Symbol emergence by combining a reinforcement learning schema model with asymmetric synaptic plasticity

A novel integrative learning architecture, RLSM with a STDP network is described. This architecture models symbol emergence in an autonomous agent engaged in reinforcement learning tasks. The architecture consists of two constitutional learning architectures: a reinforcement learning schema model (RLSM) and a spike timing-dependent plasticity (STDP) network. RLSM is an incremental modular reinforcement learning architecture. It makes an autonomous agent acquire behavioral concepts incrementally through continuous interactions with its environment and/or caregivers. STDP is a learning rule of neuronal plasticity that is found in cerebral cortices and the hippocampus. STDP is a temporally asymmetric learning rule that contrasts with the Hebbian learning rule. We found that STDP enables an autonomous robot to associate auditory input with its obtained behavioral concepts and to select reinforcement learning modules more effectively. Auditory signals that are interpreted based on obtained behavioral concepts are revealed to correspond to “signs” in Peirce’s semiotic triad. This integrative learning architecture is evaluated in the context of modular learning.