One of the pleasures of getting older is becoming wiser. Once you see the world through a new lens, it's hard to ever go back; the lessons you learn can forever change your perspective. Usually. But sometimes you realize you were wrong.
When I was in college, I had what seemed like a life-changing epiphany. My psychology professor told a story about the first Thanksgiving he spent with his in-laws when he was married. His wife was preparing the turkey, and after rinsing it in water and removing the giblets, she proceeded to cut the bird right in half. When he asked her why she cut the turkey in two, she said that was how her mother taught her. My professor, surprised by this fowl cooking technique, went to ask his mother-in-law why she cut the turkey in half before roasting it. She said that was the way her mother did it. Finally, they asked the matriarch. When grandma was raising her family, the oven was too small to roast a whole turkey, so she cut it in half so it could fit in the oven.
I don't know if that story was true, but the lesson was imparted. I did not want to go around cutting turkeys in half just because someone in the past had a small oven. I wanted to be someone who thought for myself, someone who valued ingenuity. After all, isn't humans' ability to solve problems on our own one of the traits that made us smart, evolutionarily adaptive creatures?
This is the question a team of researchers, representing fields of psychology, zoology, mathematics, evolution and neuroscience, set out to answer when they conducted a learning-strategies tournament. They invited participants from around the globe to submit strategies for the contest, which was designed to determine whether learning from peers (social learning), learning by innovation (asocial learning), or some combination of the two would lead to the maximum gains, and hence be evolutionarily advantageous.
The task was to earn as many points as possible from a ‘multi-armed bandit,' similar to a slot machine, but with one hundred arms. Each arm had a different payout, and represented a different behavior that could be selected. During each turn, there were three options: Innovate, Observe or Exploit. Innovate was the asocial learning strategy, where the player could test one of the arms to learn about its payoff, but without earning any points. Observe was the social learning strategy, where the player could learn the payoff another player had when choosing an arm in the past, but again without earning any points. Exploit was the only option that allowed the player to earn points. The risk of picking up outdated information is considered one of the weaknesses of social learning, so to account for this, the payoffs for each arm varied over time, so even if someone had a high payout from an arm in the past, it might not be so profitable now.
Evolution was accounted for in the tournament by giving players a life-death cycle. Players had a 1/50 chance of dying during each turn. If a player died, he/she was replaced by the offspring of another player. The probability of a player being chosen to produce offspring was proportional to the player's lifetime earnings. Players who earned the most should have the most offspring, and in so doing have the most evolutionary fitness.
Of the 104 entries to the tournament, most favored a strategy that evenly mixed Innovate and Observe for learning. The study's authors, along with a panel of experts who oversaw the tournament, predicted that one of these mixed strategies would win. Apparently I'm not the only one who had a bias towards originality and self-reliance. Yet, the winner was not among the entries that used a balanced strategy. The winner, along with almost all of the most successful participants, heavily favored observation: social learning. They were copycats.
Why would observation be so successful? According to the authors, "Social learning proved advantageous because individuals frequently demonstrated the highest-payoff behavior in their repertoire, inadvertently filtering information for copiers." Once someone invents the printing press, they're not going to go back to writing everything out by hand; they've already got a winning strategy. As a social learner, you get the benefit of the most useful knowledge without having to do all the ground work.
What about the risk of learning outdated strategies, such as cutting a turkey in half? The winner of the tournament placed a higher value on more recently acquired information and discounted older knowledge. This is part of the principle that makes Wikipedia so successful; the information you're getting reflects the most up-to-date understanding because it is constantly revised and edited.
One caveat about social learning is that although it was the winning strategy, it might decrease the overall fitness of a population, making everyone dumber. To test this, the authors of the study did a simulation where individuals could only learn through innovation, but not through observation. The strategies that relied heavily on innovation scored higher in this simulation than the winner of the group competition
. Don't give up on innovation, but even Isaac Newton acknowledged that his scientific success was made possible by standing on the ‘shoulders of giants,' the work of his predecessors.
Humans' evolutionary success is due in large part to our ability to learn from others, and this may be far more important than our ability to invent. After contemplating the results of this study, I began to see the turkey anecdote in a new light. Even if the constraints which dictated grandma's peculiar bird preparation are no longer a hang-up, it's not unreasonable to keep using her recipe. As long as everyone happily gobbles up Thanksgiving dinner, there's no reason to innovate with cooking.
Another revelation from the tournament, which has personal resonance for me, is that the winning strategies focused most of their energy, 80-90% of their turns, on exploiting knowledge for points rather than learning more. As I'm heading into my fifth year of graduate school, isn't it time I think about getting a real job?
Rendell L, Boyd R, Cownden D, Enquist M, Eriksson K, Feldman MW, Fogarty L, Ghirlanda S, Lillicrap T, Laland KN. Why copy others? Insights from the social learning strategies tournament. Science. 2010 Apr 9;328(5975):208-13.