Bear or Ball, What Will She Grab? Infants’ Comparison and Prediction of Others’ Goal-Directed Actions

• In Event: 1-145 - Mechanisms Underlying the Development of Action Understanding in Infancy

Integrative Statement

By 12-months, an early understanding of others’ intentional behaviour is emerging and infants are at the cusp of understanding and performing tool-reach actions. For example, around this age, infants learn to pull on a blanket to draw a toy situated on the end of it closer (Gerson et al., 2015; Krogh-Jespersen & Woodward, 2018; Sommerville et al., 2005, 2008). In our current work, we used an eye-tracking paradigm to investigate how initial action understanding is generalized to new contexts. We propose that comparisons between familiar and unfamiliar actions facilitate prediction of never before seen actions.

In our experimental condition, 12-month-olds (N = 18) viewed a series of actions that shared a matching goal structure that could be compared: a hand pulling a toy car followed by the hand using different tools to pull the car (Figure 1a). Controls (N = 18) only viewed one of the two tool-use actions, without the opportunity to compare these with a familiar hand-reach (Figure 1b). An eye-tracking paradigm was then used to assess infants’ predictions about a new tool-use action. After being familiarized with a hand using a claw to reach for one of two toys (Figure 1c), the placement of the toys was swapped and infants’ predictive saccades were measured with regards to where they expected a hand to reach (Figure 1d). Fixations to the familiarised toy or its corresponding claw were considered accurate goal predictions, whereas looks to the previous location of the familiarised object (where the unfamiliarised object was newly situated) or its corresponding claw were considered inaccurate. 

Average predictive trials per infant were calculated and compared between conditions. The difference in accurate goal predictions between experimental and control conditions was marginally significant (p = .07). Infants in the experimental condition were descriptively more accurate in their predictions than controls (Figure 2). Our experimental condition facilitated infants’ generalization of action prediction when viewing a novel tool-reach action. Though the marginal findings require caution in interpretation, this research supports the hypothesis that comparisons help infants apply their existing knowledge to understand novel actions, aiding the development of social cognition.

In addition to adding to our sample size and replicating this marginal finding, ongoing research will address how individual differences in infants’ motoric ability and in parent-child interactions relate to action predictions. Active motor experience is known to facilitate action understanding and prediction (Gerson et al., 2015; Krogh-Jespersen & Woodward, 2018; Sommerville et al., 2005). We therefore expect to find relations between proficiency in tool-use and action predictions when assessing infants’ capacities to carry out these actions. We further expect that parent-child interactions that include large exaggerated motion, slow movements, and high interactivity (i.e., motionese) will relate to action prediction via boosting infants’ knowledge about commonalities and intentionality within goal-directed actions (Brand et al., 2002, 2007; Williamson & Brand, 2014). Data collection for this research is ongoing and this new data will allow us to speak to both motoric and social mechanisms that contribute to the development of goal understanding.

Authors

• Charlotte Findlay, Cardiff University
  Presenting Author

• Sarah A Gerson, Cardiff University
  Non-Presenting Author