Reproduce “A Computational Model of Early Word Learning from the Infant's Point of View” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Satoshi Tsutsui, Arjun Chandrasekaran, Md Alimoor Reza, David Crandall, Chen Yu

A computational model developed by researchers at Indiana University and Max Planck Institute for Intelligent Systems demonstrates that a deep learning network can successfully map object names to visual objects using raw egocentric video and gaze data collected directly from infants. The model achieved up to 39.27% accuracy in ideal sustained attention scenarios, indicating that both sustained attention and object size independently contribute to learning efficacy.

Key insights

  • Raw infant egocentric video data, combined with gaze information, provides sufficient sensory input for a computational model to perform name-object mapping, validating its ecological relevance for early word learning.
  • Infants' sustained attention on a single target object during naming events mechanistically enhances learning by providing perceptually clearer and more consistent visual input.
  • Visual properties of objects, specifically their size in the infant's visual field, independently influence word learning efficacy beyond the effects of attentional focus.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.