Infants and toddlers leverage their understanding of action goals to evaluate agents who help others

Abstract

Although studies have found that infants prefer helpful individuals over unhelpful ones, the basis of such evaluations is unclear. If infants and toddlers, like adults, understand helping as fostering others’ goals, then their evaluations should depend on their ability to infer the goal of an agent in need of help. Here, 15-month-old toddlers and 8-month-old infants (n = 48) differentially evaluated acts of help, consistent with their developing understanding of means-end actions. In further experiments (n = 40), when a protagonist directly grasped an object, 8-month-old infants represented that goal and preferred an agent who facilitated it. These findings connect early evaluations of helping to action understanding.

Toddlers’ social evaluations of agents who act on false beliefs

“…Despite some failures to replicate these findings (Salvadori et al, 2015;Schlingloff et al, 2020), a meta-analysis of experiments probing early social evaluations found that infants and toddlers demonstrate a significant preference for prosocial over antisocial agents (Margoni & Surian, 2018). Since this meta-analysis, 13 additional studies have provided evidence that infants and toddlers evaluate agents based on their social actions (see Woo et al, 2022), consistent with the hypothesis that infants and toddlers prefer agents who help others to achieve their goals. Margoni and Shepperd (2020) have found, moreover, that the failed replications in this literature are consistent with the expected levels of error (i.e., false negatives) that occur when many samples are drawn from a population whose evaluations show a true preference for prosocial agents.…”

“…In lab-based experiments, similar methods have been used to assess infants' and toddlers' evaluations of agents (Colomer et al, 2020;Geraci et al, 2022;Hamlin et al, 2010;Kinzler et al, 2007;Powell & Spelke, 2018), with findings that are consistent with those of studies using reaching measures. Research conducted via video conferencing has used this visual preference method to probe early social evaluations (in situations that do not involve false beliefs), and found that looking and reaching measures converge in infants and toddlers (Woo & Spelke, 2022) (as reviewed above). Other research has used visual preference methods to probe toddlers' understanding of emotion, replicating in-lab findings (Smith-Flores et al, 2022).…”

“…Three experiments tested 15‐month‐old toddlers’ evaluations of two social agents who both either saw or did not see a protagonist's desired toy move to a new location. These experiments adapt methods by Woo and Spelke (2022), who probed toddlers’ understanding of the goals of a bear protagonist who, with help from two rabbit agents, opened a closed box and then grasped a toy inside the box. By late in the first year, infants infer that the goal of such an agent is to obtain the toy, rather than to open the box (Woodward & Sommerville, 2000).…”

Human Morality Is Based on an Early-Emerging Moral Core

“…When the agents had instead been absent, infants chose randomly between the agents. These findings have been conceptually replicated and extended in additional situations involving helping and hindering (Woo et al 2017;Woo & Spelke 2022a, 2022c.…”

The Representation of Third-Party Helping Interactions in Infancy

“…For instance, upon being exposed to an agent taking an object from a patient, infants interpret the agent's action as directed to acquiring a resource, without considering the action's consequences on the object's original possessor ; see also Yin at al., 2022). While adults likely solve this selection problem by applying their knowledge about prototypical goal states (i.e., outcomes that generate direct or derived rewards), this strategy may not be available to infants due to their limited goal repertoire (Liu, Brooks, & Spelke, 2019) and their still developing understanding of means-ends-relations (Woo & Spelke, 2020). This challenge is further compounded for second-order social goals, such as helping, insofar as these actions cannot be construed by a common reference to a single agent's utility (Hobbs & Spelke, 2015).…”