Neural Noise- Intracranial EEG can evaluate the temporal order of motor and auditory systems during speech production.1
- Motor–auditory ordering directly informs speech prostheses and decoding pipelines.1
- Temporal order of these systems matters for closed-loop BCI design.1 1
Gardner updates
- iEEG was used to evaluate the temporal order of motor and auditory systems in speech production; motor–auditory ordering matters for closed-loop speech prostheses and decoding pipelines (Nature). 1
Weekly enrichment (2026-07-20)
- The source study is Li, Chen, Luo et al., “Evaluating the temporal order of motor and auditory systems in speech production using intracranial EEG,” Communications Biology 8:1442 (published October 2025), from NYU Shanghai and Sanbo Brain Hospital, Capital Medical University (Beijing).2
- The cohort was 26 drug-resistant epilepsy patients (16 male, ages 14–46), all right-handed native Mandarin speakers, undergoing clinical stereo-EEG (sEEG) monitoring; a total of 2916 electrode contacts were implanted (8–16 contacts per electrode, 0.8 mm diameter, 2 mm length, 3.5 mm spacing), sampled at 512 Hz on a Nicolet system.2
- Participants read aloud 50 high-frequency single-character Chinese words (mean corpus frequency ~80 per million) presented visually, avoiding auditory-stimulus confounds; the average articulation reaction time was 826.8 ± 106.5 ms.2
- Contrary to the classic view that speech production is a temporal reversal of perception (auditory phonological code before motor encoding), high-gamma onset latencies showed motor IFG activating before auditory pSTG.2
- Population high-gamma onset latencies ran visual-to-auditory: vpIOG 71.3 ± 51.8 ms (n = 2), pFG 203.6 ± 48.6 ms (n = 26), IFG 368.6 ± 37.1 ms (n = 28), pSTG 617.4 ± 44.7 ms (n = 26), and HG 831.3 ± 42.0 ms (n = 18) — placing motor IFG ~250 ms ahead of auditory pSTG.2
- Critically, pSTG still activated before articulation onset (RT ≈ 826.8 ms), so an auditory phonological code is available before production even though the motor code appears to initiate first.2
- Representational similarity analysis (RSA) reproduced the latency ordering: both IFG and pSTG encode similar representations, but IFG’s is earlier, supporting a model in which the motor system begins programming speech codes before an auditory target is fully retrieved.2
- The authors interpret bidirectional IFG–pSTG interactions across gamma and high-gamma bands as consistent with predictive/internal forward-model accounts rather than a strict serial pipeline.2
- Stated limits temper BCI inference: Granger causality was single-participant, the stimulus set of ~50 Chinese sounds limits phonetic-feature modeling, sEEG coverage of some regions (e.g., visual vpIOG) was sparse, and seizure-onset-zone electrodes were excluded to reduce epilepsy confounds.3
- For speech neuroprosthetics and closed-loop decoding, the finding that motor-region activity leads auditory phonological activity argues for prioritizing frontal/motor signals in production-side BCI timing models.2