Intrinsic motivation and learning dynamics

We investigate the effects of intrinsic motivation on the dynamics of learning processes. We construct a simple model of a single agent adapting to un-known environment. Performing a repeated choice between a number of initially unexplored alterna-tives, the agent gains rewards for each selected al-ternative and in doing so gradually comprehends the environment. In our model the agent choice is governed by two stimuli. The traditional extrin-sic motive inclines the agent to maximize the cu-mulative payoff throughout the process, while the second, intrinsic one, biases the agent towards the novel options that she inherently likes. We show that the intrinsic motivation can induce an insta-bility and periodic dynamics of the learning pro-cess which is always stationary in the case of self-ish, rational agent. Interestingly, the opposite ef-fect can arise as well: when the impact of intrin-sic motivation on the agent choice is strong, the equiprobable choice equilibrium strategy becomes stable. Based on the presented results we argue that the effects of human intrinsic motivation in particular and bounded rationality in general may appear to contribute greatly to the dynamics of complex socio-economic systems and therefore de-serve much attention in the formal models of such systems.