A Property Induction Framework for Neural Language Models
This work addresses the question of how language experience shapes conceptual knowledge in AI models, providing insights into their cognitive-like abilities, but it is incremental as it builds on existing research on LMs and property induction.
The authors tackled the problem of whether neural language models can generalize novel property knowledge from one concept to others, similar to human property induction, and found that LMs show an inductive preference based on category membership, indicating a taxonomic bias in their representations.
To what extent can experience from language contribute to our conceptual knowledge? Computational explorations of this question have shed light on the ability of powerful neural language models (LMs) -- informed solely through text input -- to encode and elicit information about concepts and properties. To extend this line of research, we present a framework that uses neural-network language models (LMs) to perform property induction -- a task in which humans generalize novel property knowledge (has sesamoid bones) from one or more concepts (robins) to others (sparrows, canaries). Patterns of property induction observed in humans have shed considerable light on the nature and organization of human conceptual knowledge. Inspired by this insight, we use our framework to explore the property inductions of LMs, and find that they show an inductive preference to generalize novel properties on the basis of category membership, suggesting the presence of a taxonomic bias in their representations.