Abstract
Humans are expert decision makers, capable of assimilating information rapidly and tailoring behaviour optimally, according to task constraints and context. Thus, individuals can produce an effective response to a visual stimulus in a short time frame. Nevertheless, the mechanisms that evaluate evidence and reach decisions can sometimes select sub-optimal behaviours, with decision-making appearing to become ‘stuck in a rut’. We developed a model of learning that revealed this inertia is a naturally emergent feature of a learning system. To test the model, 30 participants (16 female, 14 male, mean age 26.8 years) completed an aiming task designed using specialised software presented on a digitizing tablet (Toshiba Portege M700-13P). A handheld stylus was used as an input device to move a cursor between two points displayed on a computer screen without hitting an obstacle blocking the route. In ‘sequential’ conditions, the obstacle was displaced to either to the left or the right of the screen and then incrementally moved 15 times away and then towards the starting positions. In the ‘random’ condition, the obstacle appeared randomly in one of the 15 positions (twice per session). Post-hoc interviews showed participants were unaware that the obstacle moved from trial-to-trial. In the random condition, participants showed a high bias towards selecting the shortest route between the points. In the sequential conditions, participants showed a bias towards the previous selected route (a phenomenon that can be termed hysteresis) even though this was a sub-optimal route avoided in the random condition. The learning model predicted precisely this qualitative pattern of decision-making and demonstrated that the emergent hysteretic effect evident during sequential conditions does not develop with randomisation of the motor task sequence. These results suggest that an understanding of responses to visual stimuli requires a consideration of the learning mechanisms underpinning skilled behaviours.