Dynamics of error-related activity in deterministic learning - an EEG and fMRI study

The Role of the Human Cerebellum for Learning from and Processing of External Feedback in Non-Motor Learning: A Systematic Review

This review aimed to systematically identify and comprehensively review the role of the cerebellum in performance monitoring, focusing on learning from and on processing of external feedback in non-motor learning. While 1078 articles were screened for eligibility, ultimately 36 studies were included in which external feedback was delivered in cognitive tasks and which referenced the cerebellum. These included studies in patient populations with cerebellar damage and studies in healthy subjects applying neuroimaging. Learning performance in patients with different cerebellar diseases was heterogeneous, with only about half of all patients showing alterations. One patient study using EEG demonstrated that damage to the cerebellum was associated with altered neural processing of external feedback. Studies assessing brain activity with task-based fMRI or PET and one resting-state functional imaging study that investigated connectivity changes following feedback-based learning in healthy participants revealed involvement particularly of lateral and posterior cerebellar regions in processing of and learning from external feedback. Cerebellar involvement was found at different stages, e.g., during feedback anticipation and following the onset of the feedback stimuli, substantiating the cerebellum's relevance for different aspects of performance monitoring such as feedback prediction. Future research will need to further elucidate precisely how, where, and when the cerebellum modulates the prediction and processing of external feedback information, which cerebellar subregions are particularly relevant, and to what extent cerebellar diseases alter these processes.

Slip or fallacy? Effects of error severity on own and observed pitch error processing in pianists

Errors elicit a negative, mediofrontal, event-related potential (ERP), for both own errors (error-related negativity; ERN) and observed errors (here referred to as observer mediofrontal negativity; oMN). It is unclear, however, if the action-monitoring system codes action valence as an all-or-nothing phenomenon or if the system differentiates between errors of different severity. We investigated this question by recording electroencephalography (EEG) data of pianists playing themselves (Experiment 1) or watching others playing (Experiment 2). Piano pieces designed to elicit large errors were used. While active participants' ERN amplitudes differed between small and large errors, observers' oMN amplitudes did not. The different pattern in the two groups of participants was confirmed in an exploratory analysis comparing ERN and oMN directly. We suspect that both prediction and action mismatches can be coded in action monitoring systems, depending on the task, and a need-to-adapt signal is sent whenever mismatches happen to indicate the magnitude of the needed adaptation.

Dorsal posterior cingulate cortex responds to negative feedback information supporting learning and relearning of response policies

Many challenges in life come without explicit instructions. Instead, humans need to test, select, and adapt their behavioral responses based on feedback from the environment. While reward-centric accounts of feedback processing primarily stress the reinforcing aspect of positive feedback, feedback’s central function from an information-processing perspective is to offer an opportunity to correct errors, thus putting a greater emphasis on the informational content of negative feedback. Independent of its potential rewarding value, the informational value of performance feedback has recently been suggested to be neurophysiologically encoded in the dorsal portion of the posterior cingulate cortex (dPCC). To further test this association, we investigated multidimensional categorization and reversal learning by comparing negative and positive feedback in an event-related functional magnetic resonance imaging experiment. Negative feedback, compared with positive feedback, increased activation in the dPCC as well as in brain regions typically involved in error processing. Only in the dPCC, subarea d23, this effect was significantly enhanced in relearning, where negative feedback signaled the need to shift away from a previously established response policy. Together with previous findings, this result contributes to a more fine-grained functional parcellation of PCC subregions and supports the dPCC’s involvement in the adaptation to behaviorally relevant information from the environment.

Measuring Task-Related Brain Activity With Event-Related Potentials in Dynamic Task Scenario With Immersive Virtual Reality Environment

Measurement of event-related potentials (ERPs) in simulated and real environments is advantageous for understanding cognition and behavior during practice of goal-directed activities. Recently, instead of using task-irrelevant “probe stimuli” to elicit ERPs, extraction of ERPs directly from events that occur in simulated and real environments has drawn increased attention. Among the previous ERP studies using immersive virtual reality, only a few cases elicited ERPs from task-related events in dynamic task settings. Furthermore, as far as we surveyed, there were no studies that examined the source of ERPs or correlation between ERPs and behavioral performance in 360-degree immersive virtual reality using head-mounted display. In this study, EEG signals were recorded from 16 participants while they were playing the first-person shooter game with immersive virtual reality environment. Error related negativity (ERN) and correct-(response)-related negativity (CRN) elicited by shooting-related events were successfully extracted. We found the ERN amplitudes to be correlated with the individual shooting performance. Interestingly, the main source of the ERN was the rostral anterior cingulate cortex (ACC), which is different from previous studies where the signal source was often estimated to be the more caudal part of ACC. The obtained results are expected to contribute to the evaluation of cognitive functions and behavioral performance by ERPs in a simulated environment.

Microstates-based resting frontal alpha asymmetry approach for understanding affect and approach/withdrawal behavior

The role of resting frontal alpha-asymmetry in explaining neural-mechanisms of affect and approach/withdrawal behavior is still debatable. The present study explores the ability of the quasi-stable resting EEG asymmetry information and the associated neurovascular synchronization/desynchronization in bringing more insight into the understanding of neural-mechanisms of affect and approach/withdrawal behavior. For this purpose, a novel frontal alpha-asymmetry based on microstates, that assess quasi-stable EEG scalp topography information, is proposed and compared against standard frontal-asymmetry. Both proposed and standard frontal alpha-asymmetries were estimated from thirty-nine healthy volunteers resting-EEG simultaneously acquired with resting-fMRI. Further, neurovascular mechanisms of these asymmetry measures were estimated through EEG-informed fMRI. Subsequently, the Hemodynamic Lateralization Index (HLI) of the neural-underpinnings of both asymmetry measures was assessed. Finally, the robust correlation of both asymmetry-measures and their HLI’s with PANAS, BIS/BAS was carried out. The standard resting frontal-asymmetry and its HLI yielded no significant correlation with any psychological-measures. However, the microstate resting frontal-asymmetry correlated significantly with negative affect and its neural underpinning’s HLI significantly correlated with Positive/Negative affect and BIS/BAS measures. Finally, alpha-BOLD desynchronization was observed in neural-underpinning whose HLI correlated significantly with negative affect and BIS. Hence, the proposed resting microstate-frontal asymmetry better assesses the neural-mechanisms of affect, approach/withdrawal behavior.

A tradeoff relationship between internal monitoring and external feedback during the dynamic process of reinforcement learning

Effective behavior monitoring, including internal monitoring/error detection and external monitoring/feedback, is very pivotal for reinforcement learning. However, less attention has been paid to internal monitoring and the dynamic learning performance in reinforcement learning, and there is still a heated debate on which kind of external feedback is relied on in the reinforcement learning. In order to address these questions, an adaption probabilistic selection task was used to examine the effect of the internal monitoring, external feedback and the relationship between them for approach learners and avoidance learners during dynamic learning process of reinforcement learning and behavior adaption. Error-related negativity (ERN), feedback-related negativity (FRN) and feedback-related P300 are three ERPs components, which can be used as the indexes of internal monitoring, external feedback and behavior adaption. For our results, the ERN effect of avoidance learners become large in block 3, which is earlier than approach learners (block 4). This phenomenon suggests that avoidance learners learned faster than approach learners. In addition, the FRN amplitude of avoidance learners in block 4 was significantly smaller than the other three blocks. The aforementioned results demonstrated a tradeoff relationship between the ERN and FRN effects.

A condition-independent framework for the classification of error-related brain activity

The cognitive processing and detection of errors is important in the adaptation of the behavioral and learning processes. This brain activity is often reflected as distinct patterns of event-related potentials (ERPs) that can be employed in the detection and interpretation of the cerebral responses to erroneous stimuli. However, high-accuracy cross-condition classification is challenging due to the significant variations of the error-related ERP components (ErrPs) between complexity conditions, thus hindering the development of error recognition systems. In this study, we employed support vector machines (SVM) classification methods, based on waveform characteristics of ErrPs from different time windows, to detect correct and incorrect responses in an audio identification task with two conditions of different complexity. Since the performance of the classifiers usually depends on the salience of the features employed, a combination of the sequential forward floating feature selection (SFFS) and sequential forward feature selection (SFS) methods was implemented to detect condition-independent and condition-specific feature subsets. Our framework achieved high accuracy using a small subset of the available features both for cross- and within-condition classification, hence supporting the notion that machine learning techniques can detect hidden patterns of ErrP-based features, irrespective of task complexity while additionally elucidating complexity-related error processing variations. Graphical abstract A schematic of the proposed approach. (a) EEG recordings in an auditory experiment in two conditions of different complexity. (b) Characteristic event related activity feature extraction. (c) Selection of feature vector subsets for easy and hard conditions corresponding to correct (Class1) and incorrect (Class2) responses. (d) Performance for individual and cross-condition classification.