Flexibility of learning in complex worlds

Learning to adjust to changing environments is an important aspect of behavioral flexibility. Here we investigate the possible advantages of flexible learning rates in volatile environments, using learning simulations. We compare two established learning mechanisms, one with fixed learning rates and one with flexible rates that adjust to volatility. We study three types of ecological and experimental volatility: transitions from a simpler to a more complex foraging environment, reversal learning, and learning set formation. For transitions to a complex world, we use developing cleaner fish as an example, having more types of client fish to choose between as they become adult. There are other similar transitions in nature, such as migrating to a new and different habitat. Performance in reversal learning and in learning set formation are commonly used experimental measures of behavioral flexibility. Concerning transitions to a complex world, we show that both fixed and flexible learning rates perform well, losing only a small proportion of available rewards in the period after a transition, but flexible rates perform better than fixed. For reversal learning, flexible rates improve the performance with each successive reversal, because of increasing learning rates, but this does not happen for fixed rates. For learning set formation, we find no improvement in performance with successive shifts to new stimuli to discriminate for either flexible or fixed learning rates. Flexible learning rates might thus explain increasing performance in reversal learning, but not in learning set formation. We discuss our results in relation to current ideas about behavioral flexibility.