Fig. 6.

Comparison between vocal suppression and compensation predictions. a Scatter plot comparing unit vocal responses (RMI) during normal and shifted feedback showing decreased suppression across units. Equal value line (unitary slope) is indicated. FB +2ST evoked an increase in activity compared to normally suppressed vocal responses (RMI change: 0.24 ± 0.31, mean ± std; p < 0.001, Wilcoxon signed-rank), as did FB −2ST (0.27 ± 0.37; p < 0.001). b Feedback effects plotted as mean Feedback-Normal RMI differences and binned by RMI. Suppressed units (RMI < −0.2) exhibited larger feedback increases than excited units (RMI > 0) which showed more balance effects. This trend of larger activity increases in suppressed units was present for both +2ST (r = −0.53, p < 0.001) and −2ST (r = −0.53,  p < 0.001) feedback. (Error bars: standard deviation) c Scatter plot comparing vocal correlation coefficients and vocal suppression, showing no obvious relationships (+2ST: r = −0.02, p = 0.65; −2ST r = 0, p = 0.97). d Mean vocal correlation coefficients binned by RMI, shown separately for all data (black) and for only units with significant correlations (orange). e, f There was also no obvious relationship between frequency tuning (CF) and feedback effects (e) or vocal correlation coefficients (f). g, h, Histograms comparing feedback effects (g) and vocal correlation coefficients (h) between hemispheres (RH: right, LH: left). The right hemisphere showed a slight increase in feedback sensitivity over the left (p = 0.02, Wilcoxon rank-sum), but there were no differences for vocal correlation (p = 0.84). Sample sizes and statistical details may be found in the Supplementary Note 1