Do we bend it like Beckham?

This post is part of a small series of posts on social learning and cooperation.

Jean-Baptiste's reaction to Laurent Lehmann's (and his colleagues') criticism of Boyd and Richerson's models made me brood over the notion of prestige-biased imitation. This notion is central to their whole system: both to their theory of cutural inheritance, and to their theory of cooperation through cultural group selection. But let me explain…

Robert Boyd likes to illustrate prestige-biased imitation with this picture of David Beckham lending his name and face to sell razorblades. Advertisement moguls (not exactly the epitome of scientific rigor) claim that having Beckham on the ad greatly improves its efficiency, and big business executives (not exactly the embodiment of rationality either) are willing to follow them. Let us suppose that they're right. Let us make the additional supposition that Beckham's face does not work merely by attracting people's attention on the ad (for it would imply that our friends, the rational businessman and the rigorous ads mogul, would be wasting too much money on a trivial result). Then, we have to conclude, people are buying these razorblades on the recommendation of David Beckham.

Now, Robert Boyd remarks, David Beckham is just a footballer – sometimes a poorly-shaven one at that. Why would one copy his choice of razorblades?

"Prestige-biased imitation" is the explanation. In our ancestral environment, the story goes, it was much more difficult to find an appropriate response to life's challenges, than to simply copy the most successful individuals around us. These individuals were skillful and knowledgeable: they knew useful techniques and bits of information that helped them survive this famously harsh Pleistocene environment of ours. Because of that, they were successful: they had more power, and food, and mates. And because of their success, they were prestigious: they elicited signs of deference from others, even when they were not in a position to dominate them.

Since prestige is such a good indicator of success, and success, in turn, such a good predictor of survival skills, it made sense for us to imitate our fellow humans on the basis of the prestige they had in our society (now is not the time for a Proustian note on the social relativity of prestige – huge variations in the value assigned to individual from one social setting to the next). Evolution thus gave us a motivation to imitate prestigious and/or successful, and/or skillfull individuals. At the end of the day, the theory predicts that if a fellow human is sufficiently presitigious, successful and skillful, we tend to imitate everything he does. Like buying razorblades or marrying Posh Spice.

If true, this psychological hypothesis obviously has momentuous consequences for cultural evolution. Among other things, it predicts that, if David Beckham becomes a secluded Buddhist Monk, or a suicide bomber, a small but significant proportion of the world population will do so. And conversely, if many humans do things like giving away their lives for religious or political causes – something you do not see among other animals, but quite often among humans, chances are that somewhere, sometime, a Pleistocene David Beckham decided to give away his life for a cause.

The bearded Beckham

Is prestige-biased imitation based on prestige?

In order for the theory to work, Laurent Lehmann et al. have shown (in accordance with previous work by Boyd and Richerson) that prestige-based imitation must obey two criteria: first, it has to be independent of the actual content of the behavior that is copied (in the words of the theory, it has to be context-dependent, not content-dependent).

In other words, when you imitate David Beckham by buying razorblades, your decision has to be based on David Beckham's charisma alone, and not on some inference you made about what the razorblades do. Suppose, for instance, that you chose to buy the razorblades because you believe they are responsible for Beckham's subtle, artful, faux négligé three-days beard – after all, TV lore teaches that there is a direct causal link between a good razorblade and a great sex life (I see our friend the ad mogul nodding).  Then your decision is based on information that ultimately has to do with razorblades – and sex – but not with Beckham. The advertisement team has wasted money paying him: a nobody with a great 3-days beard woul have done the deal.

You might also have bought Beckham's razorblades because you were under the impression that they contributed significantly to his winning the heart of a former Spice Girl – which many people consider a success in life, at least in the UK. In that case, the advertisers lost their money on Beckham all the same: any other former-Spice-Girl's husband would have done the deal, at a cheaper price.

In both these cases, your decision did not originate in your admiration for Beckham, but in the perceived effects of buying a razorblades, as evidenced by the case of David Beckham. You were influenced either by David Beckham's supposed skills in choosing rasors, or by David Beckham's success.

To many contemporary defenders of the theory, like Joseph Henrich or Richard McElreath, this argument cuts little ice. Whether it is Beckham's skills, his success or his prestige that influenced you, it remains that David Beckham's presence on the ad prompted you to imitate him. There is, they argue, a total continuity between skills, success and prestige: it makes no difference to the human mind and its deeply-rooted imitation instincts. The rest is  philosophical picking.

For example, to these authors (I am using an example from McElreath), if I copy farmer X because I have seen him plant a new crop that gives far better yields than other crops, which of course caused him to gain much money, this is an instance of prestige-biased imitation, also known as success-biased, skills-biased imitation. It has everything to do with what I think about farmer X (even if that's the only thing I know about him). It has nothing to do with what I learnt about the new crop when observing farmer X.

There may be more than a conceptual difficulty here. This broad definition of prestige, as including skills and success, is a way to dispose of many embarrassing cases where imitation is content-dependent (i.e. mindful of skills and of their outcomes, and not easily swayed by prestige), and define them away as prestige-biased imitation, hence completely context-dependent. In that way, most cases where our imitation is based on information that we get about the world while observing others, trying to find out what strategies are best for us,  free from social prejudice, are pushed under the rug. Social influence is everything.

What happens if we don't push all these instances of imitation not based on prestige under the rug? This: if your decision to buy the same razors as Beckham was not based on Beckham himself, but rather on his good looks, or on his decent (by Girls-Band standards) romantic life, then you have no reason whatsoever to imitate him when he decides to become a suicide bomber or a Buddhist Monk. And advertisers lost their money.

The stupidity argument

A possible line of defence might be the following: we might dissociate Beckham's skills and successes from his prestige, if we were clever enough to do so. But we're not. Evolution has programmed us once and for all with a motivation for which prestige trumps everything else. The kind of razors that Beckham uses are completely irrelevant to his football success – but we are too dumb to see that. If somebody is prestigious, our imitation instinct will just tell us to imitate him in all other respects. Joseph Henrich remarks that prestigious people are deemed better than other on a wide range of performances, even in fields so far away from their original domain of skill and success that we have no good reason not to trust them there (I have been told that Juliette Binoche's recent dancing performance confirmed Henrich's view).

To put it another way, evolution has equipped us with the starkest rule of thumb to survive in society: copy the most prestigious individual, whatever he does. In order for this rule of thumb to work well enough to keep us alive as a species, two assumptions would have to be met: first, skills should be tightly correlated with mating success and power; second, revision rules that may prevent us from following an unskilled leader, by letting us realize his incompetence, should be so costly to maintain that it would never evolve, even though  its absence would cost the death of a great number of individuals.

Skills not rewarded

We all know that, in a given society, skills and success/prestige may not be present together. At least George Bush Jr. taught us that. How much decorrelated can they be? If the correlation has to be between a person's prestige and the majority of her skills (opening a bottle, making allies, singing, swallowing Bretzels, etc.), as the theory demands, it seems  obvious that, while she might derive much success from a handful of special skills in which she is gifted, there is no reason to expect her to be significantly more gifted than all her social inferiors in the majority of their skills.

This problem is especially vivid since Boyd, Richerson and Henrich often claim that their theory works best under conditions of extreme uncertainty (humans, the story goes, developped their imitating habit as a way to acquire cultural knowledge, that helped them adapt to a fast-changing Pleistocene climate). The problem with extremely uncertain environments, in which randomness plays a great part, is that they tend to decorrelate skills and success. Think of it this way: if I am a good hunter hunting in a very steady population of rabbits with regular habits, the result of my hunt does not have much to do with the rabbit population. I will regularly outperform less gifted hunters. But suppose the rabbit population begins to fluctuate violently, with rabbits popping up at the most unexpected times and places: it will increase the odds that I come back from the hunt without a rabbit to show for my great skills. If, however, someone comes back from hunting with rabbits aplenty, this will not be a good indicator of his hunting skills. He might just have gotten lucky. The informational value of hunting success will decrease: they will no longer be good indicators of hunting skills.

I will just sketch my second argument: it is extremely costly to imitate David Beckham when he decides to become a Monk or a suicide bomber, because that reduces one's fitness to zero. We do have the capacity to see the consequences of Beckham's idealistic actions and the detrimental consequences of imitating him: if we are not blind to success or skills, we should similarly be aware of failures and incompetence. If prestige-imitation causes us to disregard them (and it has to, otherwise fitness dictates us not to imitate stupid people and failures, prestigious though they might be), we should assume that, for some reason, it is impossible for our prestige-biased imitation mechanism to take them into account.

Is imitation ever genuinely blind and prestige-biased ?

I have no proper conclusion to offer, only a challenge: I am not aware of any good study unambiguously showing a case where imitation is clearly detrimental to the imitator, and clearly motivated by the model's prestige alone (that is, excluding his perceived skills or his success, real, expected or assumed). The only example I have been given is waves of suicide among teenagers following the deaths of Marilyn, Kurt Cobain, or the litterary suicide of Young Werther – from whom these waves got their names of "Werther effects". I find the evidence of such waves quite sloppy and anecdotal, with many biases in the treatement of data – discounting the fact that the suicide rate was already on the rise before the allegedly imitated suicide took place is a frequent move in suicide wave studies. The only case that seems genuine to me, is the first one – Werther. Sadly, Young Werther was neither prestigious, nor successful, nor skillful, nor real.

The version of dual-inheritance theory this post refers to can be found in chapter 2 of Henrich, J. and Henrich, N., Why Humans Cooperate: a cultural and evolutionary explanation,. Oxford University Press, 2007. Their formulation has the advantage of being recent and extremely clear. I am not sure to what extent other theorists like Robert Boyd endorse some aspects of it, like skill-biased imitation being equivalent to prestige-biased imitation.


  • comment-avatar
    Jean-Baptiste André 26 November 2008 (16:57)

    Great post. A short comment on the issue of star-based commercials (admitting that they do work). Surprisingly, in order for such a commercial to work, we do not really need to believe that the prestigious individual in it really uses the product he praises. No one really believes that Beckham uses this or that razor-blade. And, in fact, no one really cares. This is not the point. Does not this make the success of commercials even more surprising? My personal interpretation is as follows. I believe that the “solution” to this paradox is probably to be looked for in the direction of coordination. Commercials might help us solve coordination problems. At least in the occidental world, the products you use are social signals. You need to know which products will make you look good/rich/hype etc. And, in such a coordination problem, the information that can serve to coordinate does not necessarily need to be about anything “objective”. A good information to coordinate on product X is an information that makes you think that others will choose this product, and that makes you think that others will think that you will choose this product, etc. Such information, that makes the choice of X salient, can be objective information about the product’s quality (robustness, efficiency, luxury etc.), but they need not be. Any information that merely associate, in a convincing way, the product with something positive and desirable will do. The association needs to be convincing, but not in an operational way. The difference is slight, but I believe it explains why the mere idea to associate a successful male with a razor-blade can make other males coordinate in the choice of this razor-blade, which will become the “cool” razor-blade to have. And this works even if no one believe that the razor is actually responsible for Beckham’s success in football (what a joke!!), and even if no one believes that Beckham actually uses the razor-blade! Of course this is not the only effect involved in star-based commercials. Another is commitment. By definition, prestigious individuals have a lot to loose in term of prestige. So they cannot afford praising a very bad product that would, for instance, harm their buyers. Anyway, there are many possible explanations for the effects of prestige on imitation. And, I believe, like you Olivier, that assuming the existence of a general preference for all the behaviors expressed by prestigious individuals is not necessary.

  • comment-avatar
    guest guest 28 November 2008 (07:29)

    why can it not be that I see Beckham in the advert and I think ”Wow, that company was willing to pay my favourite star for that ad, so I shall return the favour and support them because I know some small part of my purchase will go to help make Beckham a better footballer” or whatever it is Beckham actually does. If they part with some pennies in the direction of Alan Guth (the astrophysicist) and I know that Alan maybe must need the cash or he wouldn’t be there, well, I can take a shot at their razor blades. I mean, hey, how different can their blades be from what I’m using now? So I might as well spend the expense for Alan’s benefit, right? My point being there could be any or all of ten thousand reasons behind the motivation to buy based on the clear association with a celebrity. I don’t think any rational human actually believes any superstar uses any gear for any other purpose other than the celebrity endorsement deal, we’re just not that stupid. But we can be willing to play along if we think we (or our star) can get a piece of the action…

  • comment-avatar
    Christophe Heintz 29 November 2008 (00:39)

    To Jean Batiste’s social signal factor and mrG’s support-my-idol and prestige-based-guarantee (he’s got too much to loose to lie) factors, I would like to add the attract-one’s-attention factor for explaining the success of ads featuring famous people. Because the faces of well known people attract attention (see also Ophelia’s post on fame), one is led to read the add. In the face of the competition for attention (cognitive ressources), it is already quite an achievement for an ad. Then, the recognition heuristic does the rest of the work when shoppers have to decide between 10 brands of razors (without much information, I choose the one that ’rings a bell’). If this last factor is the most important, then having Darth Vador rather than Beckman in the ad would be little detrimental to its success. If the Darth Vador ad had as much, or more, impact as Beckman one, them the prestige biased hypothesis would have to face an important counter-example. To which extent it is it important if the person in the add is prestigious or notorious?

  • comment-avatar
    Dan Sperber 30 November 2008 (20:26)

    Suppose there is a prestige-based bias as argued by Boyd, Richerson, Henrich. Surely, this is only one bias among several. Other biases may be context-dependent or content-dependent. So, in adopting or not adopting a specific trait exhibited by a prestigious individual, prestige-based bias may well be only one of the factors. Other factors may weigh in the same or in the opposite direction. As suggested in Jean-Baptiste’s comment, if the trait is, for instance, involved in coordination and if there is a bias (or some more cognitively-rich disposition) to use saliency in coordination, then prestige (which happens to confer saliency) and saliency per se will weigh in the same direction (a point well made by Christophe). (Razors, by the way, are not a particularly good example of this because my success in shaving does not depend on my coordinating with others.) If the trait, for instance suicide-bombing, is detrimental to the agent and there are biases against such detrimental traits, then the prior prestige of some prestigious suicide bomber would only tilt, at best, a few people who were already tempted and, for whatever reason, not wholly deterred by biases against it (in fact, if you think of it, if a football star prestigious in the population among which suicide bombers are being recruited became himself a suicide bomber, this could well help with the recruitment of new candidates). I n such conditions, it is rather hard to ascertain whether there exists a specific content-independent prestige-bias. One would need to separate its contribution from that of other factors. Not impossible, but has it been done in a satisfactory manner? One way is to find a trait that spreads even though there is no conceivable content-based bias in its favour (one anecdotal example that comes to mind is speech defects such as mumbling or stuttering spreading among the disciples of a prestigious professor). Another way is to have good enough evidence of the spread of a trait in the absence of a prestige factors and see what happens when prestige is added. Still, it is hard to think of a case free of uncontrollable possible confounds. The conclusion is not that there is no such thing as prestige-based bias – and there are theoretical arguments as to why it might exist – but that its existence – rather than the plausibility of its existence and easy would-be illustrations – is hard to establish.

  • comment-avatar
    Olivier Morin 30 November 2008 (21:45)

    Many thanks to all of you who left comments. A few quick responses: 1. Concerning advertisement and celebrities: Mr G. is right, of course, few people think that Beckham actually uses the razorblades – he just endorses them. Christophe is right to suggest that the ad may work merely by attracting the viewer’s attention (a point that is made in the post). I am tempted by Jean-Baptiste’s suggestion that celebrities provide a focal point for coordinated actions. Still, famous people in ads was merely an illustration for this post, and I feel I have to add that I am not aware of any good piece of work showing that, ceteris paribus, celebs-based advertisements are more efficient than no advertisement at all (come to think of it, is there any well-controlled study showing that advertisement is more efficient than nothing?). Even if there was, attentional biases, skills-related biases and success-related biases would be tricky to control for. But the technique of randomized internet advertisements, made popular by google ads and now widely used, should make these issues testable. 2. I agree with Dan in admitting that pure prestige biases (i.e. distinct from skill-based and success-based biases) might exist, but, if pure, would have to prevail in the face of other, conflicting biases. Yet I do not think that Boyd, or at any rate, Henrich would really agree with this statement: they tend to treat imitation based on for example, success, as though it was purely based on prestige, which amounts to ignoring the (confounding or competing) factors you allude to. Other psychological factors that might make prestige biases much more fragile, such as (in the case of a spread of altruistic behavior) hunger or the fear of death, may simply be left out of the models. 3. All in all, we all agree that what matters is not whether or not we bend it like Beckham, but what would motivate us to do so.Yet, if the question has any meaning at all, then imitating Beckham or any other celebrity is not a primordial, basic instinct, but the fragile result of a conflict of motivations, Beckham’s prestige being only one among many potentially adverse motivations. Would models of cultural group selection still hold if that assumption was taken seriously?