Figure 2
Summary of the effects of polar-extremity in the normative valence scores of the stimulus on modeling performance. (A) Estimation of the contribution of fixed effects, separately for heart rate deceleration and SVM predictions as a function of the width of the threshold used to exclude “neutral” stimuli. This threshold is reported in units of the 9-point Likert scale within which the IAPS image set was originally acquired. *Denotes a non-significant fixed effect at that level of thresholding. (B) Mean normative valence scores of the positively and negatively valent stimulus sets created by excluding neutrally valent images as a function of the exclusion threshold. Note, the black dots depict the mean normative valence scores of the positively and negatively valent image sets analyzed in Bradley et al.7. The neutral valence threshold representative of this stimulus set in our dataset was transferred to panels A and C, respectively, for interpretation of these findings. Also note, horizontal gray lines denote mean valence scores for positively and negatively valence stimulus sets reported for other prior work reporting heart rate change discrimination of valenced image stimuli (see Supplemental Table 1 for reference). (C) Fraction of full stimulus set included as a function of scaling.
