Brosnan and de Waal reply

It is unlikely that inequity aversion appeared de novo in humans. It almost certainly evolved because individuals who responded to inequality disadvantageous to themselves increased their relative fitness compared with those who did not. We recognize several potential evolutionary precursors to disadvantageous inequity aversion (S. F. B., H. C. Schiff and F. B. M. de W., manuscript in preparation). First is the ability to recognize that rewards and efforts differ between individuals, which is also required for social learning, a skill present in capuchins3. Second is the propensity to react if another individual receives a better reward for a specific task. Third is sacrifice to alter another individual's outcome.

Our study mainly concerned the second ability, showing that capuchin monkeys react negatively when another individual gets a better reward for the same or less effort on a specific task. This finding suggests that precursors to inequity aversion are present in animals from which our lineage split millions of years ago. Although capuchins may be reacting somewhat differently from adult humans, we have still learned something about the behaviour's possible evolutionary trajectory.

Regarding the cross-cultural study, the lowest mean offer by a proposer in the ultimatum game was 26% of the total, whereas the lowest modal offer was 15%, both by the Machiguenga of Peru4. Such relatively high offers would not seem to be consistent with completely selfish individuals who lack any conception of fairness5.

As stated earlier1, although the mere presence of a higher-value reward affects the capuchins, their reaction is not the same as when a conspecific receives the higher-value reward. To ignore the differences between the inequality test and the food-control test is unwarranted. Our Fig. 1 does not permit any conclusions about the effect of the food-control test and was not used for this purpose; it is the data in our Fig. 2 that inspired our claim.

The frequency of refusals across trials increases when a partner receives the reward and decreases when a reward is merely visible. The conservative statistic we chose did not allow significance (P<0.05)1, but we have since subjected these data to a comparison of the slopes of the linear regressions across trials for each test6. This re-analysis shows that refusals in the food-control test decrease across time, whereas those in the inequality test and effort-control condition increase (F2,69=28.71, P<0.001). Our subjects therefore discriminate between a situation in which higher-value food is being consumed by a conspecific and one in which such food is merely visible, intensifying their rejections under only the former condition.