I Congreso de Postgrado fcfm: ingeniería, ciencias e innovación

148 Santiago, 10 al 12 de agosto, 2022 OPTIMAL FOOD SHARING LOOKS LIKE A SOCIAL NETWORK Francisco Plana 1* , Jorge Pérez², Andrés Abeliuk¹ , ³ ¹Departamento de Ciencias de la Computación, FCFM, UChile, Santiago, Chile. ²Instituto Milenio Fundamentos de los datos, Santiago, Chile. ³Centro Nacional de Inteligencia Artificial, Santiago, Chile. *Email: franciscoplana@gmail.com RESUMEN It has been argued that hunter-gatherers' food-sharing may have provided the basis for a whole range of social interactions, and hence its study may provide important insight into the evolutionary origin of human sociality. Motivated by this observation, we propose a simple network optimization model inspired by a food-sharing dy- namic that can recover some empirical patterns found in social networks. We focus on two of the main food-shar- ing drivers discussed by the anthropological literature: the reduction of individual starvation risk and the care for the group welfare or egalitarian access to food shares, and show that networks optimizing both criteria may exhibit a community structure of highly-cohesive groups around special agents that we call hunters, those who inject food into the system. These communities appear under conditions of uncertain and scarce food, which sug- gests their adaptive value in this context. We have additionally obtained that optimal welfare networks resemble social networks found in lab experiments that promote more egalitarian income distribution, and also distinct distributions of reciprocity among hunters and non-hunters, which may be consistent with some empirical re- ports on how sharing is distributed in waves, first among hunters, and then hunters with their families. Regarding our methods, we employ an original pipeline of state-of-art clustering algorithms to analyze the multiple network optima of our model, which were computed by evolutionary algorithms. Finally, our model also relies on an orig- inal formulation of starvation risk, and it may contribute to a formal framework to proceed in this discussion regarding the principles guiding food-sharing networks. MOD E L AM I E N TO MAT EMÁT I CO 13