The creative walrus: Novel sound production via contingency learning

Abstract

Walruses (Odobenus rosmarus) have a range of anatomical specializations that can provide plasticity to their vocal emissions. These include respiratory adaptations related to breath-holding and buoyancy and soft tissue adaptations of the mouth, lips, and tongue related to suction feeding. The aim of this study was to examine the extent to which contingency learning using food reinforcement could modify the sound production of walruses. One male and three female captive walruses participated in the current study. Animals were first trained in air in a series of four experimental steps: (1) induce and reinforce natural sound production, (2) establish stimulus control over sound production and inhibition, (3) establish stimulus control over different sound types by mapping them to different discriminative stimuli, and (4) encourage the production of novel sounds by withholding reinforcement for previously emitted sounds and reinforcing only novel variants. Subsequently, two of the walruses were transferred from airborne to underwater testing. In the absence of additional training, both individuals, when prompted to vocalize, emitted a range of underwater sounds, many of which were qualitatively different from those produced in air and those recorded in baseline conditions under water. Thus, walruses, like cetaceans, can learn to modify their sound production.