Online self-evaluation of fMRI-based neurofeedback performance

Abstract

This study explores the subjective evaluation of supplementary motor area (SMA) regulation performance in a real-time functional magnetic resonance imaging neurofeedback (fMRI-NF) task. In fMRI-NF, people learn how to self-regulate their brain activity by performing mental actions to achieve a certain target level of blood-oxygen-level-dependent (BOLD) activation. This setup offers the possibility to study performance monitoring in the absence of somatosensory feedback. Here, we studied two types of self-evaluation expressed before receiving neurofeedback: performance predictions and perceived confidence in the prediction judgement. We hypothesized that throughout learning, participants would (1) improve the precision of their performance predictions about the actual changes in their BOLD response, and (2) that reported confidence would progressively increase with improved metacognitive precision. Participants completed three sessions of SMA regulation in a 7T fMRI scanner, performing a drawing motor imagery task. During each trial, they modulated their mental drawing strategy to achieve one of two different levels of target fMRI signal change. They then reported a performance prediction and their confidence in the prediction before receiving delayed BOLD-activation feedback. Results show that participants' performance predictions improved with learning throughout the three sessions, and that these improvements were not driven exclusively by their knowledge of previous performance. Confidence reports on the other hand showed no change throughout training and did not differentiate between the better and worse predictions. In addition to shedding light on mechanisms of internal monitoring during neurofeedback training, these results also point to a dissociation between self-evaluation of performance and corresponding reported confidence in the presence of feedback.

Working out the brain

“…Although much attention has been given to the success of self-regulation and its benefits, the underlying mechanisms of neurofeedback training and self-regulation success have been researched to a lesser extent (for an overview see ). Recent metaanalyses for example determined a network related to self-regulation and examined whether training success can be predicted based on pretraining (e.g., localizer) brain activation ; other studies explored which brain regions are related to the accuracy of regulation (Skottnik et al, 2019), how participants self-monitor during regulation (Muñoz-Moldes et al, 2021), and what are the effects of different strategies on regulation success (Schulz et al, 2019).…”