Machine Learning Reveals Modules of Economic Behavior from Foraging Mice

Complex ethological behaviors could be constructed from modules that are discrete functional units of behavior with a genetic basis. Here, we test this idea for foraging, and develop a paradigm to dissect foraging patterns in mice. We uncover discrete behavioral modules linked to round trip excursions from the home. Machine learning reveals 59 modules across different genetic backgrounds and ages. Different modules develop at different ages and are linked to different aspects of economic behavior, including memory, reward, risk and effort responses. Crosses of distant mouse strains reveal that parental and genetic effects shape foraging differently, and parental effects grow stronger with age. Specific behavioral modules, genes and pathways are found to be sensitive to parental effects. One candidate gene, Magel2, is linked to Prader-Willi Syndrome and shaped the expression of discrete modules in an age-dependent manner. Our results reveal building blocks for normal and abnormal economic behavior patterns.\n\nHIGHLIGHTSO_LIIdentification of 59 economic behavior modules underlying foraging\nC_LIO_LIDiscrete modules are linked to memory, reward, risk and effort responses\nC_LIO_LIGenetic and parental effects shape foraging by changing module expression\nC_LIO_LIMagel2, a Prader-Willi Syndrome gene, affects specific modules at specific ages\nC_LI