this is such a stunning display of the Halo effect in action—as if there have never been a single handsome KGB agent, or all “rat-like” people belong in a a cell.

I realize this was an insult hurled at Putin, but it’s extremely poorly worded and with some unpleasant implications. Would a good looking KGB agent be preferable? Is Putin that physically repulsive, or is matt33 misjudging his appearance due to the horn effect?

[pnrjulius]

But what if it means we exclude, say, Stephen Hawking from candidacy due to his paralysis and resulting unattractive appearance, despite the fact that he is one of the most brilliant men alive?

[BlueAjah]

I think all those traits correlate, even when measured independently to avoid that effect.

Which makes sense for many reasons.

One reason: who are people going to marry? People of the same worth as themselves, but not necessarily from the same category. Smart rich men get to marry beautiful women, or the kindest women, or the most honest women, whichever they prefer. So the positive traits get mixed with each other, and the negative traits get mixed with each other.

[Rixie]

I wonder if this really one hundred percent bias? I hate saying this, but when I moved to a new school 3 years ago I immediately noticed one person that I found extremely unattractive, and he later turned out to be one of the “bad kids”, and did measurably bad things with two of his friends that no one besides them did. I don’t think it was hindsight, I remember the exact moment when I first saw him and thought that he wasn’t that attractive.

Could there possibly be some correlation between attractiveness and some other good qualities?

[Shield]

It’s only when reading about these sort of experimental results when the full atrocious stupidity of an education system that doesn’t educate people about logical fallacies begins to hit me.

They have this much of an impact on people in such critical situations where absolute neutrality is completely vital and yet no one seems to consider making it in any way mandatory to teach children how to THINK.

[christopherj]

Research has shown that we automatically assign to good-looking individuals such favorable traits as talent, kindness, honesty, and intelligence

The next question, of course, is do we assign good-looking individuals these traits more, or less often than nature does? For example, there is a correlation between physical attractiveness and intelligence. A pretty big one.

Beautiful People Really ARE More Intelligent

If you want to estimate someone’s intelligence without giving them an IQ test, you would do just as well to base your estimate on their physical attractiveness as you would to base it on their years of education.

[BakaPotato]

This article seems to be from quite a while ago and I’m only 15 (so I may not have as much experience if my thoughts are off) but I just wanted to point out that people who know things like this can use it to their advantage as well as it being a disadvantage to some. As for the situation with the workplace or a job...If a person working there is seen often and isn’t closed off in an office corner...That person is also more likely to attract more customers who are also affected by this “effect.” right?

[orthonormal]

Interesting case of an evolved heuristic gone wrong in the modern world.

Mutational load correlates negatively with facial symmetry, height, strength, and IQ. Some of these are important in assessing (desirability or inevitability of) leadership, and others are easier to externally verify. So in a tribe, you could be forgiven for assuming that the more attractive people are going to end up powerful, and strategizing accordingly by making favor with them. (Bit of a Keynesian beauty contest there, but there is a signal at the root which keeps the equilibrium stable.)

However, in modern society, we’re not sampling randomly from the population; the candidates for office, or for a job, have already been screened for some level of ability. And in fact, now the opposite pattern should hold, because you’re conditioning on the collider: X is a candidate either because they’re very capable or because they’re somewhat capable and also attractive!

Since all tech interviews are being conducted online these days, I wonder if any company has been wise enough to snap up some undervalued talent by doing their interviews entirely without cameras...

[Caperu_Wesperizzon]

Do we even want to stop giving attractive people all manner of advantages in all domains of life? Sure, sometimes it may be in your best interest to claim you do, but that’s a whole different matter.

[papetoast]

What possible advantages do you have in mind? I think it is just a bad, irrational thing to automatically assume attractive people to be smart or honest.