<Emphasis Type="Italic">Cebus apella</Emphasis> Tolerate Intermittent Unreliability in Human Experimenters

Monkeys form expectations for outcomes based on interactions with human experimenters. Capuchins, a cooperative New World monkey species, not only anticipate receiving rewards that the experimenter indicates, but also apparently anticipate rewards based on what the experimenter has given to their partners. However, this could be due to subjects responding to either outcomes or experimenters. Here we examine whether capuchins will continue to interact with human experimenters who are occasionally unreliable. We tested 10 monkeys with a series of familiar human experimenters using an exchange task. The experimenters had never before participated in exchange studies with these monkeys, hence the monkeys learned about their behavior during the course of testing. Occasionally experimenters were unreliable, failing to give a reward after the monkey returned the token. The monkeys did recognize these interactions as different, responding much more quickly in trials following those that were nonrewarded than in other situations with the same experimenter. However, subjects did not change their preference for experimenters when given the opportunity to choose between the unreliable exchanger and another exchanger, nor did subjects learn to prefer reliable experimenters from watching other monkeys’ interactions. Instead, subjects returned the tokens to the same location from which they received it. These results indicate that capuchins may not be sensitive to isolated instances in which experimenters are unreliable, possibly because of a strong bias to returning the token to the location from which it was donated.