Fig. 3: Model predictions and comparison.
From: Controllability changes pain perception by increasing the precision of expectations
a Schematic group-level predictions of the winning model (in grey) follow closely the empirical average pain ratings (green to yellow). b Model comparison using the robust version of leave-one-out cross validation. Expected log point-wise predictive density (ELPD) was compared between models. Higher ELPD values indicate better model fit. The winning precision-change model implemented a dynamic prior mean in the unpredictable condition and shared posterior means for the controllable and predictable condition. Precisions were allowed to vary between the conditions. c Model predictions for pain ratings in two exemplary single subjects. Black squares indicate actual ratings and blue circles indicate model predictions.
