Table. 3.

Direct model comparisons between neural scores

Base model	Composite model	Improvement of composite model over base model (χ2)	p
Informational network	Informational network + RSA	1.28	0.26
Informational network	Informational network + univariate	0.003	0.96
Searchlight RSA	Informational network + RSA	7.29	0.0069
Whole-brain univariate	Informational network + univariate	7.59	0.0059

Each alternative neural score was directly compared with the informational network neural using stepwise linear mixed model comparisons (likelihood ratio test). The predictive strength of the informational network neural score model was not improved by the addition of either the RSA or the univariate neural score as a second predictor. By contrast, both the RSA and univariate base models were significantly improved by the addition of the informational network model as a second predictor