Mind Control of LLMs via Closed-Loop Stimulation: Out-of-Distribution Robustness
Abstract
Brains and large language models need not be considered equivalent systems to reveal an interpretable commonality: both exhibit high-dimensional, nonlinear dynamics whose behavior can only be modeled imperfectly. This creates a shared challenge whenever we seek to intervene in their dynamics. A controller may either rely on a model identified around a particular operating regime and risk failure as the system moves away from that regime, or explicitly account for uncertainty and a changing internal state. Neurostimulation has confronted an analogous problem for decades. The limitations of fixed, open-loop stimulation have motivated closed-loop and state-dependent approaches that condition intervention on the ongoing state of the neural system. Inspired by this principle, we formulate LLM steering as a feedback-control problem in which the network's internal state, rather than a fixed intervention alone, determines how the system is perturbed. We further account for the mismatch between tractable dynamical models and the nonlinear transformer they approximate through robust control. As one principled realization of this framework, we develop an H∞-based steering controller that limits sensitivity to disturbances and linearization mismatches. Across multiple steering benchmarks, the controller improves performance over existing baselines, with particularly pronounced gains under out-of-distribution conditions and varying length complexities. Beyond controller performance, we introduce robust steerability, Srob, a pre-deployment measure of tolerance to dynamics mismatch; across model-task pairs, it exhibits task-specific structure and varies systematically with model scale. Beyond robust LLM steering, our measure asks not merely whether control is possible, but how much uncertainty it can survive; a principle that may span artificial and biological neural systems.
Related
- Exploring the robustness of permutation entropy analysis to differentiate between closed-eyes and open-eyes resting states
- CLEAR-FORCE: A reservoir computing framework for closed-loop brain stimulation
- Closed-Loop Acoustic Stimulation of Slow Oscillations Using a Wireless EEG Headband: A Polysomnography-Validated Feasibility Study.
- BOWIe: Wireless Headstage for Optical Stimulation in Closed-Loop Optogenetics Experiments in Freely-moving Rodents
- Beyond the attractor: closed-loop stimulation design for learning multistable neural dynamics
BibTeX
@misc{fallah2026mind,
title = {{Mind Control of LLMs via Closed-Loop Stimulation: Out-of-Distribution Robustness}},
author = {Kasra Fallah and Rudramani Singha and Hannah Ghanei and Paul Sajda and James Anderson and Erfan Zabeh},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.06418},
eprint = {2611.06418},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.06418},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Mind Control of LLMs via Closed-Loop Stimulation: Out-of-Distribution Robustness
AU - Kasra Fallah
AU - Rudramani Singha
AU - Hannah Ghanei
AU - Paul Sajda
AU - James Anderson
AU - Erfan Zabeh
PY - 2026
DA - 2026/11/04
PB - arXiv
DO - 10.48550/arxiv.2611.06418
UR - https://arxiv.org/abs/2611.06418
AB - Brains and large language models need not be considered equivalent systems to reveal an interpretable commonality: both exhibit high-dimensional, nonlinear dynamics whose behavior can only be modeled imperfectly. This creates a shared challenge whenever we seek to intervene in their dynamics. A controller may either rely on a model identified around a particular operating regime and risk failure as the system moves away from that regime, or explicitly account for uncertainty and a changing internal state. Neurostimulation has confronted an analogous problem for decades. The limitations of fixed, open-loop stimulation have motivated closed-loop and state-dependent approaches that condition intervention on the ongoing state of the neural system. Inspired by this principle, we formulate LLM steering as a feedback-control problem in which the network's internal state, rather than a fixed intervention alone, determines how the system is perturbed. We further account for the mismatch between tractable dynamical models and the nonlinear transformer they approximate through robust control. As one principled realization of this framework, we develop an H_∞-based steering controller that limits sensitivity to disturbances and linearization mismatches. Across multiple steering benchmarks, the controller improves performance over existing baselines, with particularly pronounced gains under out-of-distribution conditions and varying length complexities. Beyond controller performance, we introduce robust steerability, S_rob, a pre-deployment measure of tolerance to dynamics mismatch; across model-task pairs, it exhibits task-specific structure and varies systematically with model scale. Beyond robust LLM steering, our measure asks not merely whether control is possible, but how much uncertainty it can survive; a principle that may span artificial and biological neural systems.
ER -