Perceptual organization of sound begins in the auditory periphery

Abstract

Segmenting the complex acoustic mixture that makes a typical auditory scene into relevant perceptual objects is one of the main challenges of the auditory system [1], for both human and nonhuman species. Several recent studies indicate that perceptual auditory object formation, or "streaming," may be based on neural activity within the auditory cortex and beyond [2, 3]. Here, we find that scene analysis starts much earlier in the auditory pathways. Single units were recorded from a peripheral structure of the mammalian auditory brainstem, the cochlear nucleus. Peripheral responses were similar to cortical responses and displayed all of the functional properties required for streaming, including multisecond adaptation. Behavioral streaming was also measured in human listeners. Neurometric functions derived from the peripheral responses predicted accurately behavioral streaming. This reveals that subcortical structures may already contribute to the analysis of auditory scenes. This finding is consistent with the observation that species lacking a neocortex can still achieve and benefit from behavioral streaming [4]. For humans, we argue that auditory scene analysis of complex scenes is probably based on interactions between subcortical and cortical neural processes, with the relative contribution of each stage depending on the nature of the acoustic cues forming the streams.

Figure 1. Illustration of the sound sequences and cochlear nucleus single-unit responses

(A) The frequency difference, Δf, between A and B tones is 1 semitone. Sequences of ABA- tones were presented for 10s, and the frequency of the A tone was chosen equal to the unit’s best frequency. The neuron displayed here was classified as a multipolar cell with a transient-chopper response and a best frequency of 2.63 kHz (see Supplemental Experimental Procedures). The post-stimulus time histograms (binwidth: 5ms) show that the unit responded to both A and B tones. In this case, listeners tend to hear a single stream. (B) As (a), but the frequency difference is now 6 semitones. The responses to the B tones are reduced because of frequency selectivity and forward suppression. In this case, listeners have a probability of hearing 2 streams that increases over the duration of the sequence.

Figure 2. Multi-second adaptation is present in the cochlear nucleus

(A) Firing rates are displayed for each tone of the triplets, as a function of the time within the sequence. Left, middle and right panels show responses to the first A tone, B tone, and second A tone of the triplets, resp. Single-unit firing rates were averaged for all of the Multipolar cells of our sample (chopper-transient and chopper-sustained response types). Error bars represent +−1 standard error around the mean. Each line represents a single frequency difference, Δf, as identified in the figure legend. Multi-second adaptation is observed for all tones and all Δfs. (B) Same as (A), for bushy cells (primary-like response types).

Figure 3. Responses from the cochlear nucleus predict the behavioural build-up of streaming

(A) Neurometric (solid lines) functions for the Multipolar cells subpopulation, and psychometric functions in human listeners (dashed lines), for the Δfs used in the experiment. Neurometric functions were estimated by a “grouping by co-activation” model, which predicts a one-stream percept if A and B tones recruit a same neuronal population, and a two-stream percept otherwise (see Supplemental Experimental Procedures). Psychometric functions were obtained from normal hearing human listeners. Error bars show 95% confidence intervals around the mean. There is a good correspondence between the two, neurometric functions are within the confidence intervals for the psychometric functions. (B) Same as (A), but for the Bushy cells subpopulation.