Fig. 5: Phonetic errors are correlated with delayed, reduced θ–α STN-cortex coupling.

A Comparison of the spike-phase coupling (SPC) maps (N = 827, 74 neurons in 18 participants) between trials with and without phonetic errors (see “Methods”). In error trials (red), θ spike-phase coupling before speech production onset was significantly lower than in correct trials (green) (cluster-based permutation test). B t-SPC events occurrence in trials with and without phonetic errors grouped by frequency band (θ–α: left, β: right). Shaded areas illustrate the 5th and 95th percentiles of the bootstrapped distribution (1000 bootstraps) of t-SPC events occurrence. The thick line denotes the mean. Black bars denote time bins in which error t-SPC occurrence is different between correct and error trials (cluster-based permutation test). C Cumulative distribution of the θ–α t-SPC onset and offset (see Fig. 2C and “Methods”) in error and correct trials. D Comparison of the median θ–α t-SPC onset at the single-neuron level (N = 20/46 neurons in 13 participants). Black and magenta vertical dashed lines denote auditory cue (AC) and speech production (SP) windows. E Comparison of the median θ–α t-SPC duration at the single-neuron level before (N = 17/46 neurons in 13 participants) and after (N = 8/46 neurons) speech duration. Note that in (D, E) we only included neurons with significant θ–α t-SPC events in both accurate and error conditions. Two-sided permutation t-tests were used to compare t-SPC onset and duration within neurons.