Fig 1.
Illustration of expression intensity displayed during the experiment, ranging from 0% (neutral) to 130%.
Morphing at equal stages was used to create intermediary intensities between a neutral (0%) and a naturally high facial expression of pain (100%). Higher intensities were also created (up to 130%), which gave more caricatured expressions.
Fig 2.
Illustration of a trial sequence.
Here, the video reached 90% intensity (VideoEI90%) and the TestEI presented 65% pain intensity. The participants had to memorize the final expression intensity (90%) and compare this memorized expression with the TestEI. The static TestEI was displayed until the participants had given their responses via the keyboard.
Fig 3.
Example of behavioral results from one participant, in response to the VideoEI90% condition.
Participants presented a negative memory bias, with a PSE of 65.9%, for a video stopping at 90% of the maximal intensity. Triplets of TestEIs are represented by gray dots.
Fig 4.
The data from six participants were recorded using 32 electrodes, whilst the data from the remaining 19 participants were recorded using 64 electrodes. For the purposes of statistical analysis, we grouped electrodes into four ROIs, based on anatomical and functional criteria: Frontal (F), Centro-Parietal (CP), Temporal (T) and Occipital (O).
Fig 5.
Average ERPs, by hemisphere (with congruity conditions combined).
Apart from the baseline, data recorded before the 0 ms time value (i.e., data recorded during the presentation of the dynamic facial expression and the mask) are presented for illustrative purposes only. In the T ROI (panel C), the average amplitude of the EEG signal (across all time windows) was more negative in the right hemisphere (M = -3.81 μV) than in the left hemisphere (M = -2.98 μV). In the O ROI, significantly greater deflections were observed in the right hemisphere than in the left hemisphere, for both P100 and N170 waves (panel D).
Fig 6.
Average ERPs, by congruity condition (with hemispheres combined).
Apart from the baseline, data recorded before the 0 ms time value (i.e., data recorded during the presentation of the dynamic facial expression and the mask) are presented for illustrative purposes only. Taking all time windows together, the mismatch effect was significant in the T (panel C) and O (panels D and E) ROIs. It was also marginally significant in the F ROI (panel A). In contrast, the mismatch effect in the O ROI varied across time windows. It showed marginally weaker P100 amplitude in the incongruent condition than in the congruent condition, and marginally greater N170 amplitude in the incongruent condition than in the congruent condition. However, the difference in the mismatch effect on the LPP component was significant in the left O ROI and not in the right O ROI.
Fig 7.
Source localizations (LORETA) for each congruity condition and the mismatch effect, per time window.
The LORETA suggested that the main source of the P100 wave was in the occipital lobe in the congruent condition (A1) and in the posterior temporal lobe in the incongruent condition (B1), with the source of the mismatch effect located in the posterior temporal lobe (C1). The sources of the N170 and LPP waves were mainly localized in the posterior temporal lobe in both the congruent condition (A2 and A3) and the incongruent condition (B2 and B3). The strongest sources of the N170 and LPP mismatch effect were in the temporal lobe (C2 and C3), and the frontal lobe, respectively.
Table 1.
Significant effects are highlighted in bold.
Fig 8.
Scatterplots of the distribution of the mismatch effect (A) and of the hemisphere effect (B) across the participants, for the F ROI and T ROI, as well as each hemisphere of the O ROI, for each ERP component (P100, N170, and LPP).
Fig 9.
Path analysis mediation model testing for the spatio-temporal unfolding of the mismatch effect from occipital P100 up to Left occipital LPP.
This serial multiple mediator model includes three mediators (F_N170, T_N170, and F_LPP). Standardized (β) coefficients (in bold) are provided together with unstandardized coefficients (between parentheses). The unstandardized coefficients are in μV unit, together with the significance level of effects as §p < .10, *p < .05, ** p < .01, *** p < .001. Significant indirect effects are illustrated with thicker lines.