This article helped me understand the game a bit better:
http://mindyourdecisions.com/blog/20.../#.U5rcjSjuzkc
Bottom of page 20 in this
http://www.econ.ucsb.edu/~tedb/Cours...rerChimps.pdf:
Quote:
Note that the mixture probabilities for one player depend only on the payoffs of the other player. For example, if the (Left, Left) Matcher payoff was X rather than 4), the NE mixtures would be p*=.5 (i.e.,the Matcher is not predicted to change behavior at all) and q*=1/(X+1) (i.e., the Mismatcheris predicted to choose Left less often, as if to deny the Matcher the high X payoff). This is a counter-*‐intuitive feature. Any learningalgorithm that is guided by received payoffs (such as reinforcement learning) will therefore adapt, at least in the short-*‐run, in the wrong direction.
Does that basically say that humans learn, maybe faster than chimps, but the learning leads to a bad result in this experiment? And also if that's true could it be said that humans are better at learning than chimps in some sense but in this game it leads to a bad result? Similar to someone trying to apply sauce/galfond high level concepts at nl10 at failing miserably (probably due to bad application as well)?
e: Also wrt poker a fish as the Matcher in the asymmetric and inspection games would always choose left cos winning a big pot is yay, and then GTO Mismatcher would win less than an exploiting Mismatcher
Last edited by Finnisher; 06-13-2014 at 07:54 AM.