Table 3.

Different mechanisms underlying resolution of reward and affective PEs

	Round 1 – 2 Update (Expectations)	Rounds 2-5 Update (Expectations)	Round 1 – 5 Update (Experience)
	t-score (p-value)	$\beta \pm SE$ (p-value)	$\beta \pm SE$ (p-value)
Fair partner
Reward	9.53 (<0.001)	0.0$1\pm$0.005 (0.02)	–
Valence	9.23 (<0.001)	0.0$1\pm$0.01 (0.44)	−0.04$\pm$0.01 (<0.001)
Arousal	5.00 (<0.001)	–0.001$\pm$0.02 (0.94)	–0.08$\pm$0.02 (<0.001)
Neutral partner
Reward	–3.00 (0.005)	–0.0$1\pm$0.01 (0.09)	–
Valence	–1.44 (0.16)	-0.02$\pm$0.01 (0.16)	0.02$\pm$0.01 (0.052)
Arousal	–1.45 (0.15)	–0.03$\pm$0.01 (0.05)	−0.01$\pm$0.01 (0.35)
Unfair partner
Reward	–14.19 (<0.001)	–0.0$2\pm$0.01 (0.006)	–
Valence	–10.36 (<0.001)	0.01$\pm$0.02 (0.51)	0.07$\pm$0.02 (<0.001)
Arousal	–2.28 (0.03)	−0.01$\pm$0.02 (0.55)	–0.003$\pm$0.01 (0.82)

Rounds 1–2 update (expectations) shows the results of paired t-tests comparing expectation values from round 2 to round 1. Rounds 2–5 update (expectations) shows the results of LMMs comparing how expectation values change between rounds 2–5. Rounds 1–5 update (experience) shows the results of LMMs comparing how emotional experiences change between rounds 1–5. Reward experiences were defined by task parameters and are stable across rounds (see Methods). Full statistical info can be found in Supplementary Table 8–10. All statistical tests are two-sided with no adjustments for multiple comparisons