Can Agents Without Concepts Think? An Investigation Using a Knowledge Based System

Grid-World is a working computer model which has been used to investigate the search capabilities of artificial agents that understand the world in terms of non-conceptual content. The results from this model show that the non-conceptual agent outperformed the stimulus response agent, and both were outperformed by the conceptual agent. This result provides quantitative evidence to support the theoretical argument that animals and pre-linguistic children may use non-conceptual content to understand the world. Modelling these ideas in an artificial environment provides an opportunity for a new approach to artificial intelligence.