Table 2.

Quality of model fits for computational models of altruistic decision-making in sessions 1 and 2

Model	Description	Parameters per subject	Exceedance probability (session 2)	BIC (session 2)	BIC (session 1)
F1.1	$\kappa$,$\lambda$	2	0	3708	7697
F1.2	$\kappa$, $\alpha$, $\lambda$	3	0.71	3223	6690
F1.3	$\kappa$, $\beta$, $\lambda$	3	0.29	5269	6892
F1.4	$\kappa$, $\alpha$,$\lambda$	3	0.0007	8470	14,367
F1.5	$\kappa$, $\alpha$, $\beta$, $\lambda$	4	0	4328	8090
F2.1	$\kappa$, $m$,$\lambda$	3	0.0005	4456	7843
F2.2	$\kappa$, $m$, $\alpha$, $\lambda$	4	0	3369	7719
F2.3	$\kappa$, $m$, $\beta$, $\lambda$	4	0.0001	5140	11,841
F2.4	$\kappa$, $m$, $\alpha$, $\lambda$	4	0	6211	12,620
F2.5	$\kappa$, $m$, $\alpha$, $\beta$, $\lambda$	5	0	5802	11,881
F3.1	$\kappa$, $m$, $p$, $\lambda$	4	0	4042	7076
F3.2	$\kappa$, $m$, $p$, $\alpha$, $\lambda$	5	0	3537	6971
F3.3	$\kappa$, $m$, $p$, $\beta$, $\lambda$	5	0.0001	5307	12,009
F3.4	$\kappa$, $m$, $p$, $\alpha$, $\lambda$	5	0	6379	12,788
F3.5	$\kappa$, $m$, $p$, $\alpha$, $\beta$, $\lambda$	6	0	5970	12,049

BIC, Bayesian information criterion. BIC scores are summed across subjects. Model F1.2 was favored across both sessions. All models have an inverse temperature parameter $\mathrm{\lambda }$. κ, relative weighting parameter for others' benefits (altruistic preference) in models F1.1– F1.5; κ, weighting parameter for others' benefits in models F2.1–F2.5 and F3.1–F3.5; $m$, weighting parameter for one's own costs in models F2.1–F2.5 and F3.1–F3.5; $p$, weighting parameter for interaction between benefit and cost in models F3.1–F3.5; $\alpha$, power exponent which modulates the nonlinearity of others' benefits; $\beta$, power exponent which modulates the nonlinearity of one's own costs.