Modeling Object Knowledge from Child Visual Experience
eScholarship (California Digital Library) · 18.06.2025
DOI: 10.48448/z630-0920 ↗Kurz gefasst
Fachartikel in der Zeitschrift eScholarship (California Digital Library), erschienen am 18.06.2025. Verfasst von Cognitive Science Society 2025 und 2 weiteren Autor:innen.
Abstract (Originalsprache)
The distributional approach to language has been helpful in understanding and making predictions about children’s semantic and linguistic development. In the current research, we apply similar techniques to study children’s conceptual development using their first-person visual experiences. This study investigates the distributional properties of objects encountered in early visual experience and how they may contribute to the learning of concept organization. Frames were extracted from head-mounted camera videos at regular intervals, segmented into objects, and manually annotated with their superordinate and subordinate categories. We then built a distributional model of object–image co-occurrences and computed the similarity of different objects based on their distributional statistics. We show that objects’ distributional patterns would allow children to make useful predictions about objects’ high-level semantic categories, such as foods, appliances, and electronic devices. These results highlight that early distributional experiences may facilitate category formation, with implications for developmental theory and computational modeling.
Schlagwörter
Zitieren
APA
2025, C. S. S., Ozcan, R., & Willits, J. A. (2025). Modeling Object Knowledge from Child Visual Experience. eScholarship (California Digital Library). https://doi.org/10.48448/z630-0920
Vancouver
2025 CSS, Ozcan R, Willits JA. Modeling Object Knowledge from Child Visual Experience. eScholarship (California Digital Library). 2025. doi:10.48448/z630-0920
BibTeX
@article{2025,
author = {2025, Cognitive Science Society and Ozcan, Rojda and Willits, Jon A.},
title = {Modeling Object Knowledge from Child Visual Experience},
journal = {eScholarship (California Digital Library)},
year = {2025},
doi = {10.48448/z630-0920}
}
EUVIMED