Infants’ Expectations and Evaluations of Agents Who Inconsistently Help and Hinder Others

• In Event: Inferences drawn from moral actions in infants and children

Abstract

Infants prefer agents who consistently facilitate rather than hinder others’ goals (Hamlin & Wynn, 2011; Margoni & Surian, 2018). However, individuals in the real world perform both helpful and unhelpful acts across time. The present experiments seek to address whether 11-month-olds can evaluate inconsistent agents, and whether they expect agents to behave consistently. Specifically, if infants base their evaluations on an averaging strategy of agents’ variable behaviors, then they should prefer agents who have been more helpful. Furthermore, if infants have expectations for consistency, there should be a recovery in looking time when an agent changes its behavior.

Steckler et al. (2017) probed 9-month-olds’ evaluations of agents who inconsistently helped and hindered a protagonist’s attempts to open a box. Infants were familiarized to either (i) a Consistent Actor who always helped and an Inconsistent Actor who mostly hindered or (ii) a Consistent Actor who always hindered and an Inconsistent Actor who mostly helped. Across experiments, infants did not preferentially reach to more helpful actors. Thus, 9-month-olds appeared to have difficulty incorporating inconsistency into their social evaluations.

In Experiment 1 (in-person), we again probed infants’ responses to inconsistent actors, adjusting Steckler et al.’s methods in three ways. First, we examined older (11-month-old) infants, who may better reason about inconsistency. Second, we adopted a different paradigm to demonstrate helping and hindering, involving a protagonist who loses control of its ball, based on findings from a meta-analysis that evaluations here may be particularly strong (Margoni & Surian, 2018). Third, we provided infants with more evidence of how an actor usually behaves before its behavior changes, to support their evaluations. Nevertheless, infants again did not prefer the “more helpful” actor (ps < .578); however, infants looked numerically longer when an actor’s behavior changed (b = 0.44, t = 2.50, p = 0.12). There were also suggestive, non-significant trends that infants more strongly responded when an actor’s behavior changed from helping to hindering than the reverse.

We next sought to replicate these findings and determine whether non-significant trends were meaningful. Due to the COVID-19 pandemic, we transitioned to online testing. In Experiment 2, we examined the validity of online testing by attempting to replicate the finding that infants prefer actors who consistently help a protagonist retrieve a lost ball over actors who consistently take the ball away. Here, infants (n = 24) preferred looking to the helper over the hinderer (b = 3.47, p < .001). Thus, we are using online testing for Experiment 3.

In Experiment 3, we are again examining 11-month-olds' responses to inconsistent actors. Unlike in Experiment 1, pilot data suggested that infants may prefer a more helpful actor. Data collection is ongoing and on track to finish (n = 48) by Spring 2023.

Together, these findings provide evidence that infants may notice changes in agents’ prosocial vs. antisocial behaviors and demonstrate that it is feasible to collect meaningful data on infants’ evaluations online. Results from Experiment 3 will further shed light on how infants make sense of inconsistency in social behavior.

Authors

• Chiu Kin Adrian Tsang, University of British Columbia (UBC)
  Presenting Author

• Brandon Matthew Woo, Harvard University

• J. Kiley Hamlin, University of British Columbia (UBC)