Difficulties in the learning of same-different discrimination using multiple-item displays in pigeons (Columba livia)

The primary goal of the present study was to investigate how the location of changed items affected pigeons’ performance on same-different discrimination between two successively presented multiple-item displays using the same training procedure as Gibson et al. (2011). However, after encountering pigeons’ difficulty in learning the discrimination (Experiment 1), we shifted our focus to discrimination training itself. The poor performance during training and the resulting pattern in the simplified test suggested that entropy, or degree of heterogeneity, of each display may have been more salient to the pigeons than the abstract same-different relationship between them. In Experiment 2, after observing that replacing the stimuli with those used in Gibson et al. did not improve learning the discrimination, we switched to the training procedure based on Katz & Wright (2006). This procedure, in which two displays were presented simultaneously, may facilitate learning by allowing pigeons to compare overall entropy rather than relying solely on abstract relational concepts. In this revised task, one display was presented above the other, enabling perceptual comparison between the two displays. Although we observed slight accuracy improvement when single-item displays were used, it remained highly improbable that pigeons would achieve same-different discrimination between multiple-item displays under the current methodology. We discuss our task manipulations and the corresponding changes in pigeons’ performance, which may provide useful insights for developing robust training methods that enhance same-different discrimination learning in multiple-item displays.