Perceptual salience of sounds depends on the acoustic context in which they appear. Neural correlates of contextual sensitivity can be investigated efficiently by analysing the responses of neurons to complex sounds using nonlinear-linear “context models” (Williamson et al. 2016 Neuron). Context models capture the spectrotemporal sensitivity of a neuron in a principal receptive field (PRF) and its sensitivity to local acoustic context in a contextual gain field (CGF). The PRF is analogous to a classical receptive field estimate; it describes how the firing rate of a neuron depends on particular features of the input stimulus (e.g., tone frequency and time of tone occurrence). The CGF, in contrast, describes how the gain of the neuron’s response to a particular input depends on the context (e.g., on other sounds that occurred recently or simultaneously). Thus, this input-specific gain modulation captured in the CGF is essentially a neuronal measure of contextual sensitivity.
In this talk I will summarize both published and recent unpublished results from analysis of cortical and thalamic responses to complex sounds using context models. Previous work revealed strong input-specific gain modulation in the auditory cortex and thalamus of anaesthetized mice. CGF structure varied from neuron to neuron, but generally featured suppression of response gain by preceding sounds at similar frequencies (nonlinear forward suppression) and enhancement of response gain by simultaneous sounds at different frequencies (nonlinear off-frequency facilitation). In more recent work, we have investigated the stability of CGF structure in awake animals, using spike waveform-matching techniques to track single neurons across multiple days of recording. We find that in awake as well as in anaesthetised mice, CGFs are robust features of neuronal responses to complex sounds. Notably, CGF structure in individual neurons remains remarkably consistent across many days of recording. We conclude that contextual sensitivity is an essential and stable feature of the neural code in auditory cortex of awake animals.