“Straightness” versus “briefness” in binaural cue extraction

Abstract

The neural mechanisms that detect and encode interaural delays have elicited intense interest for many decades, from Jeffress’s [J. Comput. Physiol. Psychol. 41, 35–39 (1948)] original “place theory of sound localization,” its modeling via frequency-specific cross-correlation starting with Colburn [JASA 54, 1458–1470 (1973)], and continuing to this day. An important advance was the emphasis of cross-correlation peaks that align across frequency [“straightness” ; Stern et al., JASA 84, 156–165 (1988)]. While narrowband cross-correlation produces multiple peaks and is therefore inherently ambiguous, straightness detection accounts for resolution of that ambiguity in broadband stimuli by combining binaural outputs across frequency. An alternative account is suggested by the effects of temporal fluctuations, such as onsets, on the inputs to binaural processing. Zurek [JASA 67, 953–964 (1980)] described how emphasizing such events (e.g., by “windowing” the input) also resolves interaural ambiguities. Recent psychophysical and physiological evidence supports that view, strongly suggesting that binaural processing does, in fact, occur in brief windows triggered by envelope fluctuations such as onsets and intrinsic fluctuations in bands of noise. This talk investigates the possibility that the resulting temporal sparsity of binaural inputs (“briefness” ) might account for bandwidth effects even within single frequency channels, i.e. without a second level of “straightness” detection.