Proceedings of The Physiological Society
University of Oxford (2011) Proc Physiol Soc 23, PC52
Behavioural and Neurophysiological Thresholds for Two Pitch Discrimination Tasks
A. Davies1, K. M. Walker1, J. K. Bizley1, A. J. King1, J. W. Schnupp1
1. Physiology, Anatomy & Genetics, University of Oxford, Oxford, United Kingdom.
Pitch thresholds (mean + standard deviation) of ferrets (white bars; n = 3 and n = 5) and humans (black bars; n = 5) on a 2-alternative forced choice (2AFC; left bars) and Go/No-Go (GNG; right bars) pitch discrimination task.
Histograms of the “neurometric” pitch discrimination thresholds of populations of cortical neurons (n = 47) on a pitch direction judgment (solid line) and pitch change detection (dashed line) task.
Pitch perception is an essential part of vocal communication in many species. Appropriate animal models of human pitch perception are needed in order to examine pitch processing at the cellular level. Human pitch thresholds have often been compared to those of other mammals, but not on the same tasks. In animals, pitch thresholds are measured on a Go/No-Go (GNG) task in which the animal must detect a pitch change in a sound sequence. Human pitch thresholds, on the other hand, are measured on 2-Alternative Forced Choice (2AFC) tasks, which usually require subjects to label a pitch shift across 2 sounds as “increasing” or “decreasing”. Here, we compared the pitch thresholds of 5 ferrets and 5 human listeners on a GNG task with thresholds from the same 5 humans and 3 other ferrets on a 2AFC task. Stimuli were artificial vowels, with a 400-Hz reference. We also examined how the responses of ferret auditory cortical neurons to the same sounds may support performance on both these tasks. Cortical recordings were carried out in 5 ferrets under general anaesthesia (0.022 mg/kg/h i.v. medetomidine, with 5 mg/kg/h i.v. ketamine). P-values for Wilcoxon Rank Sum Tests are reported. The discrimination thresholds of ferrets were poorer on the 2AFC task than the GNG task (p = 0.036), while humans performed similarly on the two tasks (p = 0.063, Figure 1). Humans outperformed ferrets on the 2AFC (p = 0.036) task, but performance on the GNG task did not differ across species (p = 0.056). We designed neurometric algorithms that used the recorded responses of auditory cortical neurons to either (a) detect a change in the pitch of two artificial vowels, or (b) determine the direction of a pitch shift across two vowels. We found that among the neural populations (n = 47) which could perform the pitch discrimination tasks to criterion, pitch thresholds did not differ across the two tasks (p = 0.115, Figure 2). Human listeners, as well as ferret auditory cortical neurons, provided similar thresholds across the two pitch discrimination tasks. These results suggest that the large difference in ferrets’ performance on the GNG and 2AFC task is mainly due to differences in the tasks’ cognitive demands, rather than their perceptual demands. This study emphasizes the importance of task design in sensory neuroscience, in both estimating perceptual thresholds and in interpreting neurophysiological responses.
Where applicable, experiments conform with Society ethical requirements