Noise-invariant Neurons in the Avian Auditory Cortex: Hearing the Song in Noise
Vertical axis in A–C shows the noise invariance in the neural response. Each neuron (each point on the scatterplots) is represented by its STRF (0.25–8 kHz on the vertical axis, 0–60 ms on the horizontal). A. Invariance vs STRF Model Invariance. The solid line has slope 1.0, showing equal performance between the STRF model and the neural response. Neurons with significantly different performance (p<0.05, two-tailed t-test) have their receptive fields outlined. Dashed line shows regression fit (slope = 0.40, Adjusted R2 = 0.12, p = 0.034), indicating the positive correlation between the invariance predicted by the STRF and actual invariance. B. Invariance vs Spectral Modulation Tuning. Neurons sensitive to higher spectral modulations are more invariant (Adjusted R2 = 0.192, p = 0.007). C. Invariance vs Temporal Modulation Tuning. Neurons sensitive to lower temporal modulations are more invariant (Adjusted R2 = 0.15, p = 0.015). D. Ensemble modulation transfer functions for neurons grouped by invariance. Low invariance neurons (left panel, invariance<0.3, n = 11) respond to high temporal and low spectral frequency modulations. Neurons with moderate invariance (middle panel, 0.3<invariance<0.4, n = 11) transmit faster, sharper modulations. Neurons with high invariance (right panel, invariance>0.2, n = 10) respond mostly to slower and spectrally sharp sounds.