Does the cognitive bias test in dogs depend on spatial learning?

Abstract

Cognitive bias tests have been used to measure affective states in dogs. The results may also involve spatial learning and memory when location is used as the cue. In this study, we evaluated the performance of 16 Labrador retriever dogs with a cognitive bias test that used spatial location as the cue and compared the results to the dogs’ performance on a delayed nonmatching to position (DNMP) operant procedure. All dogs completed the cognitive bias test, whereas only 9 of 16 completed DNMP acquisition criteria within a maximum of 300 trials (mean ± SEM for successful dogs = 187.8 ± 19.2 trials). A negative correlation was observed between the number of trials required for DNMP acquisition and the mean latencies to the near positive (P = 0.001, r2 = 0.56, n = 16) and near negative (P = 0.013, r2 = 0.39, n = 16) ambiguous locations on the cognitive bias test. For the 9 dogs who met DNMP acquisition criteria within 300 trials, this correlation remained, with a significant negative correlation observed for the near positive (P = 0.01, r2 = 0.64) and near negative (P = 0.055, r2 = 0.43) ambiguous positions and the number of DNMP trials. The observed association between a classic test of spatial working memory (DNMP) and the cognitive bias test suggest that the cognitive bias test may also rely on spatial learning in some dogs. This finding suggests that the effect of spatial learning ability may need to be acknowledged in tests that use location as the discriminatory cue.