| Investigation of the neural bases of fear conditioning has historically involved the use of lesions in animal models to find the critical region important for learning CS-UCS associations and producing conditional responses. Recent advances in non-invasive imaging have allowed the assessment of fear conditioning in the intact human brain while associations are learned. In two similar event-related functional magnetic resonance imaging (FMRI) differential conditioning experiments evoked brain activity was assessed while stimulus contingencies were acquired, then violated. In the acquisition phase of both experiments, one visual stimulus was followed by shock (CS+), while another was not (CS−). In experiment 1, the acquisition phase was followed by reversal, where stimulus contingencies were switched. In experiment 2, the acquisition phase was followed by extinction, where neither stimulus was followed by shock. In addition to brain activity, explicit knowledge of stimulus contingencies and implicit autonomic activity were measured throughout training. Behavioral measures showed higher learning-related evoked conditional responses (CRs) to the presentation of the CS+ than the CS−. Signal deconvolution, analysis of variance, and partial least squares analyses were used to detect learning-related changes in stimulus-related and behavior-related brain regions. Brain regions that showed learning-related changes were classified into those uniquely involved in CS input, explicit and/or implicit CR output, or both. A network was created from brain regions that showed learning-related activity (medial temporal lobe, cingulate cortex, limbic thalamus, extrastriate cortex, and prefrontal cortex) as well as with behavior nodes that represented neural influences on implicit and explicit behavioral measures. Analysis of communication in this network using structural equation modeling (i.e. path analysis) revealed evidence for neural influences that changed across phases according to CS type. Connectivity patterns varied between participants with different levels of explicit, but not implicit learning. Changes in influences between regions were characterized as to their involvement in different processes relevant for conditioned fear such as formation and retrieval of conditional associations, CR production, inhibitory associations, and working memory. |
