Response learning confounds accurate assays of inhibitory control

Abstract

The ability to inhibit prepotent actions towards rewards that are made inaccessible by transparent barriers has been considered to reflect capacities for inhibitory control (IC). Typically, subjects initially reach directly, and incorrectly, for the reward. With experience, subjects may inhibit this action and instead detour around barriers to access the reward. However, assays of IC are often measured across multiple trials, with the location of the reward remaining constant. Consequently, other cognitive processes, such as response learning (acquisition of a motor routine), may confound accurate assays of IC. We measured baseline IC capacities in pheasant chicks, Phasianus colchicus, using a transparent cylinder task. Birds were then divided into two training treatments, where they learned to access a reward placed behind a transparent barrier, but experienced differential reinforcement of a particular motor response. In the Stationary-Barrier treatment, the location of the barrier remained constant across trials. We therefore reinforced a fixed motor response, such as always go left, which birds could learn to aid their performance. Conversely, we alternated the location of the barrier across trials for birds in the Moving-Barrier treatment, and hence provided less reinforcement of their response learning. All birds then experienced a second presentation of the transparent cylinder task to assess whether differences in the training treatments influenced their subsequent capacities for IC. Birds in the Stationary-Barrier treatment showed a greater improvement in their subsequent IC performance after training compared to birds in the Moving-Barrier treatment. We therefore suggest that response learning aids IC performance on detour tasks. Consequently, non-target cognitive processes associated with different neural substrates appear to underlie performances on detour tasks, which may confound accurate assays of IC. Our findings question the construct validity of a commonly used paradigm that is widely considered to assess capacities for IC in humans and other animals.