Infant understanding of agency and prediction of goal-directed actions: Associations between looking time and predictive gaze
Final Report Abstract
Human adults are very apt in understanding the goals and intentions that underlie other persons’ behaviors, although these mental states cannot be directly observed but have to be inferred, e.g., from the goal-directed actions performed by the other. This social-cognitive ability is very important because it enables the observer to explain and to anticipate the behavior of other persons. Although there is evidence that infants can detect the goal-directedness of others’ actions already in their first year, the mechanisms by which infants ascribe agency (i.e., the ability to perform goal-directed actions) or by which infants cognitively process or predict others’ action goals are not well understood. Thus, the major goal of this research project was to further understand the development of cognitive processes that underlie the attribution of goal-directedness to others’ actions and the prediction of action goals in human infants. There are two main measures to assess action understanding in preverbal infants. Looking-time studies have revealed that when infants repeatedly observe a goal-directed action (e.g., a hand grasping a ball on the left side of the screen), they (from 6 months of age) typically look longer to an action event displaying a change of goal (e.g., grasping a toy bear on the left side) than to a change of the path or trajectory (e.g., grasping a ball on the right side). This pattern of looking times is taken to indicate that infants realize the goal-directedness of the observed action. Eyetracking studies have revealed that when infants repeatedly observe a goal-directed action (e.g., a hand reaching for and grasping a toy), they (from 11 months of age) can perform predictive gazeshift, that is, they shift their gaze from the moving hand to the goal object before the hand arrives at the goal. In the present project, we used both measures to investigate how specific aspects of information (e.g., which agent performs the movement, in which way does the agent move, how salient and/or predictable is the action goal) influence infants’ cognitive processing of observed goal-directed actions. Study Series 1 comprised several looking-time studies with 9-month-old infants. Infants were presented with 3-D animation films showing goal-directed motions of a nonhuman agent (i.e., a ball). In several experimental conditions, we systematically varied behavioral cues that may signal agency and goal-directedness. The important finding was that the 9-month-olds looked longer to a goal-change than to a path-change (indicting the attribution of goal-directedness), when the nonhuman agent displayed the agency cues self-propelledness (i.e., the ability to move on its own), obtained a salient action-effect, and remained visible next to the goal. This supports socalled cue-based accounts of infant action understanding, which claim that the presence of behavioral cues may indicate agency and thus trigger infants’ attribution of goal-directedness, independent from the agent’s external appearance. Study Series 2 combined the measurement of looking times and of predictive gaze-shifts in 9-, 12-, and 24-month-olds. The results confirmed previous research indicating that the two measures assess different aspects of cognitive action processing, which develop at different age. Looking-times are measured when the action has been finished, so infants have complete action information, which may facilitate goal attribution. Predictive gaze-shifts are measured while the movement is still ongoing, so the action information is incomplete, making goal prediction more difficult. Accordingly, looking-times indicated that the 9-month-olds attributed goal-directedness to the observed actions, but it was not before 24 months that infants produced predictive gaze-shifts. Moreover, our research showed that infants’ ability to perform goal-directed gaze shifts is influenced by context information, such as the relative informativeness of the location (e.g., was the goal on the left or right side) or the identity (e.g., was the goal a duck or a penguin) of the goal object. Study Series 3 comprised several eyetracking studies with 12-month-old infants. Here, we found that infants performed goal-directed predictive gaze shifts when a human hand grasped a large, as compared to a small, goal object (impact of goal salience), and when a human hand repeatedly grasped the same out of three object, as compared to grasping of different objects (impact of goal certainty). Infants did not show predictive gaze-shifts when observing the same grasping movements performed by a mechanical claw (impact of agent familiarity). This confirmed theoretical accounts claiming that infants’ prediction of action goals is based on their own action experience, and it added new insights into the cognitive mechanisms that underlie infants’ understanding of other persons’ actions.
Publications
- (2012). Goal salience affects infants’ goaldirected gaze shifts. Frontiers in Psychology, 3, Article 391
Henrichs, I., Elsner, C. Elsner, B., & Gredebäck, G.
(See online at https://doi.org/10.3389/fpsyg.2012.00391) - (2014). Behind the curtain of action understanding: What influences infants‘ predicttive gaze during the observation of goal-directed actions? PhD thesis. University of Potsdam
Henrichs, I.
- (2014). Goal certainty modulates infants’ goal-directed gaze shifts. Developmental Psychology, 50(1), 100-107
Henrichs, I., Elsner, C., Elsner, B., Wilkinson, N., & Gredebäck, G.
(See online at https://doi.org/10.1037/a0032664) - (2016). Goal saliency boosts infants’ action prediction for human manual actions, but not for mechanical claws. Infant Behavior & Development, 44, 29-37
Adam, M., Reitenbach, I., Papenmeier, F., Gredebäck, G., Elsner, C., & Elsner, B.
(See online at https://doi.org/10.1016/j.infbeh.2016.05.001)
