November 1, 2011 | 1
Sci saw this paper being tweeted around the internets recently, a paper on the evolution of overconfidence. It seemed like a really interesting and cool paper, and I was kind of surprised that no one had covered it. I picked up the paper and took a look for myself, and I realized WHY. This paper is chok full of variables and functions and statistical modeling. It’s enough to make your head spin (though of course, I could be wrong and there was cooler science out there last week). But it’s a very interesting paper, which has some interesting explanations for human behavior. The net RESULT of the paper is clear enough, and the complicated math can be explained pretty simply.
All you need is a shot of Jager, and Pauly D and The Situation from The Jersey Shore.
Welcome to the blog boys!
Johnson and Fowler. “The evolution of overconfidence” Nature, 2011.
Ah, overconfidence. It can be blamed for black eyes, broken limbs, really bad relationship, housing bubbles, financial ruin, at least a couple of wars, and you can’t forget that time Ceaser up and crossed the Rubicon (overconfident much there, Julius?). I personally blame overconfidence for several running injuries and grad school. And every time we’re smacking our heads against our desks over being too overconfident, AGAIN, you kind of have to wonder WHY we feel this way? Wouldn’t it be a lot better for all of us if we were more realistic about our capabilities and what we could handle?
So you have to wonder, with all the costs of overconfidence, what kind of benefits it actually entails. And you have to wonder how overconfidence might be passed on and might become common in humans, if the fails are so incredibly dramatic. You’d think the benefits would have to be pretty dramatic, too. The authors of this study decided to come up with a model to see whether overconfidence is a evolutionarily feasible option. Does it maximize individual fitness, giving us greater access to resources and ability to produce offspring?
The result of this question was a four page paper, in which every other sentence introduces a new variable. It’s a lotta function. But it actually breaks down to be a good bit simpler than you’d think.
Let’s start with our Jersey Shore boys.
The two Jersey Shore boys are in their natural habitat, “the club”. They are in the bar area, and spy an undefended, and apparently free, shot of Jagermeister.
This is a Resource. And both of them want it. Now, there are three options that can happen here, say to Pauly:
1) They fight. The Situation gets distracted by a mirror containing his reflection. Pauly knocks him out and wins the Jager.
2) They fight. The Situation smacks Pauly against his legendary abs, causing a concussion. Pauly loses the Jager.
3) The Situation, confronted by the chemical wall of Pauly’s hair gel, becomes nauseated and decides he didn’t really want Jager anyway, and Pauly claims the uncontested shot.
Win, lose, or scare the other person off. Now, we assume here that people don’t really WANT to fight. Fighting is costly, if you win OR lose, someone is bound to get hurt.
So does Pauly stake a claim to the Jager? Well, he’s going to have to choose whether to do it or not based on how likely he thinks he is to win the fight. If one guy was obviously stronger than the other, presumably the weaker would head out in favor of saving his own neck. But what if you don’t KNOW?! What if Pauly and The Situation have never been in a fight before. Pauly knows the Sitch looks pretty mean and flashes those abs everywhere…but can he win the fight?
Add to this Pauly’s own assessment of his fighting potential. Sure he spends a lot of time at the gym, but does he have a lot of fighting experience? Does he have more than the Situation? Will his hair gel load potentially throw off his balance? And should he be OVERconfident? Realistic? Or underconfident?
Whether Pauly is overconfident will depend on the Jager, and potential injury from the fight. If the Jager is SUPER important, and injuries from losing the fight will be minimal, it pays to be overconfident in all situations. After all, losing won’t be that bad. If the value of the Jager is about as important as the pain from the potential injuries, you could have three possibilities. You could have overconfidence, UNDERconfidence, and lack of bias. When the odds are 50/50, all three conditions can exist. Finally, if the Jager doesn’t really matter, and the concussion might be really bad, well, then it pays to be underconfident, and just give up while you’re ahead. So in 2/3 of all situations (if they are evenly distributed), the resource will matter enough to make overconfidence a useful strategy.
As you can see here, when resources are valuable, overconfidence might win you the shot, and thus overconfidence will win a disproportionate amount of the time. Overconfidence is advantageous because it causes you to go for the resources when you might not otherwise win, and potentially bluffing your way out of it, and stronger but less confident rivals decide to go after easier prey. So it’s GOOD to be overconfident…at least as long as the potential pitfalls are smaller or equal to the potential wins (or maybe when the potential losses are too far down the road for us to concern ourselves about at the time).
The model that the authors put together also comes out with something interesting. The less you know about your opponent, the more the odds of injury approach the value of the resource (in your mind, anyway). This results in two peaks in behavior (according to the model), extreme overconfidence, and extreme underconfidence, with relatively little lack of bias. We have the highest levels of confidence, either over (hubris), or under (fear) when we are faced with an opponent or strategy that we know little about. I could apply this to US politics…but I think you can do that yourself.
So you see, in direct competition, it pays for Pauly to be overconfident. The more he bluffs, the bigger the chance that the Situation will find himself another drink. And if Sitch calls his bluffs, well, fighting makes for good reality TV. The benefits of winning ourweigh the cost of losing, so overconfidence is warranted. The authors hypothesize that this state of affairs may have been commonly in human history, making us more likely to display overconfidence to get what we want. Of course, this is speculation on the evolution of a BEHAVIOR, one which may or may not have a genetic basis. Certainly it is one that can be learned and imitated, which may make the ‘evolutionary’ aspects of this kind of moot (I think that’s probably the biggest flaw in this paper, and I would have titled it something like “The Probability of Overconfidence” or “The Benefits of Overconfidence” or something). Maybe, in this case, we just learn from our peers that overconfidence gets you places. And it appears that, 2/3 of the time, it might. But out of that 1/3 of fail, we have things like WWI, the war in Iraq, the financial crisis, and my bad knee. The authors note at the end that while overconfidence might have once been profitable, today the fails might not be worth the potential wins. Think VERY carefully before you bluff for that Jager, Pauly D.
Johnson, D., & Fowler, J. (2011). The evolution of overconfidence Nature, 477 (7364), 317-320 DOI: 10.1038/nature10384