Abstracts

Network Dynamics Supporting Rapid Visuomotor Processing and Error Detection

Abstract number : 2.108
Submission category : 3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year : 2023
Submission ID : 594
Source : www.aesnet.org
Presentation date : 12/3/2023 12:00:00 AM
Published date :

Authors :
Presenting Author: Riyo Ueda, MD, PhD, – Wayne State University

Kazuki Sakakura, MD, PhD, – Wayne State University; Yu Kitazawa, MD, PhD, – Wayne State University; Yongje Jeon, Medical student – Yokohama City University; Yoshio Shiimoto, Medical student – Yokohama City University; Naoto Kuroda, MD – Wayne State University; Masaki Sonoda, MD, PhD, – Yokohama City University; Aimee Luat, MD, PhD, – Wayne State University; Eishi Asano, MD, PhD, – Wayne State University

Rationale:
Humans have the capability to quickly process visual information and adjust to tasks that demand frequent switches. Alongside this, humans possess an ability to identify errors and take a cautious approach to prevent repeated mistakes in subsequent attempts. Since this critical error detection functionality should be preserved after epilepsy surgery, we aim to identify and characterize the neural networks that underlie rapid visuo-motor processing and error detection.

Methods: We studied eight right-handed patients, aged 8 to 20 years, who had a diagnosis of drug-resistant focal epilepsy. During extraoperative intracranial EEG monitoring, patients participated in a cognitive training game, "Speed Match," offered by Lumosity on an iPad. Each participant completed five game sessions, each lasting 45 seconds. During the game, one of five symbols was randomly presented, and each patient had to tap a "Yes button" if the incoming symbol matched the previous one, and a "No button" otherwise. We measured the high-gamma amplitude (70-110 Hz) in alignment with the onset of stimulus or response. Mixed model analysis determined the independent effects of various factors on high-gamma amplitude. These factors encompassed [1] a ‘No button’ trial, [2] a task switch trial, [3] a trial immediately preceded by an erroneous response, and [4] the trial number itself. We also evaluated the influence of high-gamma amplitude on response time and accuracy.

Results:
The average number of trials completed per patient was 184, ranging from 117 to 232. An error in the preceding trial substantially influenced response time and accuracy. The average overall response time was 1,025 ms, with trials preceded by an error showing longer response times averaging 1,385 ms. The overall correct response rate was 88.0%, with a drop to 77.2% in trials following an error. Mixed model analysis revealed that trials preceded by an error exhibited sustained high-gamma enhancement in extensive, right-hemispheric dominant networks, encompassing the superior frontal, supramarginal, and posterior middle frontal gyri. Enhanced high-gamma activity in the right superior frontal (maximum mixed model t-value: +3.54 at -350 to -150 ms pre-stimulus onset) and left fusiform (maximum t-value: +4.82 at -300 to -100 ms pre-stimulus onset) and left medial temporal (maximum t-value: +5.35 at -300 to -100 ms pre-stimulus onset) regions was associated with increased response time. These effects were independent of the influences of the "No button," task switch, and trial number. Yet, logistic mixed model analysis failed to demonstrate that enhanced high-gamma activity in right superior frontal, left fusiform or left medial temporal regions was associated with improved response accuracy (p-values >0.05).

Conclusions:
Enhanced high-gamma activity immediately following erroneous responses, observed within the right-hemispheric dominant network, specifically involving the right superior frontal gyrus, may reflect the processes related to error detection.

Funding:
NIH R01 NS064033 (to E.A.), KAKENHI Grant JP23KJ2197 (to R.U.), KAKENHI Grant JP 22J23281 (to N.K.)



Neurophysiology