Representing object categories by connections: Evidence from a mutivariate connectivity pattern classification approach

Abstract

The representation of object categories is a classical question in cognitive neuroscience and compelling evidence has identified specific brain regions showing preferential activation to categories of evolutionary significance. However, the potential contributions to category processing by tuning the connectivity patterns are largely unknown. Adopting a continuous multicategory paradigm, we obtained whole-brain functional connectivity (FC) patterns of each of four categories (faces, scenes, animals and tools) in healthy human adults and applied multivariate connectivity pattern classification analyses. We found that the whole-brain FC patterns made high-accuracy predictions of which category was being viewed. The decoding was successful even after the contributions of regions showing classical category-selective activations were excluded. We further identified the discriminative network for each category, which span way beyond the classical category-selective regions. Together, these results reveal novel mechanisms about how categorical information is represented in large-scale FC patterns, with general implications for the interactive nature of distributed brain areas underlying high-level cognition. Hum Brain Mapp 37:3685-3697, 2016. © 2016 Wiley Periodicals, Inc.

Multivariate Classification of Major Depressive Disorder Using the Effective Connectivity and Functional Connectivity

“…In our study, we found that the linear SVM classifier achieved the best performance in all classification process, which suggested that the linear SVM was indeed a good classifier for fMRI data. This finding was consistent with majority of previous studies (Craddock et al, 2009 ; Wang et al, 2016 ; Saccà et al, 2017 ). For example, using the linear SVM classifier, patients with depression were successfully distinguished from healthy volunteers (R. Cameron Craddock), One recent study achieved high accuracy in object categories classification task using functional connections from task-related functional neuroimaging as features and SVM as the classifier.…”

Emotion Representation in Whole-brain Functional Connectivity Patterns: An fMRI Study

“…Using multivariate pattern analysis and machine learning, we found that emotion information could be successfully decoded from the whole-brain FC patterns. Our results added to the recent FC-based decoding studies which found functional connectivity patterns could be used to discriminate different populations [14], task or mental states and different object categories [13], and further highlighted the effects of the whole-brain functional connectivity patterns in the emotion perception. Overall, our results provide new evidence that large-scale functional connectivity patterns also contain rich emotion information and effectively contribute to the recognition of emotions.…”

“…Then a component-based (CompCor) strategy was used to remove the non-neural confounders. The data were temporally filtered with a previously used for task-relate connectivity analysis 0.01 -0.1 HZ band-pass filter [13].We conducted ROI-to-ROI analysis to assess pairwise correlations between ROIs. Finally, we got 3 connection matrices (112 × 112) for each participant, one per emotion.…”

“…In accordance with the previous study [13], in this part, we only used positive FCs as inputs, which had values significantly higher than zero with one-sample t-test across participants and corrected with multiple comparisons with false discovery rate (FDR) q=0.01. We performed emotion-pairwise classification (joy VS anger, joy VS fear, and anger VS fear) with a leave one subject out cross-validation (LOOCV) scheme [13,14]. In each iteration, data of one subject was used as testing data while the data of the remaining subjects were used for training.…”

Self-esteem Modulates the P3 Component in Response to the Self-face Processing after Priming with Emotional Faces

“…CAFPS was established in a similar way to the International Affective Picture System except that the models were all Chinese. CAFPS has been widely used in previous research ( Duan et al, 2010 ; Luo et al, 2010 ; Zhang et al, 2015 ; Wang et al, 2016 ). There are four males and four females faces in each category.…”