Signaling smarts? Revealed preferences for self and social perceptions of intelligence

We design a laboratory experiment to test for image motives in a setting where decisions signal intelligence to a social audience. Money-maximizing behavior in the experiment sorts subjects by academic ability, as measured by performance on verbal analogy questions, across two levels of question difficulty. Sorting behavior is publicly revealed in our “audience” treatment, facilitating social signaling. In the “intrinsic only” treatment, the signaling mechanism was explained but decisions were kept private. In the control, there was no discussion of the signaling mechanism and all decisions were private. We find that intrinsic only subjects were significantly more likely to choose the high-type action than the control. In comparison, subjects were significantly less likely to choose the signal in the audience treatment, when doing so was publicly observed. The effects are more pronounced in males. The results suggest that social observation can demotivate individuals when decisions signal intelligence, despite evidence that the underlying trait was privately considered desirable. Audience effects have a less predictable impact on behavior in this setting as compared to the near universally positive findings from the altruism and trust literature. Our experimental design can be easily adapted to study image motives in a broad set of environments using revealed preferences.