In your phase: neural phase synchronisation underlies visual imagery of faces

Abstract

Mental imagery is the process through which we retrieve and recombine information from our memory to elicit the subjective impression of "seeing with the mind's eye". In the social domain, we imagine other individuals while recalling our encounters with them or modelling alternative social interactions in future. Many studies using imaging and neurophysiological techniques have shown several similarities in brain activity between visual imagery and visual perception, and have identified frontoparietal, occipital and temporal neural components of visual imagery. However, the neural connectivity between these regions during visual imagery of socially relevant stimuli has not been studied. Here we used electroencephalography to investigate neural connectivity and its dynamics between frontal, parietal, occipital and temporal electrodes during visual imagery of faces. We found that voluntary visual imagery of faces is associated with long-range phase synchronisation in the gamma frequency range between frontoparietal electrode pairs and between occipitoparietal electrode pairs. In contrast, no effect of imagery was observed in the connectivity between occipitotemporal electrode pairs. Gamma range synchronisation between occipitoparietal electrode pairs predicted subjective ratings of the contour definition of imagined faces. Furthermore, we found that visual imagery of faces is associated with an increase of short-range frontal synchronisation in the theta frequency range, which temporally preceded the long-range increase in the gamma synchronisation. We speculate that the local frontal synchrony in the theta frequency range might be associated with an effortful top-down mnemonic reactivation of faces. In contrast, the long-range connectivity in the gamma frequency range along the fronto-parieto-occipital axis might be related to the endogenous binding and subjective clarity of facial visual features.

Figure 1

Diagram of an experimental trial for the visual imagery and control conditions. Visual imagery and control conditions had a parallel structure. (A) In the visual imagery condition, a celebrity face was visualised mentally after seeing the corresponding name that was presented for 300 ms. Participants pressed a key as soon as they had generated a vivid mental image, and 200 ms after this keypress, a test face appeared for 300 ms which was of the same person whose face has been mentally visualised before. (B) In the control condition, after seeing the name of a celebrity, participants were asked to press a key after seeing a grey oval on the screen when they felt the urge to do so. Then, 200 ms after the keypress, a test face appeared (300 ms) which was congruent with the name of the celebrity presented before.

Figure 2

Gamma phase-synchrony between frontal and parietal electrodes. wPLI (30–60 Hz) for visual imagery (A) and control (B) conditions. (C) Cluster-based permutation test comparing visual imagery and control conditions. The area highlighted in the time–frequency chart depicts a significant wPLI cluster between conditions (visual imagery minus control) in the gamma band (30–60 Hz). (D) Region of interest (ROI) for phase-synchrony analysis. wPLI values are expressed in standard deviations (z-scores) in reference to the baseline (− 1500 to − 1250 ms). Trial length (− 1500 ms) is relative to response time (0 ms). (D) Topographical representation of frontoparietal wPLI channel pairs and single-participant wPLI values for the imagery (left) and control (right) conditions. Each arc represents a pair of channels, and the height of the arc is its normalised value. Grey circles represent single-participant wPLI values for the cluster depicted in (C) for the imagery (left panel) and control (right panel) conditions. The red horizontal line represents the group mean, the rectangle represents SEM, and the red-dashed horizontal line represents the group median.

Figure 3

Theta phase-synchrony between inter-hemispheric frontal electrode pairs. wPLI (1–10 Hz) for visual imagery (A) and control (B) conditions. (C) Cluster-based permutation test comparing visual imagery and control conditions. The area highlighted in the time–frequency chart depicts a significant wPLI cluster between conditions (visual imagery minus control) in the theta band (5–7 Hz). (D) Region of interest (ROI) for phase-synchrony analysis. Only interhemispheric frontofrontal pairs of channels were considered for the analysis (see Sect. 2.8 for details). wPLI values are expressed in standard deviations (z-scores) in reference to the baseline (− 1500 to − 1250 ms). Trial length (− 1500 ms) is relative to response time (0 ms). (D) Topographical representation of frontofrontal wPLI channel pairs and single-participant wPLI values for the imagery (left) and control (right) conditions. Each arc represents a pair of channels, and the height of the arc is its normalised value. Grey circles represent single-participant wPLI values for the cluster depicted in (C) for the imagery (left panel) and control (right panel) conditions. The red horizontal line represents the group mean, the rectangle represents SEM, and the red-dashed horizontal line represents the group median.

Figure 4

Gamma phase-synchrony between occipital and parietal electrodes. wPLI (30–60 Hz) for visual imagery (A) and control (B) conditions. (C) Cluster-based permutation test comparing visual imagery and control conditions. The area highlighted in the time–frequency chart depicts a significant wPLI cluster between conditions (visual imagery minus control) in the gamma band (30–60 Hz). (D) Region of interest (ROI) for phase-synchrony analysis. wPLI values are expressed in standard deviations (z-scores) in reference to the baseline (− 1500 to − 1250 ms). Trial length (− 1500 ms) is relative to response time (0 ms). (D) Topographical representation of occipitoparietal wPLI channel pairs and single-participant wPLI values for the imagery (left) and control (right) conditions. Each arc represents a pair of channels, and the height of the arc is its normalised value. Grey circles represent single-participant wPLI values for the cluster depicted in (C) for the imagery (left panel) and control (right panel) conditions. The red horizontal line represents the group mean, the rectangle represents SEM, and the red-dashed horizontal line represents the group median.

Figure 5

Multiple and single regressions between wPLI and behavioural scales. (A) Beta coefficients for three separate multiple regressions using Likert scales (vividness, contour definition, and emotional expression) as an outcome variable and frontoparietal and occipitoparietal wPLI as predictors. (B) Pearson’s correlation between frontoparietal and occipitoparietal wPLI values and the corresponding scale scores.