Trait inferences from token faces do not predict behavior in the trust game
Author(s):
Chris Said
Field:
Social
Name/description of effect:
Perceived traits in neutral faces influence economic behavior of the perceiver.
Year(s) of attempted replication(s):
2005
Number of experiments:
1
Replication status:
Abject failure
Sample size(s):
44
Brief summary of experiment(s):
Each subject played an iterated trust game against 24 computer opponents, with the subject in the role of proposer and the computer opponents in the role of responder. Each computer opponent was assigned a name and a face, which the subjects believed were included only as a memory aid. On each round, the subject could propose up to 10 symbolic U.S. dollars to the opponent. Subjects knew that upon making the proposal, the money would double in value, and the opponent would return some fraction of the doubled amount.
Returns were selected from a normal distribution around a fixed mean for each opponent, with half the opponents returning a mean of 45% and half returning a mean of 55%. Subjects played 6 rounds against each opponent. At the end of the experiment, one round was randomly selected to count as real money so that subjects were paid the returned amount plus any of the initial $10 that they chose to keep.
While there was a significant effect of returner type, there was no significant effect of perceived trustworthiness of the faces, as rated by a separate group of subjects.
Of course, since the subjects were told the faces were tokens, this is not a failure to replicate findings in which subjects believed the faces belonged to the opponents.