PB 13 Errors elicit high-gamma responses in the human cerebral cortex

Abstract

The nature of human error processing and the underlying cerebral mechanisms are of great interest in neuroscience, and error-related brain responses may provide information useful to improve the accuracy of brain-machine interfaces (BMIs). So far, many studies focused on error-related responses below 30 Hz. We were interested in error-related effects in the high-gamma band (HGB, 50–150 Hz), as this frequency range is thought to reflect local cortical information processing more directly than lower frequencies. 30 healthy subjects performed a flanker task as classically used to study error processing ( Gehring et al., 1993 ). Under time pressure, the subjects had to use the left or right index finger to respond to the respective flanker. Recordings of 128-channel EEG were acquired within an optimized setting for non-invasive EEG high-gamma mapping in an electromagnetically shielded cabin with full optical decoupling of all devices and high-resolution binocular eye tracking. Additionally, 9 patients with pharmacoresistant epilepsy and implanted stereo-EEG (SEEG) electrodes performed the identical task. As our main result, we show for the first time error-related HGB spectral power modulations up to 120 Hz in non-invasive EEG ( Fig. 1 A), which could also be used for decoding on a single-trial basis. Additional SEEG data ( Fig. 1 B) revealed possible sources of these effects, including the premotor cortex. Additional responses were seen in areas not reflected in the non-invasive EEG, such as hippocampus and insula. Importantly, based on the eye-tracking data, we show that the error-related effects in the gamma range cannot be explained by differences in eye movements including micro-saccades, which had a different spatial distribution compared to the error-related effects. Our findings open a new window for investigation of the complex neuronal processes related to error perception and ensuing behavioral adaptation. The similarity of intracranial and non-invasive measurements further show that data of healthy subjects and epilepsy patients is comparable and physiological high-gamma activity can be found in both cases.