How You Learn Depends on When You Learn: Experimental Evidence From Cournot Contests

We conducted controlled laboratory experiments to investigate how humans adapt the decision rule (DR) they use in repeated strategic interactions in light of new information becoming available. We do so by providing -- along different paths -- more and more information over time, so as to trigger adoption and abandonment of different DRs along the way. We focus on 'Cournot contest' games and consider the three leading DRs in game theory: payoff-based learning, best-response, and imitation. Our study reveals path dependence in how people use heuristics in strategic decision-making, falsifying the salient assumption that DRs are triggered by the available information alone.