Figure 3. ITD statistics predict discriminability of spatial deviants indexed by MMN responses.

(A) Left, passive oddball sequence protocol, in which subjects listened to frequent ‘standard’ stimuli embedded with rare spatial ‘deviants’. In each condition, two tones were presented with the same frequency and distinct ITDs. Right, MMN response within the 100–200 ms latency range of the deviant-minus-standard trace (black line) is shown for the midline frontal electrode (FZ) along with standard (green) and deviant (purple) event related potential traces, averaged across conditions and subjects. Inset on the bottom-right shows the topography of the MMN response. (B) Hypothesis (top) and null hypothesis (bottom) of an adapted neural code underlying MMN responses to spatial deviants tested in this study. Under a neural code relying on natural ITD statistics, the correlation between amplitude of MMN responses and difference between deviant and standard ITD is expected to show a synergistic effect of ITD statistics. (C) Left, coefficients of correlations between MMN amplitude and different predictor equations adjusting ITD difference between standard and deviant by ITD statistics, as a function of the relative weight of the standard stimulus (w_s), relative to the weight of the deviant (w_d). Middle, best prediction of MMN amplitude in the model relying on √FI_ITD, weighting standard more than deviant (80%:20%). Right panel, changes in MMN peak amplitude as a function of the difference between ITD of deviant and standard show stronger negative linear slopes for conditions where the weighted average of √FI_ITD was higher, compared to conditions with lower √FI_ITD.