Cognitive Ability, Character Skills, and Learning to Play Equilibrium

In this topic, we aim to explore how cognitive ability and nature skills 

control mode, progress and the development of play towards Nash equilibrium in returned strategic interactions.1 Many economic interactions are strategic and repeated Given that cognitive ability and nature skills range greatly at the population level, to understand fully the dynamics of how people behave and acquire in reformed strategic interplays we need to ascertain the roles of cognition and character. Despite well-documented heterogeneity in cognitive ability and character skills, to the best of our knowledge, we are the first to study how cognition and character affect how people learn to play equilibrium. Our research was created to explain the following questions: (i) does cognitive ability influence the way that groups of agents learn to play equilibrium overtime when they join in repeated strategic interaction?; (ii) how do subjects’ learning processes depend both on their own cognitive ability and that of their opponents?; (iii) do character skills influence how people learn to play equilibrium?; and (iv) how do the effects of character skills compare to those of cognitive ability? We first measured the cognitive ability of our 780 subjects using a 60 question non-verbal Raven test. We classified each subject as either of high cognitive ability or of low cognitive ability according to whether her test score was in the top or bottom half of the distribution of scores in her session. We then matched our subjects into groups of three. In own-matched groups, all three members were of the same cognitive ability type. In cross-matched groups, 

the three members were of combined ability (either two high ability and one low ability subject, or vice versa). Topics found out their own cognitive ability type as well as the ability type of the other two group members.2 This design feature allows us to conduct an analysis of how behavior responds to the cognitive ability of competitors. Each group of three then played the p-beauty contest ten times without rematching and with feedback.3 In our p-beauty contest, the three group members simultaneously chose an integer between 0 and 100 inclusive, and the subject whose choice was closest to 70% of the average of the three numbers won $6. In the unique Nash equilibrium, all subjects choose 0; however, the game is well suited to study learning since best responses to non-equilibrium opportunities are often above 0, but with repetition, behavior tends to move towards equilibrium. Our focus on repeated interaction makes it natural to keep group membership constant over the ten steps of the experiment; surely, only with constant groups can we conduct our analysis of how often different groups converge to equilibrium play. We chose a group size of three to maximize the number of independent observations while ensuring that the game has an interesting strategic