Can we replicate real human behaviour using artificial neural networks?

Figures & data

Table 1. Possible observables that can be used in a generic utility function

	absolute	relative
self	my payoff	my payoff/payoff of others
others	payoff of others	uniformity of payoffs

Table 2. Personal values that relate to the relevance of the observables from Table 1

	absolute	relative
self	self interest	conformity
others	altruism	fairness

Figure 1. Agent behaviour for different values of altruism. The y-axis shows the contribution of the agent, the x-axis shows the average contribution of all other agents.

Figure 2. Agent behaviour in the simulation (left) and in the experiment (right).

Figure 3. The difference between the presented framework (left) and strict reinforcement learning (right).