From the Journals

‘Error neuron’ EEG findings could open up future clinical applications



Single neurons in the human medial frontal cortex appear to be involved in the signaling of self-monitored errors, and this activity can be tracked through a scalp EEG pattern called error-related negativity, according to findings from experiments carried out during intracranial EEG recordings of candidates for surgical treatment of epilepsy.

An illustration of the brain Epifantsev/Thinkstock

“Our results suggest that coordinated neural activity can serve as a substrate for information routing that enables the performance-monitoring system to communicate the need for behavioral control to other brain regions, including those that maintain flexible goal information, such as the lateral prefrontal cortex and the frontal polar cortex,” first author Zhongzheng Fu, a PhD student at the California Institute of Technology in Pasadena, Calif., and Cedars-Sinai Medical Center, Los Angeles, and his colleagues reported in Neuron.

The findings offer insights that could lead to treatments for conditions in which the important executive function task of error self-monitoring is unbalanced, such as obsessive-compulsive disorder and schizophrenia, the authors noted in a press release.

“We discovered that the activity of error neurons correlates with the size of the ERN [error-related negativity],” Mr. Fu said. “This identifies the brain area that causes the ERN and helps explain what it signifies. This new insight might allow doctors to use the ERN as a standard tool to diagnose mental diseases and monitor responses to treatment.”

Error neuron firing and intracranial ERN occurred first in pre-supplementary motor area (pre-SMA), then in the dorsal anterior cingulate cortex (dACC) about 50 ms later, with significant correlations between firing and intracranial ERN in both locations. In dACC, this activity, with error-integrating neuron responses, correlated with magnitude of post-error slowing (PES).

Previous research suggested a link between “the detection of self-generated errors, as reflected in the ERN, with changes in cognitive control, as exhibited behaviorally in PES,” the investigators wrote. “However, several electroencephalogram (EEG) studies have failed to find a significant relationship between PES and ERN.”

The present study involved intracranial EEG of 29 candidates for surgical treatment of epilepsy and scalp EEG of 12 control participants, with each modality measuring activity in the frontal cortex. Both cohorts performed a rapid version of the color-word Stroop task, in which the words “red,” “green,” or “blue” were printed either in corresponding or noncorresponding colors of red, green, or blue. Subjects were presented various color-word combinations while being asked to click one of three buttons indicating the color of the word as quickly as possible. The investigators monitored neuronal activity throughout, discarding responses that were too slow.

As found in previous trials, the subjects demonstrated the “Stroop effect,” which refers to a slower response when word and color are incongruent (224.9 ms difference; P less than .001). As anticipated, correct responses following correct responses were faster than were correct responses following erroneous responses, which defines PES.

In the intracranial EEG group, the investigators isolated 1,171 neurons, of which 618 were located in dACC and 553 in pre-SMA. Using a Poisson regression model and correlations with erroneous responses, the investigators identified 99 “type I” error neurons in dACC and 118 in pre-SMA, based on higher frequency of firing during erroneous responses than during correct responses. At a single-cell level, error neuron mean spike rates were highest when intracranial ERN amplitude was greatest, such that error neuron firing in dACC and pre-SMA had maximal likelihood ratios of 7.9 (P = .01) and 15.1 (P less than .001), respectively. The strength of correlation between intracranial ERN and error neuron firing rate was directly related to PES magnitude exclusively in the dACC (maximum likelihood ratio of 13.9; P = .015). In post-error trials, faster error-integrating neuron firing rates in dACC predicted greater PES (maximal likelihood ratio of 18.3; P less than .001).

The study was funded by the National Institutes of Health, the McKnight Endowment for Neuroscience, and the National Science Foundation. The authors declared no conflicts of interest.

SOURCE: Fu Z et al. Neuron. 2018 Dec 4. doi: 10.1016/j.neuron.2018.11.016

Next Article: