UniCog: Uncovering Cognitive Abilities of LLMs through Latent Mind Space Analysis

TL;DR

UniCog introduces a novel framework analyzing LLM cognition through a latent space, revealing shared reasoning cores and ability-specific signatures, and improving reasoning performance with a latent-informed strategy.

Contribution

The paper presents UniCog, a unified latent space framework for interpreting LLM cognition, uncovering a Pareto principle and enhancing reasoning accuracy.

Findings

Shared reasoning core in LLMs identified

Ability-specific signatures discovered in latent space

Latent-informed strategy improves reasoning performance by 7.5%

Abstract

A growing body of research suggests that the cognitive processes of large language models (LLMs) differ fundamentally from those of humans. However, existing interpretability methods remain limited in explaining how cognitive abilities are engaged during LLM reasoning. In this paper, we propose UniCog, a unified framework that analyzes LLM cognition via a latent mind space. Formulated as a latent variable model, UniCog encodes diverse abilities from dense model activations into sparse, disentangled latent dimensions. Through extensive analysis on six advanced LLMs, including DeepSeek-V3.2 and GPT-4o, we reveal a Pareto principle of LLM cognition, where a shared reasoning core is complemented by ability-specific signatures. Furthermore, we discover that reasoning failures often manifest as anomalous intensity in latent activations. These findings opens a new paradigm in LLM analysis, providing a cognition grounded view of reasoning dynamics. Finally, leveraging these insights, we introduce a latent-informed candidate prioritization strategy, which improves reasoning performance by up to 7.5% across challenging benchmarks. Our code is available at https://github.com/milksalute/unicog.