We use a novel experimental design to examine the role of reputational concerns in explaining conditional cooperation in social dilemmas. By using the strategy method in a repeated sequential prisoners' dilemma in which the probabilistic end is known, we can distinguish between strategically and non-strategically motivated cooperation. Second movers who are strong reciprocators ought to conditionally cooperate with first movers irrespective of whether the game continues or not. In contrast,

more »

... tegically motivated second movers conditionally cooperate only if the game continues and they otherwise defect. Experimental results, with two different subject pools, indicate reputation building is used around 30% of the time, which accounts for between 50% and 75% of all realized cooperative actions. The percentage of strong reciprocators varied between 6% to 23%.