Neutral reference stimuli improve interpretation of choice behavior in touchscreen pairwise discrimination in male mice.

The touchscreen pairwise discrimination (PD) paradigm is a standardized behavioural assay for assessing cognition in rodents. PD performance is typically quantified by accuracy, but under disease models or pharmacological manipulations, accuracy alone may mask heterogeneity in behavioral strategies, limiting interpretation of the behavioral processes contributing to task performance. Here we introduced dynamic neutral images as reference stimuli to operationally compare two components of observed choice behavior: reward-guided selection of S+ and suppression of responding to S-. When the rewarded correct stimulus (S+) was paired with five different neutral images (S-), mice selected S+ in > 80% of trials across all pairings. In contrast, when the unrewarded incorrect stimulus (S-) was paired with the same five neutral images (S+), selection of the incorrect stimulus was highly variable (20-60%). We then extended training in conditions in which the same non-rewarded stimulus was repeatedly presented against different rewarded stimuli. Extended training reduced incorrect-stimulus (S-) choices to < 20%; however, when the incorrect stimulus (S-) was re-paired with neutral images (S+), incorrect-stimulus (S-) choice rebounded (20-40%), with mice preferentially choosing neutral images (60-80%). These results indicate that, under a reward-only contingency, high PD accuracy in healthy mice primarily reflects reinforced approach to S+ rather than suppression of responding to S- . Neutral reference patterns reveal the behavioral basis of task performance and improve the interpretability and reliability of PD outcomes.