Attentional Processes of Selective Learning

• In Event: 1-086 - Cognitive Mechanisms of Selective Learning

Integrative Statement

Transmission of knowledge is highly dependent on social learning, as we continuously acquire novel information from people around us. To do so successfully, we use prior experiences to judge whether people are trustworthy. The sensitivity for trustworthiness develops early in life: Already at 14 months, children selectively learn from competent over incompetent others (Zmyj, Buttelmann, Carpenter, & Daum, 2010). However, to date, little is known about cognitive processes underlying this selective learning. One possibility for the observed differences in learning could be, that during a learning phase, attention allocation to novel information is different when coming from a competent model compared to an incompetent model, leading to a lower learning outcome.

To test this hypothesis, we conducted a combined behavioral and eye-tracking study with children at four years of age. Forty children were familiarized with either a competent (labeling a chair a chair) or an incompetent (labeling a chair a table) model. Afterwards, in a novel information phase, all children were taught novel words from the same model. In a subsequent recall phase, the children were asked to choose the objects matching the novel words. We expected differences in allocation of attention mostly during presentation of novel information and focused our current analysis on the novel information phase. To assess differences in processing of information from the two models, we applied recurrence quantification analysis (RQA) to detect complex structures in eye-tracking data collected during the novel information phase. We assessed the RQA value of determinism of fixation patterns to the AOIs of target object, distractor objects and model. Determinism estimates the probability of transitions between different AOIs and is a measure for predictability of looking patterns. Learning success was assessed by correct object choice in the recall phase.

We analyzed the data using cumulative link models for learning success and linear mixed effect model for allocation of attention. Children in the competent group were slightly better at recalling novel words compared to children in the incompetent group (estimate = -0.3927, se = 0.2146, p = 0.067). Furthermore, the groups differed in their degree of determinism (estimate = 0.08, se = 0.026, p = 0.002), in that children in the incompetent condition showed higher determinism towards the model. Finally, we found an interaction between competence and determinism predicting learning success (estimate = -12.97, se = 5.7, p = 0.023) showing that lower determinism towards the model led to increased learning in the incompetent, but not the competent group. Thus, more flexibility in engaging and disengaging looks towards the model indicates higher probabilities to correctly encode the novel information.

We conclude that experience with a models’ competence does influence how children observe a model during presentation of novel information, leading to a difference in learning success. These results are contrary to suggestions in the literature stating, attention is not an influencing factor of selective learning (Sabbagh & Shafman, 2009).

Authors

• Julia Brehm, University of Zurich
  Presenting Author

• anja gampe, university of zurich
  Non-Presenting Author

• Moritz M Daum, University of Zurich
  Non-Presenting Author