Unraveling the key to innovative problem solving: a test of learning versus persistence

The possibility that variation in the propensity to forage innovatively is attributable to variation in cognition is a matter of debate. Motor flexibility and persistence offer alternative viewpoints. The present study used a computational model to evaluate the relative contribution of these mechanisms to the innovation process. We modeled the effects of low and high motor flexibility on problem-solving performance, which provided a baseline against which to examine how performance changed when combined with operant learning or persistence. We titrated our models through a wide range of parameter values in order to explore where in the outcome space biologically meaningful effect sizes are likely to be detected. The baseline model accurately reproduced an enhancement of performance when relative frequencies of motor expression were balanced (high motor flexibility) rather than skewed (low motor flexibility). Operant learning enhanced performance, but only when agents persisted until they solved and only when motor flexibility was low. In scenarios where agents gave up even if they had not solved, persistence in response to occurrence of secondary cues improved problem solving in both motor flexible and motor inflexible individuals. In scenarios, where the benefits of persistence and learning were compared directly, the benefits of persisting were typically equal, if not greater than those of learning. Given the high metabolic cost of neural tissue, our simulations predict that selection for enhanced problem solving should select for processes that increase persistence (e.g., personality changes) rather than learning.