Nectar of the Bots: Evolving Bidirectional Referential Communication

Referential communication is central to social and collective behaviour, for example honey bees communicating nectar locations each other or co-workers gossiping about a colleague. Since such behaviour typically considered be ‘representation hungry’, it often assumed require the possession of complex cognitive machinery capable manipulating symbolic representations world. However, series simulation studies have shown that can achieved by very simple embodied artificial agents controlled evolved recurrent neural networks are challenging interpret in symbol-processing terms. In this paper, we extend paradigm explore scenarios which pair agents, privy different piece private information, must jointly solve task requires both pieces information communicated, compared acted upon, i.e., agent simultaneously play role signaller receiver during an unstructured referential interaction bidirectional. We demonstrate able task, analyse extent their situated, embedded communicative step towards understanding minimal basis human language.