Back

Training Harbor Seals to Participate in Vocal Learning Experiments in a Zoo

Ravignani, A.; Duengen, D.

2024-08-27 animal behavior and cognition
10.1101/2024.08.27.609954 bioRxiv
Show abstract

Harbor seals (Phoca vitulina) are a common zoo species that show a scientifically valuable propensity for vocal learning. Under human care, the seals can be trained to associate vocalizations with cues. This ability is termed vocal usage learning and is characterized by learning to use a vocalization in a specific context. Among mammals, seals are prime candidates to investigate vocal learning. Yet, only a handful of reports exist on harbor seal vocal learning abilities, and even fewer document how these were trained or tested. Here, we investigate how, and if, two experimentally naive harbor seals under human care can be trained to participate in scientific experiments. We describe the training and testing of two seals in two basic vocal learning experiments. We trained the animals to vocalize upon the presentation of discriminative stimuli (SD) through operant conditioning methods and tested their abilities to i) vocalize and refrain from vocalizing on two distinct SDs, and ii) produce two different vocalizations upon the presentation of two different SDs. Both seals learned the tasks: the first task was achieved within 118 trials (22 errors to criterion) and 220 trials (40 errors to criterion), the second task within 480 trials (158 errors to criterion) and 380 trials (94 errors to criterion), respectively. Our results confirm that harbor seals are capable of vocal usage learning and further suggest that associating individually distinct vocalizations with different SDs may be more cognitively demanding than vocalizing and being silent on SD.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.