Table 2.

Details of the three “Social” versus “Non-social” comparisons (based on observer reports) performed as part of the GLM analysis

Analysis Number	“Social” responder group	“Non-social” responder group	Rationale for the analysis	Statistical comparisons for fMRI	ID of the analysis used in figures and text
1	RANDOM MECHANICAL (Within animation, between participant)		Controls for low-level input Most ambiguous animation	Two-sample t test	RANDOM MECH
2	COAXING	BILLIARD	Likely similar decision times during animation (as estimatedfrom online RT experiment)	Paired t test	COAXING-BILLIARD
	(Between animation, within participant)
3	All “Social”	All “Non-social”	Maximizes power by comparing all “Social” responses with all“Non-social” responses within participants	One-sample t test after averaging runwise estimates	ALL ANIMATIONS
	Coded 1 (“Social”) and −1 (“Non-social”), in runwise GLMs