Abstract Selective breeding for specialized functional roles has resulted in consistent differences among domestic dog breeds in a range of cognitive capacities, as demonstrated by standardized cognitive assessments. What remains uncertain is whether these breed differences extend beyond initial task performance to influence how dogs adjust their behaviour or improve their performance when they are re-tested on the same tasks. In the present study, we used a structured hand-touch learning task to examine changes in Discrimination, Reversal learning, and resistance to Extinction performances across two testing sessions in 105 dogs representing five breed clades. Statistical analyses accounted for demographic variables, reward responsiveness, impulsivity, and training history. Dogs showed limited changes in learning performance across sessions, with statistically significant improvement observed only in Reversal learning from Test 1 to Test 2. Most breed-clade differences in average performance found during the initial test were no longer evident at Test 2. There was also limited evidence that breed clades differed in the extent of their performance change over time. Taken together, these results suggest that initial performance differences across breed clades likely reflect breed-specific cognitive profiles that shape how dogs engage with novel learning tasks. Following re-exposure to the task, average performance tends to improve and potentially mask the initial between-clade differences, although the magnitude of improvement appears broadly similar across breed clades. These findings highlight the importance of considering both initial performance and prior experience when interpreting breed differences in learning and cognitive capacities.
Azadian et al. (Fri,) studied this question.