Typicality Effect

The typicality effect describes the phenomenon where some instances of a category are judged as more representative than others, mirroring human cognitive processes. Current research investigates how well machine learning models, particularly deep learning architectures like BERT, CLIP, and various vision and language models, capture this effect, often comparing their predictions to human judgments across diverse concepts and modalities. This research aims to better understand the alignment between human and artificial concept representations, with implications for improving model interpretability and advancing the development of more human-like AI systems.