Abstract
Punishment is claimed as an important mechanism to stabilise costly cooperation in humans, but its importance in social animals has been questioned recently due to both conceptual considerations and a lack of empirical evidence (only few published studies). We empirically tested whether there is evidence for punishment in communally nursing house mice (Mus musculus domesticus, direct descendants of "wild" animals). Communally breeding females pool their litters and raise all offspring together, indiscriminately caring for own and other offspring. Such a situation resembles a public good and provides scope for exploitation if females vary in their relative contributions to the joint nest (offspring number). We allowed two females to communally breed and conducted removal experiments both in the presence and absence of pups. We aimed to test whether reduced investment by one of the females (induced through separation from the partner and their combined offspring for 4 or 12 hours) leads to increased aggression by the other female after the reunion. We found no evidence for punishment, on the contrary, females increased socio-positive behaviours. The costs of losing a partner in a communally breeding species might be too high and hinder the evolution of punishment. Our findings add to a growing list of examples questioning the role of punishment in cooperating non-human animals and emphasise the importance of empirical testing of its assumptions and predictions.