The study of dolphin cognition involves intensive research of animal vocal- izations recorded in the field. In this dissertation I address the automated analysis of audible dolphin communication. I propose a system called the signal imager that automatically discovers patterns in dolphin signals. These patterns are invariant to frequency shifts and time warping transformations. The discovery algorithm is based on feature learning and unsupervised time series segmentation using hidden Markov models. Researchers can inspect the patterns visually and interactively run com- parative statistics between the distribution of dolphin signals in different behavioral contexts. The required statistics for the comparison describe dolphin communication as a combination of the following models: a bag-of-words model, an n-gram model and an algorithm to learn a set of regular expressions. Furthermore, the system can use the patterns to automatically tag dolphin signals with behavior annotations. My results indicate that the signal imager provides meaningful patterns to the marine biologist and that the comparative statistics are aligned with the biologists’ domain knowledge.

Dolphins use stereotyped, individually distinctive, frequency modulated whistles, referred to as signature whistles, in order to broadcast their identity. In this study, we trained six dolphins at Kolmården Zoo, Sweden, to be called over, either upon hearing their own signature whistle (SW) or upon hearing a biologically irrelevant ”trivial” sound (TS), with the aim to prove the Whistle caller concept. The Whistle caller concept is based on the fact that dolphins occasionally use other dolphins’ signature whistles in order to address specific group members and convene.  Our hypotheses were that (1) dolphins call-over trained using their SW would learn the behaviour faster than dolphins trained using TSs, and (2) dolphins trained with their SW would be able to discriminate between different SWs better than dolphins trained with a TS would be at discriminating between different TSs.  Three out of three dolphins were successfully call-over trained using their SW, and two out of three dolphins using their assigned TS. When discriminating between different sounds, two of the dolphins trained using their SW performed significantly better than one of the dolphins trained using a TS. However, there were large intra-group differences in the results, indicating that we cannot eliminate the possibility that these results stem from individual differences in these dolphins’ ability to learn new behaviours overall, rather than an understanding of the sounds they heard. We suggest that future studies focus on (1) male-female differences in discrimination success when applying the Whistle caller concept, (2) how the characteristics of the trivial sounds affect discrimination success, and (3) the option of calling more than one animal at a time by sending out several SWs in succession.