Figures and data
Experimental Procedure.
Structure of the experiment including pre-measurements and the main experiment. Expectations were generated using a sham BCI, i.e., participants were told that they would receive real-time feedback regarding their pain sensitivity (verbal instructions) and experienced the validity of this feedback (conditioning). In the conditioning phase, green cues were paired with lower pain intensities compared to red cues unbeknownst to the participants. In the test phase, the stimulation temperature was always the same, regardless of the cue. The presentation of the condition cue varied from trial to trial.
Expectation ratings, pain ratings and skin conductance response by condition.
Mean expectation (A) and pain ratings (B) on a visual analogue scale separately for each condition. (C) Mean skin conductance responses in the three conditions. White dots = mean, horizontal lines = median, thick grey vertical lines = upper and lower quartile, coloured dots = pain ratings of individual participants per condition. * p < .05. ** p < .01. *** p < .001.
SIIPS scores by condition.
(A) Mean SIIPS score per condition for all time-points during pain perception. White dots = mean, horizontal lines = median, thick grey vertical lines = upper and lower quartile, coloured dots = pain ratings of individual participants per condition. * p < .05. ** p < .01. *** p < .001. (B) Mean SIIPS score per condition plotted over the duration of the whole trial. The mean SIIPS scores shown in A were extracted from the grey marked period.
Differentiation of effects during the anticipation and pain phase.
(A) Common effects during pain anticipation (expectation > no expectation) at p < .001 (uncorrected for display purposes) show widespread higher activity for both positive and negative expectations compared to the control condition. Bottom: Distinct effects (positive > negative) during pain perception are shown, indicating broadly higher activity for positive compared to negative expectations. (B) Left: For selected areas, the overlap between common effects of expectations during the anticipation phase (yellow) and distinct effects of positive and negative expectations during the pain phase (green) in the respective area is shown. Right: The corresponding activation levels of positive and negative expectations (i.e., beta weights from the FIR model) baselined by the control condition are plotted for each phase at the respective peaks (peak coordinates in parentheses). The visualization highlights the differentiation of effects following the onset of pain.
Representation of Expectations in the Angular Gyrus.
(A) Overlap between common effects of expectations during the anticipation phase (expectation > no expectation; yellow) and distinct effects of positive and negative expectations during the pain phase (positive > negative; green) is shown for the angular gyrus at p < .001 (uncorrected for display purposes). (B) The corresponding activation levels of positive and negative expectations (i.e., beta weights from the FIR model) baselined by the control condition are plotted for each phase at the respective peaks (peak coordinates in parentheses).
Relation of fMRI activity with EEG oscillatory power.
Correlation of single-trial hemodynamic responses with time-frequency resolved EEG activity in the left DLPFC (A), left anterior insula (B), and left ACC (C) during the anticipation phase, ordered by the timing of observed correlations as indicated by the arrow on the right. Single-trial beta weights were extracted from spherical ROIs (10 mm radius) centered around the peak voxels based on the comparison of beta weights from the FIR model between expectation and no expectation during the anticipation phase, as shown on the left (p < .001 uncorrected for display purposes). On the right, the cluster-corrected correlation of oscillatory power with fMRI activity averaged over all cluster electrodes is depicted. Non-significant time-frequency points are masked.
