Conceptual and Unbiased Reasoning in Language Models

arXiv:2404.00205v1 Announce Type: new
Abstract: Conceptual reasoning, the ability to reason in abstract and high-level perspectives, is key to generalization in human cognition. However, limited study has been done on large language models' capability to perform conceptual reasoning. In this work, we bridge this gap and propose a novel conceptualization framework that forces models to perform conceptual reasoning on abstract questions and generate solutions in a verifiable symbolic space. Using this framework as an analytical tool, we show that existing …

abstract arxiv bridge capability cognition cs.cl framework gap however human key language language models large language large language models novel perspectives reason reasoning study type unbiased work