Do We Learn from Our Own Experience or from Observing Others?

Learning in real life is based on different processes. Humans learn to a certain extent from their own experience but also learn by observing what non directly related others have done. In this paper we propose a generalized payoff assessment learning (GPAL) model which enables us to evaluate the relative influences of these two different models of learning. We apply GPAL to a homogeneous good Bertrand duopoly experiment with random matching and population pricing information. The model explains the observed pricing and learning behavior at least as well and often better than learning models from the literature but has the advantage that the relative influence of the learning models can be estimated. We find that the own experience overwhelmingly dominates learning, despite the useful information about behavior of potential future opponents contained in the population price distribution.