Learning and Retrieval of Memory Elements in a Navigation Task

摘要: Desert ants when foraging for food, navigate by performing path integration and exploiting landmarks. In an earlier paper, we proposed a decentralized neurocontroller that describes this navigation behavior. As real ants, landmarks are recognized depending on the context, i.e. only belong to towards current goal (food source, home). version, neither position nor quality of food sources can be learnt, memory is preset. article, present new whose elements allow learning place vectors quality. When agent meets it updates value, if source already known, or stores quality, new. Quality values used select visited. one has too low also finds shortcut another known source.