par Di Stasi, Romain ;Kabdebon, Claire;Delhaye, Quentin ;De Heering, Adélaïde
Référence UNI General meeting 2025 (10th: 2025-05-16: Belgium)
Publication Publié, 2025-05-16
Poster de conférence
Résumé : Introduction Curiosity sparks exploration, and through exploration, learning starts in the earliest stages of life (Kidd & Hayden, 2015). We also know from computational models that novelty plays a critical role in shaping curiosity (Oudeyer et al., 2016) but not always, which is very ecological since we would otherwise be totally overwhelmed in environments saturated with new information (Oudeyer et al., 2016). For this project, we will specifically explore which aspects of novelty most effectively trigger infants’ curiosity in a context where they have to solve a task, and what are the cognitive and physiological mechanisms supporting this.To explore so, we start from the observation that 8-month-old infants show greater curiosity for stimuli presenting an intermediate level of uncertainty — a phenomenon known as the Goldilocks effect (Kidd et al., 2012, 2014) — but which has never been applied to any problem-solving task. About the underlying mechanisms connecting curiosity to novelty, we also know that emotional arousal enhances curiosity when it reaches an intermediate intensity (Berlyne, 1966). Similarly, a stimulus elicits surprise and fosters curiosity when it is neither too strong nor too weak (Berlyne, 1978).MethodsFor this project, 18- to 24-month-old infants will observe an experimenter pressing the top button of a cube 8 times, triggering a sequence of sounds. The cube will then be handed back to the infant, now silent. This action configuration will then be repeated with four additional cubes, each producing the series of sounds with different probabilities (0, 1, 4, or 8 out of 8 presses) to manipulate uncertainty. Infants' curiosity will be indexed via exploratory (e.g., varied button presses) and exploitative (e.g., repeated pressing) behaviours as proxies of their curiosity (Berlyne, 1966). The same infants’ emotional arousal and surprise will also be assessed via an automated analysis of their facial expressions powered by custom algorithms. In particular, the amount of emotional arousal will be quantified across multiple facial muscle groups based on Action Unit (AU) intensity scoring (Messinger, 2002). Instead, surprise will be evaluated via an algorithm capable of detecting raised eyebrows, wide-open eyes and an open mouth (Camras et al., 2002). PredictionsWe hypothesize that, in accordance with the Goldilocks effect (Kidd et al., 2012, 2014), infants will explore more, hence be more curious, when interacting with the cube of intermediate uncertainty (4/8), compared to the highly predictable (0/8, 8/8) or the highly unpredictable (1/8) cubes. If so, this would mean that uncertainty not only drives infants' curiosity within novel contexts (Kidd & Hayden, 2015; Oudeyer et al., 2016) but that it does so, at a maximal level, at intermediate stages of certainty. Similarly, we expect intermediate levels of emotional arousal and surprise in the 4/8 condition, which would suggest that infants’ curiosity is maximal when surprise and arousal are moderate.DataPreliminary data is currently being collected.