Autonomously learning beliefs is facilitated by a neural dynamic network driving an intentional agent

Intentionality is the capacity of mental states to be about the world, both in its “action” (world-to-mind) and its “perception” (mind-to-world) direction of fit. An intentional agent must be able to perceive, act, memorize, and plan. These psychological modes may be driven by desires and be informed by beliefs. We have previously proposed a neural process account of intentionality, in which intentional states are stabilized by interactions within populations of neurons that represent perceptual features and movement parameters. Instabilities in such neural dynamics activated the conditions of satisfaction of intentional states and induced sequences of intentional behavior. Here we explore the idea that the process organization of such intentional neural systems enables autonomous learning. We show how beliefs may be learned from single experiences, may be activated in new situations, and be used to guide behavior. Beliefs may also be dis-activated when their predictions do not match experience, leading to the learning of a new belief. We demonstrate the idea in a simple scenario in which a simulated agent autonomously explores an environment, directs action at objects and learns simple contingencies in this environment to form beliefs. The beliefs can be used to realize fixed desires of the agent.