CUBE-SD - Website This repository hosts the materials, resources, and latest updates for the CUBE-SD research project, which investigates how uncertainty shapes curiosity across development, species, and computational systems.
The project is structured around three main research questions:
-
Mechanisms in human infants
What are the underlying mechanisms linking uncertainty and curiosity in early development?
To address this question, we adopt a multidimensional approach combining behavioral measures and physiological data (e.g., via wearable sensors). In particular, we investigate the roles of surprise and arousal in shaping exploratory behavior. -
Evolutionary origins of the Goldilocks effect
What is the evolutionary basis of the Goldilocks effect—the tendency to preferentially explore stimuli of intermediate complexity (i.e., neither too predictable nor too uncertain)?
This axis compares multiple human and non-human primate species to better understand the phylogenetic origins of this phenomenon. -
Computational modeling
How can the relationship between uncertainty, curiosity, and its underlying mechanisms (e.g., surprise and arousal) be formalized in computational terms?
We develop models based on reinforcement learning and Bayesian frameworks to capture these dynamics.
This repository includes several components of the project, including:
Infant Facial Analysis for the Classification of Emotion (iFACE) is an algorithm currently under development that aims to automatically classify infant facial expressions (e.g., happiness, anger, fear, surprise, sadness, disgust, neutral).
The objective of iFACE is to provide a tool for quantifying emotional and attentional responses in infants, particularly to investigate:
What role does surprise play in the relationship between uncertainty and curiosity?
🚧 This project is ongoing. Methods, datasets, and models are under active development.