A Physiologically-Based Population Rate Code for Interaural Time Differences (ITDs) Predicts Bandwidth-Dependent Lateralization

Abstract

Interaural time difference (ITDs) are the most important cue to the location of sounds containing low-frequency energy (Wightman and Kistler 1992). ITDs are encoded centrally in the medial (MSO) and lateral (LSO) superior olives which transmit the code to the inferior colliulus (IC) (Batra et al. 1997; Goldberg and Brown 1969). Each ITD-sensitive neuron is characterized by its best ITD (BD), the one producing maximal discharge rate. It is a longstanding view that these neurons are conceptually arranged in an array with best frequency (BF) on one axis and BD on the other to form a labeled-line code. According to this model, the stimulus ITD corresponds to the BD (i.e. the label) of the most active neuron in the array (Jeffress 1948). The labeled-line model is challenged by physiological data from guinea pig (confirmed in cat and gerbil) showing that the distribution of BD is highly dependent on BF, and in general does not correspond to the range of naturally-occurring ITDs (Brand et al. 2002; Hancock and Delgutte 2004; McAlpine et al. 2001). Instead, best interaural phase (BP = BD × BF) is more nearly independent of BF, such that the steepest slopes of neural rate-ITD curves tend to occur near the midline. Because the slopes, not the peaks, align near the midline (where perceptual ITD acuity is finest), it has been suggested that ITD is encoded by the discharge rate itself rather than by the locus of maximal activity (McAlpine et al. 2001). Thus, ITD may be represented by a population rate code, in which the activity of many neurons pool to form monolithic ITD channels on each side of the brain, and the stimulus ITD may be inferred by comparing the relative activity of the two channels (van Bergeijk 1962; von Bekesy 1960). Though the physiological data suggest the existence of a rate code, analysis of its viability has barely begun (Marquardt and McAlpine 2001). Here, we demonstrate that a population rate code model can account for the dependence of perceived laterality on stimulus bandwidth.