Fig. 3
Model-based physiological responses to different fear levels. Each response was computed using a GLM-based approach as implemented in the PSPM toolbox. For each subject, we specified one regressor for each of the four levels, plus one regressor for catch trials. For pupil size (A), skin conductance (B) and heart period (C), each data point corresponds to the maximum of the response for each fear regressor, with one data point per subject. For respiration amplitude (D), since the response is biphasic, each data point corresponds to the peak of the early response. For all physiological measures, we found a significant difference between low-fear (first quartile) and high-fear trials (last quartile).
