Multivariate pattern analysis detects distinct neural representations of fMRI-based lateralized neurofeedback for cognitive reappraisal training in major depressive disorder

Abstract

Neurofeedback (NF)-supported cognitive reappraisal training has shown promise for improving emotion regulation in psychiatric disorders. A previous real-time functional magnetic resonance imaging NF (rt-fMRI NF) study in patients with major depressive disorder (MDD) and healthy individuals revealed better self-regulation and more use of reappraisal strategies for left compared to right ventrolateral prefrontal cortex (vlPFC) modulation. However, univariate approaches failed to detect this lateralized effect on the whole-brain level. Multivariate pattern analysis (MVPA) may detect such effects by detecting spatially distributed neural representations beyond cluster-wise activation differences. Here, we applied MVPA to examine whether targeting different hemispheric regions (left vs. right vlPFC) leads to distinct neural representations and corresponding group differences. This secondary data analysis examined rt-fMRI NF data from 37 patients with MDD and 37 healthy controls. A whole-brain searchlight multivariate pattern classification (MVPC) decoded brain activity distinguishing left- and right-vlPFC NF. Representational similarity analysis (RSA)-based pattern consistency, defined as the ratio of mean similarity among pairwise multivoxel patterns within each condition to that of all conditions, was calculated to estimate neural encoding stability in 20 predefined emotion regulation regions. Searchlight MVPC revealed high classification accuracies in key fronto-limbic emotion regulation regions, e.g., inferior frontal gyrus (IFG), insula, and anterior cingulate cortex (ACC), as well as related cognitive and motor control area, such as supplementary motor area (SMA). Pattern consistency during left-vlPFC NF was significantly lower in frontal regions in the MDD compared to the control group. These results demonstrate that multivariate approaches can capture distributed neural differences during NF-guided emotion regulation. Searchlight findings link NF target laterality to the reconfiguration of local representational structure, which may be crucial for cognitive flexibility and executive function. Encoding instability in MDD during left-vlPFC NF may reflect underlying pathophysiology. MVPA offers a valuable tool for detecting novel neural evidence of complex cognitive and affective processes, extending the exploration of neural differences of lateralized NF-guided cognitive reappraisal from activation level to representational decoding.

Keywords: Cognitive reappraisal; Encoding stability; Lateralization; MDD; MVPA; Real-time fMRI neurofeedback.