Table 2.

GLM comparison cannot disambiguate between the causal structure learning model (GLM 1: M₁, M₂, and M₃) and the clustering model (GLM 3: RL + clustering)

GLM	Model	Parametric modulators	PXP
GLM 1	M1, M2, M3	KLstructures, KLweights (sum)	0.4009
GLM 2	M1, M2, M3	KLstructures, KLweights (MAP)	0.0993
GLM 3	RL + clustering	KLclusters, CPE	0.4006
GLM 4	RL + clustering, simple RL	CPE, FPE	0.0992

PXPs were based on whole-brain activity from all trials.