- Rating
- 1
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
The paper represents a reasoning trace as a sequence of step-level hidden-state summaries and uses dynamic mode decomposition to extract a spectral fingerprint. It compares fingerprints using a pairwise rank-truncated matching distance and studies whether spectra separate logical templates while remaining relatively stable across topics and languages. On Reasoning-Flow and five language models, the reported inter-logic to intra-logic distance ratios are approximately 2.37–2.67. Layer, rank, and context ablations are provided, together with results on orthogonal invariance and the effects of truncation.
Critical strengths
- Viewing reasoning traces through their spectral structure is an interesting direction and offers a compact way to summarize hidden-state trajectories.
- The invariance to orthogonal changes of coordinates is a useful property of the representation and is clearly motivated.
- The experiments cover several language models and include uncertainty estimates and extraction ablations. This gives the empirical observations some breadth.
Critical weaknesses
- I am not convinced that the proposed distance provides a sound basis for the claimed geometry. The pairwise truncation can discard meaningful spectral differences and even give zero distance to different spectra. Although the appendix acknowledges limitations, the main analysis still relies heavily on this construction. Its suitability needs stronger justification.
- The interpretation of the empirical results seems too strong. Separating logic templates does not by itself show that the method captures internal reasoning dynamics; template structure, trace length, and segmentation are plausible alternative explanations. The current controls do not sufficiently distinguish these possibilities.
- The added value of the dynamical analysis remains unclear. The traces are short, so the reported rank saturation is partly constrained by the number of available steps. Without convincing comparisons to simpler hidden-state summaries or controls that remove temporal order, I find the evidence for a specifically dynamical contribution limited.
Questions for authors
- Does the main separation persist when trajectories are matched for length and rank?
- How sensitive are the conclusions to replacing the truncated distance with full-spectrum optimal transport?
- How does the method compare with static hidden-state summaries, and what happens if the reasoning steps are shuffled?