Group functional magnetic resonance imaging (fMRI) studies suggest that anxiety disorders are associated with anomalous brain activation and functional connectivity (FC). However, brain-based features sensitive enough to discriminate individual subjects with a specific anxiety disorder and that track symptom severity longitudinally, desirable qualities for putative disorder-specific biomarkers, remain to be identified. Blood oxygen level-dependent (BOLD) fMRI during emotional face perceptual tasks and a new, large-scale and condition-dependent FC and machine learning approach were used to identify features (pair-wise correlations) that discriminated patients with social anxiety disorder (SAD, N=16) from controls (N=19). We assessed whether these features discriminated SAD from panic disorder (PD, N=16), and SAD from controls in an independent replication sample that performed a similar task at baseline (N: SAD=15, controls=17) and following 8-weeks paroxetine treatment (N: SAD=12, untreated controls=7). High SAD vs HCs discrimination (area under the ROC curve, AUC, arithmetic mean of sensitivity and specificity) was achieved with two FC features during unattended neutral face perception (AUC=0.88, Ppre, t₍₁₁₎=2.9, P=0.007). In conclusion, SAD is associated with reduced FC between left temporal pole and left hippocampus during face perception, and results suggest promise for emerging FC-based biomarkers for SAD diagnosis and treatment effects.

It is currently unclear to what extent cortical structures are required for and engaged during subconscious processing of biologically salient affective stimuli (i.e. the 'low-road' vs. 'many-roads' hypotheses). Here we show that cortical-cortical and cortical-subcortical functional connectivity (FC) contain substantially more information, relative to subcortical-subcortical FC (i.e. 'subcortical alarm' and other limbic regions), that predicts subliminal fearful face processing within individuals using training data from separate subjects. A plot of classification accuracy vs. number of selected whole-brain FC features revealed 92% accuracy when learning was based on the top 8 features from each training set. The most informative FC was between right amygdala and precuneus, which increased during subliminal fear conditions, while left and right amygdala FC decreased, suggesting a bilateral decoupling of this key limbic region during processing of subliminal fear-related stimuli. Other informative FC included angular gyrus, middle temporal gyrus and cerebellum. These findings identify FC that decodes subliminally perceived, task-irrelevant affective stimuli, and suggest that cortical structures are actively engaged by and appear to be essential for subliminal fear processing.

Processing of unattended threat-related stimuli, such as fearful faces, has been previously examined using group functional magnetic resonance (fMRI) approaches. However, the identification of features of brain activity containing sufficient information to decode, or “brain-read”, unattended (implicit) fear perception remains an active research goal. Here we test the hypothesis that patterns of large-scale functional connectivity (FC) decode the emotional expression of implicitly perceived faces within single individuals using training data from separate subjects. fMRI and a blocked design were used to acquire BOLD signals during implicit (task-unrelated) presentation of fearful and neutral faces. A pattern classifier (linear kernel Support Vector Machine, or SVM) with linear filter feature selection used pair-wise FC as features to predict the emotional expression of implicitly presented faces. We plotted classification accuracy vs. number of top N selected features and observed that significantly higher than chance accuracies (between 90–100%) were achieved with 15–40 features. During fearful face presentation, the most informative and positively modulated FC was between angular gyrus and hippocampus, while the greatest overall contributing region was the thalamus, with positively modulated connections to bilateral middle temporal gyrus and insula. Other FCs that predicted fear included superior-occipital and parietal regions, cerebellum and prefrontal cortex. By comparison, patterns of spatial activity (as opposed to interactivity) were relatively uninformative in decoding implicit fear. These findings indicate that whole-brain patterns of interactivity are a sensitive and informative signature of unattended fearful emotion processing. At the same time, we demonstrate and propose a sensitive and exploratory approach for the identification of large-scale, condition-dependent FC. In contrast to model-based, group approaches, the current approach does not discount the multivariate, joint responses of multiple functional connections and is not hampered by signal loss and the need for multiple comparisons correction. Brain activity is increasingly characterized by patterns of pair-wise correlations (large-scale functional connectivity) across the whole brain obtained from Blood Oxygen Level Dependent (BOLD) functional magnetic resonance imaging (fMRI). Typically this is done during resting states (i.e. no presented stimulus) to differentiate subjects based on individual variation or diagnosis. In the current work, we identify such patterns that are a sensitive signature of unattended processing of threat-related stimuli, allowing one to “brain-read” whether an individual was presented with a neutral or fearful face while they attended to non-expression-related stimulus features. These results further the understanding of the neural mechanisms sub-serving threat-detection and facial affect processing in healthy subjects, and may also help further our understanding of various disorders, such as anxiety and autism, which exhibit anomalies in these processes. At the same time, we propose an exploratory and sensitive approach for the identification of condition-dependent, large-scale functional connectivity. This approach is not based on statistical inference on functional connections averaged across subjects and contrasted between two conditions, but rather based on the informative contribution of each functional connection when attempting to predict between two conditions, using machine-learning based multivariate pattern analysis on training data from separate subjects.

A recent report claims that functional brain networks defined with resting-state functional magnetic resonance imaging (fMRI) can be recapitulated with correlated gene expression (i.e. high within-network tissue-tissue “strength fraction”, SF) (1). However, the authors do not adequately control for spatial proximity. We replicated their main analysis, performed a more effective adjustment for spatial proximity, and tested whether “null networks” (i.e. clusters with center coordinates randomly placed throughout cortex) also exhibit high SF. Removing proximal tissue-tissue correlations by Euclidean distance, as opposed to removing correlations within arbitrary tissue labels as in (1), reduces within-network SF to no greater than null. Moreover, randomly placed clusters also have significantly high SF, indicating that high within-network SF is entirely attributable to proximity and is unrelated to functional brain networks defined by resting-state fMRI. We discuss why additional validations in the original article are invalid and/or misleading and suggest future directions.