Bradly Alicea, Stefan Dvoretskii, Sam Felder, Ziyi Gong, Ankit Gupta, and Jesse Parent

pdf

Abstract: We present a framework for meta-brain models, or hybrid models that capture multiple aspects of developmental neuro- biology and behavior. A developmentally-inspired embodied learning agent architecture is combined with a contextually- explicit representational layer to form a complex artificial nervous system. The architectural description is summarized in terms of morphological differentiation, context-handling, and generalized architectural features. In conclusion, we pro- pose potential functional milestones that help further expli- cate the usefulness of our approach.

Part of the 2020 Workshop on Developmental Neural Networks