Examples of reported and modeled perception in a single subject, and adequacy of models across subjects.

<p><b>A–F</b>: A single subject's pain rating (black) and fitted models (red), for a simple stimulus (E) using first (A) and second order (C) models, and for a complex stimulus (F) fitted by first (B) and second order (D) models. <b>G–H</b>: Performance comparisons between models. G: Comparing second order model to linear models, for simple (circle) and complex (triangle) stimuli. The linear model (red) corresponds to the null hypothesis that perception follows temperature, while the threshold model (blue) assumes perception follows temperature above a threshold with the temperature threshold optimized for each subject. H: Comparing first and second order models. Second order model outperforms linear and first order models; circles (triangles) correspond to simple (complex) stimuli. Correlation fit, r (zero lag Pearson correlation), measures accuracy of models to pain perception ratings. Each symbol/category is outcome for an individual subject. Dashed lines are the identity. <b>I–J</b>: Akaike Information Criterion analysis for the fits shown in panels G–H. The vertical axis corresponds to the gain in fit accuracy of the second order model over the null hypotheses (I) and the first order model (J), when a penalty for increased number of parameters is discounted. The color/shape is the same as for panels G–H; the horizontal axis corresponds to different subjects.