Figure 5
Slope of the learned size-weight mapping for the four family objects (family regression). The family regression is a linear model relating object sizes (heights) to weight predictions (spring lengths), fit separately for each participant. All data shows mean ± 1 SEM, shown by error bars or shading. The true sizes and weights of the objects are shown with blue dots. The predicted weights in blocks 12–22 are shown in black (outlier response is shown with a dot, family responses are error bars only), and the black solid line shows the family regression to these data, which was used in computing strength of categorical encoding. The gray data points and dashed regression line show the results for blocks 6–10, prior to the introduction of the outlier.
