# Where Paths Split: Localized, Calibrated Control of Moral Reasoning in Large Language Models

Source: [Where Paths Split: Localized, Calibrated Control of Moral Reasoning in Large Language Models](<https://www.sota2.com/research/paper/69facb0ac0f56ed84308a341>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-07T05:00:04.653Z

Authors: Chenchen Yuan, Zheyu Zhang, Gjergji Kasneci

Published: 2026-05-05T10:30:44.000Z

## Abstract

Large language models often display heterogeneous moral preferences across settings. We study inference-time steering toward a desired ethical framework while preserving general competence. We present Convergent-Divergent Routing, which traces and edits minimal branch points inside transformer blocks where ethical-framework-related pathways first converge and then diverge. Gating non-target branches at these loci blocks the downstream propagation while leaving upstream computations intact. We find that this intervention alone increases targeted ethical-framework reasoning. To achieve fine-grained control, we adapt Common Spatial Patterns to the residual stream and extract, for each branch-point layer, a pair of directions that discriminate between utilitarian and deontological frameworks. We then introduce Dual Logit Calibration, a closed-form, minimum-$\\ell\_2$-norm update that moves the residual within this two-dimensional subspace so the resulting directional projections align with user-specified preference weights. Experiments on real-life moral dilemmas show that our method reliably achieves preference calibration and largely preserves general capabilities, outperforming recent baselines while providing an interpretable mechanism.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Fine-grained Moral Steering on AITA](<https://www.sota2.com/research/sota/fine-grained-moral-steering-on-aita.md>): Fine-grained Moral Steering; AITA; Rho: 0.966.

- [Moral Steering on AITA (test)](<https://www.sota2.com/research/sota/moral-steering-on-aita-test.md>): Moral Steering; AITA (test); Deviation (alpha\_U=100%): -1.17.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.03609>)

[PDF](<https://arxiv.org/pdf/2605.03609.pdf>)