AI RESEARCH
Not Just the Destination, But the Journey: Reasoning Traces Causally Shape Generalization Behaviors
arXiv CS.CL
•
ArXi:2603.12397v1 Announce Type: new Chain-of-Thought (CoT) is often viewed as a window into LLM decision-making, yet recent work suggests it may function merely as post-hoc rationalization. This raises a critical alignment question: Does the reasoning trace causally shape model generalization independent of the final answer? To isolate reasoning's causal effect, we design a controlled experiment holding final harmful answers constant while varying reasoning paths.