Experimental Evidence on Iterated Reasoning in Games

Area of Science:

• Behavioral Economics
• Game Theory
• Cognitive Science

Background:

• Iterated reasoning in games involves backward induction and interactive knowledge.
• Previous studies offer limited insights into cognitive skills and player beliefs.
• Understanding limitations in iterated reasoning is crucial for game theory applications.

Purpose of the Study:

• To experimentally investigate backward induction and interactive knowledge.
• To disentangle cognitive abilities from beliefs about co-players' rationality.
• To provide reliable estimates of subjects' iterated reasoning skills.

Main Methods:

• Experimental design to measure cognitive skills in iterated reasoning.
• Distinguishing between cognitive limitations and beliefs about co-players.
• Analysis of performance in the 'dirty faces' game.

Main Results:

• Subjects' skills in iterated reasoning are estimated pessimistically compared to prior literature.
• Beliefs about co-players' rationality do not explain limited reasoning in the 'dirty faces' game.
• Backward induction skills predict interactive knowledge skills, suggesting generalized reasoning ability.

Conclusions:

• Cognitive abilities, not beliefs about others, are key to limited iterated reasoning.
• Backward induction is a strong predictor of interactive knowledge.
• Evidence points to a generalized capacity for iterated reasoning among subjects.