Search

Learning to distinguish subtle differences in objects or events can impact how one generalizes. In some cases, training can cause novel events to appear more familiar or attractive than those actually experienced during training: the peak shift effect. This study examined whether individual differences in learning led to systematic patterns of generalization. Participants were trained to identify simulated birdsongs, and then tested on their ability to identify a target song presented among several similar songs that differed in pitch. Initial analysis showed that those attaining moderate proficiency at discriminating songs during training were more likely to shift than those performing poorly or proficiently. However, a neural network trained to output individuals' gradient dynamics using only performance during training as input found an additional set of training variables that predicted shift. Specifically, one subset of shifters had highly conservative response biases accompanied by very little change to perceptual sensitivity in training. These findings suggest that discrimination learning may only lead to generalization shifts in some individuals, and that all individuals who shift may not do so for the same reason., (Copyright © 2014 Elsevier B.V. All rights reserved.)

Humpback whales, unlike most mammalian species, learn new songs as adults. Populations of singers progressively and collectively change the sounds and patterns within their songs throughout their lives and across generations. In this study, humpback whale songs recorded in Hawaii from 1985 to 1995 were analyzed using self-organizing maps (SOMs) to classify the sounds within songs, and to identify sound patterns that were present across multiple years. These analyses supported the hypothesis that recurring, persistent patterns exist within whale songs, and that these patterns are defined at least in part by acoustic relationships between adjacent sounds within songs. Sound classification based on acoustic differences between adjacent sounds yielded patterns within songs that were more consistent from year to year than classifications based on the properties of single sounds. Maintenance of fixed ratios of acoustic modulation across sounds, despite large variations in individual sounds, suggests intrinsic constraints on how sounds change within songs. Such acoustically invariant cues may enable whales to recognize and assess variations in songs despite propagation-related distortion of individual sounds and yearly changes in songs., (Copyright © 2011 Elsevier B.V. All rights reserved.)

Ranging, the ability to judge the distance to a sound source, depends on the presence of predictable patterns of attenuation. We measured long-range sound propagation in coastal waters to assess whether humpback whales might use frequency degradation cues to range singing whales. Two types of neural networks, a multi-layer and a single-layer perceptron, were trained to classify recorded sounds by distance traveled based on their frequency content. The multi-layer network successfully classified received sounds, demonstrating that the distorting effects of underwater propagation on frequency content provide sufficient cues to estimate source distance. Normalizing received sounds with respect to ambient noise levels increased the accuracy of distance estimates by single-layer perceptrons, indicating that familiarity with background noise can potentially improve a listening whale's ability to range. To assess whether frequency patterns predictive of source distance were likely to be perceived by whales, recordings were pre-processed using a computational model of the humpback whale's peripheral auditory system. Although signals processed with this model contained less information than the original recordings, neural networks trained with these physiologically based representations estimated source distance more accurately, suggesting that listening whales should be able to range singers using distance-dependent changes in frequency content.