After this familiarization session, subjects were tested on two experimental sessions in which they were endowed with food instead of tokens. Pooling across all five subjects, monkeys chose cereal exactly as often as fruit, and therefore chose neither option any greater than chance, as confirmed by a binomial probability test (pooled proportion of choices to cereal: 50%, n=60, p=1.00, see figure 2). Our results suggest that researchers should further investigate how and in what circumstances the endowment effect could be evolutionarily useful (see Beggan 1992 and Santos & Lakshminarayanan 2008 for a similar discussion). When tested in experiment 1, capuchins preferred to eat fruit discs when they were made owners of fruit discs, but preferred to eat cereal pieces when they were made owners of these objects instead. We observed that all subjects were willing to trade a token for a single oat.
The monkeys only had to hand over a few rocks from the pile in their cage to the assistant. In addition, the endowment effect (and possibly other behavioural biases, see Santos & Lakshminarayanan 2008) appears to emerge in the absence of much experience. We presented the monkeys with a choice between one experimenter who always offered (and gave) one piece of apple, and a second experimenter who always offered two pieces of apple but half the time gave one piece, and half the time gave two.
Put in more economic terms, our capuchins prefer options that stochastically dominate, ones that tend to give them more food overall.
To date, however, fewer studies have examined the origins of these biases.