Effective connectivity in the active encoding of facial emotions: a dynamic causal modeling study

Previous studies have identified brain regions involved in the active encoding of emotional content of facial expressions (ECFEs); however, little is known about how these regions interact within a functional network. Electroencephalography data were recorded from 62 healthy university students performing a delayed match-to-sample task in which facial expressions served as both encoding and interfering stimuli. Dynamic causal modeling was applied to examine encoding-related modulations of effective connectivity among six nodes in bilateral inferior occipital, inferior temporal, and inferior frontal gyri. During the early time window (1-200 ms), encoding modulated almost all the connectivities including reduced bottom-up excitatory and strengthened top-down inhibitory visual-temporal connectivities, strengthened bidirectional visual-frontal connectivities, and reduced bidirectional temporal-frontal connectivities. During the later window (200-400 ms), only a subset of connectivities was modulated, marked by strengthened top-down excitatory and inhibitory connectivities from frontal and temporal regions to visual areas. Modulation strength was not reliably associated with behavioral performance. Exploratory Bayesian model selection within a restricted model space showed a numerical advantage for a bottom-up excitatory model during encoding and a top-down excitatory model during interference suppression under fixed-effects modeling; however, random-effects analysis indicated that this preference was modest and not uniform across participants. These findings suggest that the active encoding of ECFEs is supported, but not determined, by directional interactions among brain modules, implicating strengthened high-order top-down control over posterior sensory areas, together with functional decoupling of face-sensitive areas from ongoing visual input.