Abstract
Auditory fill-in (FI) is the perceptual completion of a sound that is interrupted by another sound. An illusion can be created when a discontinuity or a silent gap is introduced into a sound and a temporally coincident (interrupting) noise is superimposed on this gap. Under these conditions the sound is heard as continuing through the noise even though part of the sound was deleted. We hypothesized that the responses of A1 neurons to discontinuous tones interrupted by noise would be similar to the responses to pure tones (i.e., a correlate of FI). In contrast, we hypothesized that these neurons’ responses to a discontinuous tone presented with noise temporally surrounding the entire tone (masking noise) would be similar to responses to the noise by itself (i.e., a correlate of masking where one would only be able to hear the noise). Extracellular recordings were obtained from A1 neurons in 2 macaques. Most neurons could easily distinguish discontinuous from continuous tones according to signal detection theory based analyses. Although noise reduced the ability to detect the discontinuity (gap), neurons tended to be more sensitive in detecting the gap in the tone with masking than with interrupting noise, consistent with the psychophysical literature comparing masking and fill-in. Moreover, the response of many neurons to interrupting noise centered over the gap was quantitatively similar to responses to a complete tone, suggesting a neural correlate to FI. Conversely, neural responses to a gap in the tone with masking noise was similar to the response to the noise presented in isolation, as if being masked. We conclude that primate A1 is involved in the analysis of noisy auditory scenes and has responses that both represent the background noise and the illusory segment of the foreground tone.
