Fig. 2: Game structure and behavioural results.

a An α parameter determines the social context and thus the amount that each player receives on each trial. The experimental design contained three social contexts that were hypothesised to shift people’s competitiveness. In all contexts, the position of the closest participant to the target determined the total reward won. In the first context (cooperation), this reward would be equally shared among both players. In the second - intermediate - context, in every trial the winner takes all the reward available. In the third context, the closest player to the target wins twice the reward while the loser loses the reward from its endowment. b, c The strategy adopted by most participants in the cooperative context was to cooperate, and in the competitive context, to compete. In the intermediate context, participants exhibited variable responses. Responses are presented as their joint position on the x-axis (b) or over time (c). Kernel densities are presented on the right of each plot. Mean (bold line) and standard error (shaded area) are displayed across participants. d Average deviation ΔP (change from previous position) in a trial as a function of the co-player deviation in the previous trial. Each dot represents a participant. The co-player deviations are binned into large and small increases in cooperation/competition. In all contexts there is a tendency to reciprocate the co-player changes of behaviour in the next trial (tit-for-tat). This is particularly evident in the intermediate context, where participants were sensitive also to small increases in competition.