Skip to main content
. Author manuscript; available in PMC: 2021 Jan 20.
Published in final edited form as: Curr Opin Physiol. 2020 Sep 8;18:63–72. doi: 10.1016/j.cophys.2020.09.001

Figure 2.

Figure 2

The strength of perceptual contrast adaptation can be predicted from the contrast adaptation exhibited by auditory neurons. (a) Schematic of a computational model that uses neuronal responses recorded from neurons in the mouse auditory midbrain, thalamus or cortex to predict performance on a 2-alternative, forced-choice sound level discrimination task for pairs of broadband noise bursts presented in different contrast environments. Simulated responses to noise stimuli of different levels (reference, R: 70 dB SPL, target, T: 62–78 dB SPL), embedded in low- or high-contrast dynamic random chords, were derived from the contrast-dependent linear/non-linear model estimated from the individual neuronal responses recorded at each processing level. Psychometric functions were determined by asking which noise stimulus elicited most spikes across all recorded units in the simulated trial. If the reference noise elicited fewer spikes than the target noise stimulus a “louder” response was predicted. See Lohse et al. [31 •] for more details. (b) Psychometric functions produced by the model based on responses recorded in the primary auditory cortex (top) or in the midbrain from the central nucleus of the inferior colliculus (bottom) in low-contrast (20 dB, blue) and high-contrast (40 dB, red) conditions. Level discrimination improved when the contrast of the flanking sounds was low, as indicated by the steeper psychometric functions. (c) Predicted strength of contrast adaptation from neuronal responses recorded in the auditory midbrain of awake mice or in the midbrain, thalamus or cortex under anesthesia, compared with measured perceptual contrast adaptation in human listeners. Note the similarity in each case. Solid black lines connect mean values after 25 runs of the model (or across the eight participants in the measured human adaptation). Colored error bars denote 95% confidence intervals around the mean (for clarity, individual data points are displayed next to the corresponding error bars). Adapted from Lohse et al. [31 •].