Interhemispheric integration in the neural face perception network: Does stimulus location matter?

Abstract

The neural mechanisms underlying hemispheric lateralization can be investigated using neuroimaging methods and modelling techniques. In some experiments, sensory information is initially presented exclusively to one hemisphere, for example, by displaying a visual stimulus in the periphery of the contralateral hemifield. This experimental design enables, among other things, a comparison of competing theories of interhemispheric integration (e.g., interhemispheric inhibition vs. interhemispheric recruitment). However, the underlying neural models for peripheral stimulation may differ from those for central stimulation and, therefore, may not adequately describe the mechanisms associated with typical, foveal stimulus processing. To address this question, the present functional magnetic resonance imaging (fMRI) study analysed the influence of stimulus location (peripheral vs. central) on neural network connectivity, particularly interhemispheric transfer, as determined by dynamic causal modelling (DCM), for a face perception task. Face and object images were presented either peripherally or centrally to a group of healthy volunteers (N= 17). By contrasting brain activations for faces against objects, we identified bilateral face-sensitive regions, such as the left and right fusiform face area (FFA) and the occipital face area (OFA). Additionally, we extracted the bilateral primary visual cortex (V1) as the input region for our neural models. We constructed five increasingly complex models that differed only in their modulatory connectivity. Bayesian model averaging (BMA) was employed to average the parameters across all models, enabling the calculation of interhemispheric transfer difference (i.e., left-to-right minus right-to-left modulatory connectivity parameter) and the strength of interhemispheric transfer between bilateral OFA and FFA regions. Our findings demonstrate that interhemispheric integration depends on stimulus location. Peripheral presentations of faces induce different connectivity patterns compared with centrally depicted faces. Specifically, we observed larger interhemispheric transfer differences for peripheral face stimuli compared with central stimuli. In conclusion, peripheral and central presentations of faces modulate the face processing network differently, with left and right visual field presentations yielding asymmetrical connectivity patterns. Since faces are preferentially processed via the fovea, the typical face processing network likely aligns more closely with activation patterns elicited by central stimuli. In contrast, connectivity patterns triggered by peripheral stimulation may represent an atypical processing style and cannot be directly compared with those activated by central stimuli.

Keywords: DCM; FFA; OFA; fMRI; face perception; interhemispheric transfer; lateralization; network; peripheral stimulation.

Fig. 1.

Graphical depiction of the experimental task. Subjects were presented with blocks of faces or non-face objects. The stimuli were presented either at the center or periphery (left or right) of the visual field.

Fig. 2.

Model space including five models for the DCM analysis. (A) Illustration of endogenous connections (A-matrix, blue arrows) and driving input (C-matrix, yellow arrows) for stimuli (either faces or non-face object) presented centrally or in the left or right hemifield. (B) Illustration of the modulatory influence (B-matrix, large blue arrows) of face stimuli. Models M1–M5 differ in their complexity.

Fig. 3.

Single subject ROI center coordinates for the six ROIs included in the DCM analysis. The coordinates are presented on an MNI glass brain representation using BrainNet Viewer (Xia et al., 2013).

Fig. 4.

Endogenous connectivity (A-matrix) after BMA across all models and subjects. The strength of connection is shown as the mean of the averaged coupling parameters. Self-connections are omitted in this depiction. Colors indicate the valence of causal influence. Only parameters with a posterior probability > 0.95 are shown.

Fig. 5.

Modulatory (B-matrix) and excitatory (C-matrix) connectivity after BMA across all models and subjects. The strength of connection is shown as the mean of the averaged coupling parameters for LEFT (left side), CENTRAL (middle), and RIGHT (right side) stimulus presentations. Colors (violet, blue) indicate the direction of causal influence of the modulatory parameters. The excitatory parameters are shown in yellow. Only parameters with a posterior probability > 0.95 are shown. Note that the excitatory input enters only one V1 for peripheral stimulation (left / right), but both V1s for central stimulation.

Fig. 6.

Total interhemispheric connectivity strength. The strength of modulatory, B-Matrix connectivity added to endogenous “Baseline”, A-Matrix connectivity for interhemispheric connectivity between FFAs and OFAs for different stimulus locations. Modulatory effects are the main contributor to connectivity strength changes.

Fig. 7.

Interhemispheric transfer. (A)Differencebetween interhemispheric modulatory connectivity, calculated as left-to-right minus right-to-left transfer strength, presented for both OFA and FFA as well as different stimuli locations (left, center, right). (B)Averagebetween interhemispheric modulatory connectivity calculated as the sum of absolute values divided by 2.