Empathy, The Prisoner’s Dilemma and In-Groups and Out-Groups

The purpose of this research is to investigate the relationships between individual differences in (1) adult theory of mind ability and (2) interpersonal competitiveness /cooperativeness strategies as measured using the iterated prisoners’ dilemma (IPD) game, as well as in-group vs out-group players.

Measure of Competition vs Cooperation

Prisoners’ dilemma is based on game theory and game theory is based on investigation of the logic of decision making in interactive games. There needs to be at least 2 players who are informed about which choices they have and what would be the outcome of each choice depending on the interaction with the other player’s choice. Each outcome is the result of the interaction between strategies of each player. Game theory can be characterized as being an abstraction of a real social situation involving payoffs (Gummerum, Hanoch, & Keller; 2008). Prisoner’s dilemma is one kind of ‘mixed motive’ strategy game in which the players can choose either competitive or cooperative strategies. It is most beneficial to play competitively but only if other players are cooperative. If every player competes, the worst outcome will occur as a result. In a prisoner’s dilemma game that is iterated (IPD), a cooperative strategy can result in the highest payoffs (Axelrod, 1985).  Typically individuals playing the IPD game combine competitive and cooperative ‘turns’ in iterative games, but there are individual differences in competitive and cooperative strategies. For this reason, the IPD game was used to assess the relative biases of individuals towards competition or cooperation. The payoffs for the iterated game are shown in the figure below:

Measures of ‘Mindreading’

‘Cold empathy’ mindreading ability was tested with a shortened version of Kinderman’s comprehension tasks widely used in the investigation of adult theory of mind (Kinderman et al., 1998).

‘Hot empathy’ skills were tested using Baron-Cohen and colleagues’ revised version of the Reading the Mind in the Eyes Test  (2001), which involves selecting normed motivation or emotion related adjectives for photos of 36 pairs of eyes.

Hypotheses Tested in Our Initial Study

Hypothesis 1

Based on the characteristic of IPD games in terms being cooperative is more beneficial in long run (Axelrod, 1985), participants with more cooperative strategies would have higher payoffs.

Hypothesis 2

As a replication of the Majolo et al (2006) study, it was predicted that the level of cooperativeness would be higher in the IPD game with friends than with strangers.

Hypothesis 3

Higher levels of cooperation would be positively correlated with higher performance on the ‘hot empathy’ theory of mind test. This was predicted on the assumption that hot empathy is always required for cooperative behaviours, as suggested by Paal and Bereczkei (2007).

Hypothesis 4

Higher levels of cooperativeness in the IPD will be positively correlated with higher performance on the ‘hot empathy’ ‘reading the mind in the eyes’ test, but not the ‘cold empathy’ theory of mind test. This is predicted based on previous results on above average theory of mind abilities among high scoring machiavellians (Davis & Stone, 2003; Sutton, 2001).


Hypothesis 1was confirmed. Here is a scatter plot showing the relationship between mean cooperation level and total score in the Prisoners’ Dilemma game. Cooperation level was calculated by counting the total number of cooperative moves for each subject.

Hypothesis 2was not confirmed. A repeat measures ANOVA with a single within subjects factor of group (playing with whom) with three levels (friend-unknown-experimenter) revealed no significant pattern of difference in the level of cooperation (F(2) = 1.12, p = 0.33).

Hypothesis 3was confirmed. The scatter plot below shows the relationship between mean cooperation and the mind in the eyes test score. It can be seen that mean cooperation level is positively correlated with the mindreading score obtained from the emotion recognition test.

Hypothesis 4 was confirmed. The scatter plot below shows the relationship between cooperation level in the ‘friends’ condition and scores obtained from the hot empathy ‘eyes’ test. It can be seen that the level of cooperation and the mindreading scores obtained from the emotion recognition test were positively correlated. This relationship was found to be highly statistically significant (r = 0.51, p = 0.001). By contrast, cooperation levels with strangers did not correlate with the hot empathy mindreading score.

Further research

The research question we are currently addressing is why there is such a strong correlation between level of cooperation and ‘hot empathy’ skills (”reading the mind in the eyes’ test scores) among friends but not among strangers.