Evidence is presented that the reciprocal exchange of social services
among chimpanzees (Pan troglodytes) rests on cognitive abilities that
allow current behavior to be contingent upon a history of interaction.
Food sharing within a captive colony of chimpanzees was studied by me
ans of 200 food trials, conducted on separate days over a 3-year perio
d, in which 6,972 approaches occurred among the nine adults in the col
ony. The success rate of each adult, A, to obtain food from another ad
ult, B, was compared with grooming interactions between A and B in the
2 hours prior to each food trial. The tendency of B to share with A w
as higher if A had groomed B than if A had not done so. The exchange w
as partner-specific, i.e., the effect of previous grooming on the beha
vior of food possessors was limited to the grooming partner. Grooming
did not affect subsequent sharing by the groomer, only by the groomee.
The effect of grooming was greatest for pairs of adults who rarely gr
oomed. Nevertheless, the effect was general: 31 dyadic directions show
ed an increase in sharing following grooming, and only 11 a decrease.
Food possessors actively resisted approaches by individuals who had no
t groomed them. After food trials there was a significant reduction of
grooming by previous possessors towards those individuals with whom t
hey had shared. (C) 1997 Elsevier Science Inc.