Infants¿ Social Evaluation of Helpers and Hinderers: A Large-Scale, Multi-Lab, Coordinated Replication Study

To examine whether infants do indeed engage in social evaluation of prosocial and antisocial characters, we chose to replicate the hill study by Hamlin and colleagues (2007). This study was selected because: (1) it is the most widely cited demonstration of infant social evaluation in the literature, (2) it has been successfully replicated in subsequent research by the original lab (Hamlin, Wynn, & Bloom, 2010) and at least one independent laboratory (Chae & Song, 2018), but not by other others (Cowell & Decety, 2015; Scarf et al., 2012; Schlingloff et al., 2020); (3) the effect has been reported in studies employing multiple response measures (including preferential looking, anticipatory looking, and selective reaching) and different presentation formats (video stimuli presented on screen rather than live displays: Hamlin, 2015). We reasoned that video stimuli would be easier to utilize on a global scale.Consistent with the ManyBabies goal of conducting the best possible test of a given hypothesis as opposed to exact replications, we made several modifications to the original Hamlin and colleagues (2007) paradigm. First, as just noted, we utilized a filmed puppet show as opposed to a live puppet show as in the original study. Using pre-recorded stimuli standardizes the stimuli presented to infants across laboratories, thereby ensuring that any differences in results can not be attributed to variations in habituation events. Further, videotaping these events rather than producing them in real-time allowed us to, within condition, match the speed and timing of the pushing-up and down actions along with the overall exposure to the push-up/push-down characters down to the millisecond. This method of presentation also increases the number of labs eligible for participation because it substantially reduces barriers to participation, such as financial/time constraints involved in purchasing puppet stage materials, constructing a puppet stage, and training researchers to execute a live puppet show. The video stimuli used here were recorded in Hamlin¿s lab and closely matched videos used in a successful replication (Hamlin, 2015) of the live puppet show paradigm (Hamlin et al., 2007). Because these two studies (Hamlin, 2015; Hamlin et al., 2007) differed only in the stimulus presentation modality and found results with comparable effect sizes, we expected that the effect size would not be moderated by this decision. Importantly, meta-analytic results also found no moderating effect of the modality of stimulus presentation (i.e., animations, videotaped, and real events; Margoni & Surian, 2018).Second, our design implemented controls for perceptual differences between helping and hindering events that were not present in the initial Hamlin and colleagues (2007) study. As previously discussed, Scarf and colleagues (2012) argued that infants¿ preferences in the hill paradigm were due to the Climber character bouncing after being helped but not after being hindered (but see Hamlin, 2015, for evidence against these criticisms). To avoid this issue, our study utilized videos in which the Climber remained motionless, instead of bouncing, upon reaching its final position. Third, rather than including two separate age groups (6- and 10-month-olds), as in the original study, we included a single group of infants ranging from 5.5 months to 10.5 months. This age range was selected for several reasons. First, a manual reaching choice task can be used across this age window, allowing us to fully standardize the task across all infants. Second, infants in this age range demonstrate sensitivity to the causal power of agents (Liu et al., 2019), and to both successful (e.g., Woodward, 1998) and failed goal-directed actions (e.g., Brandone & Wellman, 2009; Hamlin, Hallinan, & Woodward, 2008). Third, although Margoni and Surian (2018) did not find a significant influence of age on infants¿ preference for prosocial individuals, several successful and failed replications fall within this age range (Hamlin, 2015; Hamlin et al 2011; Salvadori et al., 2015; Scarf et al., 2012). Thus, including this broad age range allowed us to assess whether there are developmental changes in infants¿ preferences for prosocial others. Finally, as recruiting participants across a broad age range is presumably easier than recruiting within a narrow age range, we selected a wide age window to maximize the number of laboratories able to participate.As in the original Hamlin and colleagues (2007) study, we included a Non-Social control condition to examine whether infants¿ preferences are driven by the social aspects of helping versus hindering actions as opposed to non-social perceptual features of the displays. In the control condition, infants viewed events similar to the helping and hindering events of the Social condition, but with several notable differences. Most critically, the Climber was replaced by an inanimate, eyeless object that did not engage in self-propelled motion. Specifically, infants viewed an inert red ball being pushed up or down the hill by triangle and square agentic characters with eyes. Based on the estimate from Margoni and Surian¿s (2018) meta-analysis, we predicted that approximately 64% of infants would choose the Helper in the Social condition where the animate climber, a red ball with eyes, demonstrated an unfulfilled goal to climb the hill. We predicted that infants would not demonstrate a preference for the character who pushed an inanimate red ball (that had no eyes and demonstrated no goal- directed behavior) in the Non-Social control condition. Relatedly, we also predicted significantly greater preference for the Helper in the Social compared to the pusher-upper in the Non-Social condition.The Social and Non-Social videos were designed to convey fundamentally different events -- helping/hindering an animate character versus pushing an inanimate character up/down -- therefore, it was necessary that they differed in several ways. First, we had to ensure that the ball was perceived as animate in the Social videos, and as inanimate in the Non-Social videos. To do so, Social videos included a hill-climbing action at the start which demonstrated the ball¿s goal to go uphill. The Non-Social videos do not have this portion of the video, since the ball is inanimate and not capable of self-propelled motion. This difference led to the Non- Social videos being 4.4 seconds shorter than the Social videos. Although timing of the Non- Social videos could have been matched to the Social videos by introducing additional still frames and/or adding in novel actions, we reasoned that these modifications might lead to inattentiveness and fussiness in the Non-Social displays, insofar as they do not add anything directly relevant to, or may even hamper the interpretability of the push-up or push-down character¿s goals. Despite the overall length across Social and Non-Social videos, the length of the videos was nevertheless equated within condition (i.e., both Social videos are 13.3s and both Non-Social videos are 8.9 s), and the amount of time the helper/push-up and hinderer/push-down characters are on stage are exactly matched within and closely matched across conditions (Social = 4.7 s, Non-Social = 5.9 s). Although we do not expect these differences in timing to impact our main results of interest, we will test for the possible influence of these timing differences by analyzing infants¿ attention (e.g., as measured by the number of habituation trials and the overall looking to the still frame events presented after each video), and explore whether (a) attention differs across conditions, and (b) differences in attention following each event relate to differences in infants¿ choices.A second difference between the Social and Non-Social conditions involves the location of the climber at the start of each event. In the Social condition the climber always has the goal to climb up a steep hill, and needs assistance in doing so. Therefore, the climber starts at the bottom of the hill during both helping and hindering events (before being pushed up or down). In the Non-Social condition, infants also see the ball being pushed up or down the hill. But, because the ball is inanimate and not-goal directed, the ball must begin at the bottom of the hill during pushing-up events, but at the top of the hill during pushing-down events. Although we could have created events where the character was trying to climb down a hill, we thought this goal might be perceived as relatively trivial in terms of costs (compared to climbing up a steep hill), rendering it more difficult to determine that the climber needs assistance. We have no reason to hypothesize that these differences in starting location across conditions would influence infants¿ choices.Accordingly, the main research question is: do infants have sensitivity towards moral valence of agents behavior?