Multiplexed encoding of frequency-modulated sweep features in the inferior colliculus

Abstract

Within the central auditory pathway, the inferior colliculus (IC) is a critical integration center for ascending sound information. While IC neurons have well-characterized receptive fields for individual sound features such as sound frequency, intensity, and location, growing evidence suggests that some neurons also use multiplexing to encode sound feature combinations. Here, we performed in vivo juxtacellular recordings in awake, head-fixed mice to examine how individual IC neurons and neuronal populations encode the speed, direction, and frequency range of frequency-modulated sweeps. To understand the strategies used by neurons to represent different sound features, we trained a support vector machine to decode sound features from different parameters of the spike train, including the firing rate, spike times relative to stimulus onset, distribution of inter-spike intervals, and first spike latency. We found that many IC neurons multiplex features of frequency-modulated (FM) sweeps using distinct temporal coding strategies rather than simple changes in mean firing rate, and that these feature representations are interdependent, yielding a combinatorial encoding of sound features within individual neurons. Accordingly, using static receptive fields for sweep frequency or direction alone yielded poor predictions of neuron responses to vocalizations that contain simple frequency changes. Lastly, we showed that encoding strategies varied across individual neurons, resulting in a highly informative population code for FM sweep parameters. Together, our results suggest that multiplexing is a common mechanism used by IC neurons to represent complex sound features.