Infants learn to follow an object’s motion with their gaze – if it features eyes

• In Event: 3-127 - Gaze Following in Infancy: Contextual Influences and Developmental Implications

Integrative Statement

Infants follow other people’s gaze from early on in development. While some have argued that instrumental learning plays a key role for gaze following development, others stress evolved mechanisms that might not rely heavily on individual learning experiences. Here, we tested typically developing 4-month-old infants with an interactive eye-tracking paradigm, providing direct feedback on infants’ looking behavior. We explored whether infants would learn to follow the turning of a non-social “agent” through reinforcement learning and, if yes, whether this would depend on the stimulus featuring eyes.

In three experiments, blocks were used as central cues that rotated toward the left or right side (see Figure 1). In Experiment 1 the “eyes” on the front of the block were contrast-reversed, i.e. white pupil on a black background, while the contrast of the “eyes” in Experiments 2 and 3 resembled that of actual eyes.

General procedures were identical in all experiments: In a baseline phase (4 trials), infants’ initial following of the turning motion was assessed while the block rotated toward one of two comic mice on each side of the screen. In Experiments 1 and 2 this was followed by an interactive training phase (8 trials). Now, every time the infant followed the turning motion of the block, the mouse started wiggling, thereby reinforcing the “gaze-following.” Finally, the test phase was identical to the baseline. Experiment 3 served as a yoked control condition for Experiment 2, in which an independent sample observed the same presentations as infants in Experiment 2, but without contingency. Thus, we tested whether the engaging wiggling of the cued mouse would suffice to reinforce “gaze following” without contingency to infants’ behavior.

The difference score between relative looking times to the cued and uncued mouse was the dependent variable. In none of the experiments did infants spontaneously follow the motion of the animated block toward one of the mice during baseline, all p's > .05. The same was true for the test phase when the block with contrast-reversed eyes was used (Experiment 1, N=20). However, for the block with schematic eyes (Experiment 2, N=21), the difference score in the test phase had a mean of 0.17 (standard error: 0.07), indicating a significant preference for the cued over the uncued mouse when tested against 0, t(20) = 2.2, p = 0.038. When contrasted directly, the difference score during test was significantly greater than the difference score during baseline, t(20) = -2.4, p = 0.028. In Experiment 3 (N=21), the difference score between cued and uncued mouse during test was also positive, mean of 0.18 (standard error: 0.08), only just failing to be significantly different from 0, t(20) = 2.0, p = 0.051.

Results suggest that reinforcement (apparently even without contingency) can indeed affect infants’ gaze behavior, but that infants do not readily learn to follow the motion of an object without eye-like features. While speaking to a certain degree of plasticity, findings also stress the importance of eyes when it comes to guiding young infants’ visual attention.

Authors

• Stefanie Hoehl, University of Vienna
  Presenting Author

• Sabina Maria Pauen, Heidelberg University
  Non-Presenting Author

• Christine Michel, Max Planck Institute for Human Cognitive and Brain Sciences
  Non-Presenting Author