Pattern analysis of neuroimaging data reveals novel insights on threat learning and extinction in humans

Abstract

Several decades of rodent neurobiology research have identified a network of brain regions that support Pavlovian threat conditioning and extinction, focused predominately on the amygdala, hippocampus, and medial prefrontal cortex (mPFC). Surprisingly, functional magnetic resonance imaging (fMRI) studies have shown inconsistent evidence for these regions while humans undergo threat conditioning and extinction. In this review, we suggest that translational neuroimaging efforts have been hindered by reliance on traditional univariate analysis of fMRI. Whereas univariate analyses average activity across voxels in a given region, multivariate pattern analyses (MVPA) leverage the information present in spatial patterns of activity. MVPA therefore provides a more sensitive analysis tool to translate rodent neurobiology to human neuroimaging. We review human fMRI studies using MVPA that successfully bridge rodent models of amygdala, hippocampus, and mPFC function during Pavlovian learning. We also highlight clinical applications of these information-sensitive multivariate analyses. In sum, we advocate that the field should consider adopting a variety of multivariate approaches to help bridge cutting-edge research on the neuroscience of threat and anxiety.

Keywords: Amygdala; Extinction; FMRI; Fear; Hippocampus; MPFC; MVPA; Univariate.

Figure 1.

Overview and applications of MVPA. A. Analytic approaches of univariate and multivariate analyses. A univariate analysis considers the mean activation in a single voxel over time, or in the case of region of interest analyses, the mean activation across voxels. In many regions, CS+ and CS− stimuli do not differ in total mean activation. Multivariate analyses are concerned with the information present in the pattern of activation across voxels in a given region. In this schematic, while the CS+ and CS− stimuli have equivalent mean activation across these 4 voxels, they each have a unique spatial pattern of activity. B. Representational similarity analysis (RSA). In RSA, the multivariate patterns of activity for each stimulus are compared to each other using a similarity metric, such as Pearson’s r. The results can be visualized in a representational similarity matrix. C. Decoding analyses. As with RSA, decoding analyses rely on the unique spatial patterns of activity for different stimuli. In a typical decoding analysis, multiple presentations of stimuli from different categories are submitted to a linear classifier as training data. Using a cross-validation procedure, one approach is to test whether the stimuli from different categories are separable in a given brain region. Another approach is to use the trained classifier to attempt to decode a novel stimulus, such as in threat generalization.