Determine the first reaching attempt. It was coded as 1, if directed towards the pro-social agent, and 21, if directed towards the antisocial agent, respectively. No obvious intended action towards one or the other toy after more than one minute was coded as zero The reliability between the two coders was.85 using Cronbach’s alpha, and.80 using Cohen’s Kappa (see Table S2 for row results). In order to analyze these responses statistically, the scores for the four trials (involving different toys) were averaged across the two coders to produce an average Choice Index (CI) for each infant (between 21 and +1): 57 of the infants had a MK-8742 msds positive Choice Index, 21 a negative one, and 21 were at zero (see Figure 2.B). An ANOVA of the Choice Index was run, declaring actor, order of the actions, and order of the patients as counterbalancing factors. The intercept of the Choice Index was significantly above zero (index = 0.138; SE = 0.051, F(1,37) = 7.9, p,.008, gp2 = .18) indicating that infants reached significantly more towards the object of the pro-social agent than the object of the anti-social agent. There was a marginal preference for one of the actors, (F(1,37) = 3.9, p = .06, gp2 = .10) and no other counterbalancing factor introduced a significant effect or interaction (p..1). Adoi:10.1371/journal.pone.0088612.tthe stage from each side, standing still facing the camera, playing with identical toys (teddy bears), and bending towards the camera as if both were simultaneously giving the toy to the infant. Through a mechanical apparatus (see Figure 2.A), the two identical toys then appeared on the table in front of the infant while it disappeared from the agents’ hands. Afterwards, the two agents appeared standing still with no toy, looking at the infant. This procedure was repeated four times (with four different teddy bears), with the agents swapping sides from trial to trial. The infants were videotaped and their initial attempts to reach for one of the toys were coded by two independent blind coders. Ethical issues. The same procedure as in Experiment 1 was used, and was covered by the same decision from the Cochin Ethical Committee. Participants. We tested fifty-four 10-month-old infants (age range = 9?0 months; 31 males, 23 females) recruited in Paris through mailing and telephone calls. The parents were fully informed about the protocol and were shown the sequences as in Experiment 1. The babies were seated in the lap of a purchase BX795 parent, whoFigure 2. Set up and average preference index for 10-month-olds. A Setup of the experiment for the 10-month-olds. Infants are seated in their parent’s lap (who are blindfolded) during the experiment. After the presentation of the sequences described in Figure 1, infants see a novel sequence, where the two actors appear on stage, stare at the camera, play with two identical toys, and present them to the infants. Simultaneously, two real toys appear on the table through a mechanical device. The infants’ first attempt to reach one of the toys is coded by two independent scorers. Infants are tested 4 times with different toys, and the actors switching places on the screen. B. Average preference index for 10-month-olds computed over the 4 response trials. A positive score indicates a preference towards the toy of the pro-social agent, a negative score, a preference towards the anti-social agent. The error bars correspond to one between-subject standard error above and below the mean. *p,.05. doi:10.1371/journ.Determine the first reaching attempt. It was coded as 1, if directed towards the pro-social agent, and 21, if directed towards the antisocial agent, respectively. No obvious intended action towards one or the other toy after more than one minute was coded as zero The reliability between the two coders was.85 using Cronbach’s alpha, and.80 using Cohen’s Kappa (see Table S2 for row results). In order to analyze these responses statistically, the scores for the four trials (involving different toys) were averaged across the two coders to produce an average Choice Index (CI) for each infant (between 21 and +1): 57 of the infants had a positive Choice Index, 21 a negative one, and 21 were at zero (see Figure 2.B). An ANOVA of the Choice Index was run, declaring actor, order of the actions, and order of the patients as counterbalancing factors. The intercept of the Choice Index was significantly above zero (index = 0.138; SE = 0.051, F(1,37) = 7.9, p,.008, gp2 = .18) indicating that infants reached significantly more towards the object of the pro-social agent than the object of the anti-social agent. There was a marginal preference for one of the actors, (F(1,37) = 3.9, p = .06, gp2 = .10) and no other counterbalancing factor introduced a significant effect or interaction (p..1). Adoi:10.1371/journal.pone.0088612.tthe stage from each side, standing still facing the camera, playing with identical toys (teddy bears), and bending towards the camera as if both were simultaneously giving the toy to the infant. Through a mechanical apparatus (see Figure 2.A), the two identical toys then appeared on the table in front of the infant while it disappeared from the agents’ hands. Afterwards, the two agents appeared standing still with no toy, looking at the infant. This procedure was repeated four times (with four different teddy bears), with the agents swapping sides from trial to trial. The infants were videotaped and their initial attempts to reach for one of the toys were coded by two independent blind coders. Ethical issues. The same procedure as in Experiment 1 was used, and was covered by the same decision from the Cochin Ethical Committee. Participants. We tested fifty-four 10-month-old infants (age range = 9?0 months; 31 males, 23 females) recruited in Paris through mailing and telephone calls. The parents were fully informed about the protocol and were shown the sequences as in Experiment 1. The babies were seated in the lap of a parent, whoFigure 2. Set up and average preference index for 10-month-olds. A Setup of the experiment for the 10-month-olds. Infants are seated in their parent’s lap (who are blindfolded) during the experiment. After the presentation of the sequences described in Figure 1, infants see a novel sequence, where the two actors appear on stage, stare at the camera, play with two identical toys, and present them to the infants. Simultaneously, two real toys appear on the table through a mechanical device. The infants’ first attempt to reach one of the toys is coded by two independent scorers. Infants are tested 4 times with different toys, and the actors switching places on the screen. B. Average preference index for 10-month-olds computed over the 4 response trials. A positive score indicates a preference towards the toy of the pro-social agent, a negative score, a preference towards the anti-social agent. The error bars correspond to one between-subject standard error above and below the mean. *p,.05. doi:10.1371/journ.