Abstract
Sensory processing can be enhanced by the emotional salience of stimuli. For example, fearful faces evoke stronger neural responses than neutral faces in multiple visual cortical areas in both monkeys and humans, including the primary visual cortex. Human and monkeys with amygdala lesions exhibit diminished neural modulation with facial valence, suggesting that the processing of facial valence in visual cortex is accomplished through feedback projections from the amygdala.
We hypothesized that fMRI activity in the visual cortex during face viewing reflects a conjunction of both feedforward stimulus-driven geniculate-striate projections and feedback signals from the amygdala, and that these independent sources of input have distinct laminar profiles.
In this study, we aimed to isolate and separately measure these two sources of input to visual cortex using ultra high-field, high-resolution fMRI. Participants viewed a series of face stimuli that were closely cropped and balanced for low-level visual properties. Face stimuli were presented for 900 ms with a 100 ms ISI, and were blocked by emotional valence (happy, fearful, neutral). Participants performed a gender judgement task unrelated to facial expression.
We measured fMRI activity using vascular-space-occupancy (VASO) fMRI at 7T (0.8x0.8x0.8 mm), and gradient-echo BOLD fMRI at both 3T and 7T (3x3x3 mm and 1.2x1.2x1.2 mm, respectively). We observed a robust and reliable facial valence effect in the amygdala, FFA, and V1 (Fig. 1A), replicating earlier findings. Measurements of the laminar profile of BOLD and VASO activity in V1 showed that the difference in response amplitude between fearful and neutral faces was only evident in the superficial cortical layers (Fig. 1B), the presumed site of amygdala feedback projections. Our results suggest that processing of facial valence in visual cortex is mediated through feedback projections to visual cortex, and that these inputs can be studied independent of the feedforward geniculate-striate projections.