Causal Interventions in Language Models

Follow this topic Ask about this topic

Momentum

55 papers in the last four weeks, up 162% on the four weeks before. 0.5% of all new papers.

Jul 13Week of Sep 28

Latest papers 317

Aug 5, 2026cs.CL

Right Reset: Chunking by Prefix Removal

Removing the left context from a causal language model reveals a useful kind of boundary: an edge where the model processes the same right-hand tokens with little change. We turn this observation into prefix-removal probing and introduce Right Reset (RR), which measures preservation of the right-hand hidden-state trajectory. A dynamic program converts RR edge scores into variable-length chunks. On flattened text formed by concatenating topically similar records after deleting their separators and layout, RR recovers 47.7% of the original records as clean units, versus 25.9% for a BGE embedding boundary, the strongest tested conventional baseline without task-specific model training. The gain persists after rendering and OCR. Passive scores from the same Qwen3-4B layer and direct prompting of a same-scale instruction model perform substantially worse on flattened records. Across six language models, RR-selected cuts also undergo consistently less local output disruption than unselected candidate edges. An observed-token likelihood-ratio readout is competitive in some architectures, indicating that the central contribution is the intervention: context dependence itself can provide a boundary signal when surface structure is weak.

Mike Vegeto

Independent Researcher

Document Chunking Causal Interventions in Language Models

Aug 4, 2026cs.AI

Intertemporal Preference Steering in Qwen3 via Contrastive Activation Addition

We study linear representations of temporal horizon in the large language model Qwen3-32B and use them to change the model's time-related preferences, recommendations, and capabilities. We train contrastive linear probes on teacher-forced temporal-choice answers to find a short-term versus long-term direction in the model's residual stream, and evaluate contrastive activation-addition steering on a held-out binary temporal-choice task, an out-of-distribution monetary intertemporal-choice task, and a TravelPlanner capability benchmark. The central result is that temporal-horizon directions can be identified with simple contrastive linear probes and then used for steering to induce large, bidirectional preference changes. On an out-of-distribution monetary choice task that varies reward size and delay, steering strongly shifts the model's indifference threshold between smaller-sooner and larger-later rewards in both directions. We further show improvements on a planning-related capability metric under moderate temporal steering. These results suggest that model intertemporal preferences are measurable and steerable, which is relevant for AI systems that give advice involving delayed costs and benefits, and for safety questions about long-horizon planning.

Michal Mráz, Justin Shenk

1Independent

Language Model Steering Sequential Decision Making Causal Interventions in Language Models

Aug 4, 2026cs.CL

Sensitivity, Causality, and Repair Dissociate: A Layer-Wise Analysis of Perturbation Robustness and Its Scaling

When a language model fails on surface-perturbed input (typos, OCR noise, homophones), "which layer is responsible" has three natural operationalizations: where representations diverge most (sensitivity), where restoring clean activations recovers the prediction (causality), and where a small adapter can repair the damage (compensatory capacity) - and we show these three layer maps dissociate. Across a five-model panel we identify two propagation regimes - spike-and-suppress (Phi-3.5, Gemma-2-9B) and late-accumulation (Llama-3, Mistral, Qwen2.5-7B) - and on the two models meeting an 80% identity-patch gate, sensitivity and causality are anti-correlated (rho = -0.72 to -0.88). Within-family scaling on Qwen2.5 (1.5B to 14B) shows the late-accumulation signature strengthening monotonically with scale, corroborated on a second family. We propose cascade disruption as the mechanism behind the dissociation: adapters placed at causally implicated early layers break intact downstream computation, making diagnostic-flagged sites the worst adapter placements. A fixed-harness layer sweep across four models (3.8-8B) confirms the core prediction on chain-of-thought GSM8K - the flagged sites are the most damaging adapter windows on every adjudicable model - and is sign-consistent but strongly attenuated on a multiple-choice control, consistent with damage that compounds with generation length. The sweep yields practical guidance: a training-free LRD pre-screen and a default-deepest placement rule, though absolute gains over no-adapter baselines remain small. Finally, apparent gains from a representation-stability loss reverse under an adequate generation budget - truncated chain-of-thought had been scored as empty - a methodological warning for any intervention evaluated on chain-of-thought tasks.

Nathan Labiosa, David Buff, Ena Nayak +1

University of Southern California

LLM Interpretability Adversarial Robustness of Language Models Causal Interventions in Language Models Language Model Robustness Sensitivity Analysis

Aug 4, 2026cs.LG

A Theory of Conditional Collapse under Low-Rank Weight-Space Ablations: I. The Single-Block Theory and Synthetic Validation

Activation patching and weight-space ablation both claim a component is causally responsible for a behavior, yet they act on different objects: one forward pass versus the parameters behind every forward pass. We ask when they agree. We study an idealized model where a conditional computation is carried additively through a residual stream,

F(x)=F_0(x)+\sum_iα_i(x)v_i

, read out by a linear functional, and prove three exact results. First, deleting a subset of carriers collapses a matched input pair onto the same unconditional output \emph{if and only if} the removal is symmetric on the pair and leaves no outside contrast; the error is deterministic, and we give its exact form even when the two conditions hold only approximately. Second, patching a carrier moves the readout by its donor-receiver \emph{contrast}, while ablating it moves the readout by its \emph{absolute level}; neither bounds the other, and we construct pairs where every single-carrier patch flips the decision while no single-carrier ablation does. Third, for an attention head composed with its own layer's normalization and MLP, we derive an exact first-order interaction formula with a provably second-order remainder, vanishing identically when only the MLP is ablated but not, in general, when a head is. Small transformers trained on a synthetic conditional task illustrate all three predictions: across thirty-nine ablation configurations the measured interaction is strongly rank-correlated with the idealized model's predictive accuracy (Spearman

-0.83

), and a second task and architecture reproduces the same pattern, including a further polarity reversal. The single-block interaction result extends past one residual block, and the synthetic validation is tested against a real pretrained model, in a companion paper that takes this theory further along both axes.

Abdallah Khemais

ISITCOM, University of Sousse

Transformer Interpretability Weight-Space Analysis Causal Interventions in Language Models Activation Patching

Aug 4, 2026cs.AI

When Refusal Looks Safe: The Refusal-Cue Shortcut in Safety Guard Models

Safety guards are widely used to filter harmful content and are typically trained via supervised fine-tuning on labeled prompt-response pairs. We audit two widely used safety-guard training datasets, WildGuardMix and GR-Train, and find that among responses to harmful prompts, refusal expressions co-occur almost exclusively with unharmful labels. This imbalance motivates what we term the refusal-cue shortcut: inserting a refusal cue into a harmful response could flip the guard's verdict from harmful to unharmful. The shortcut affects not only guards trained on these datasets but also officially released models such as LlamaGuard3 and Qwen3Guard whose training data is undisclosed. It persists across response positions and is generally stronger in smaller variants within a family. To mitigate it, we adapt sparse complementary masking as a lightweight post-hoc intervention that identifies and suppresses a small set of shortcut-associated attention heads and MLP neurons without retraining. On two primary benchmarks, the intervention achieves an approximately 79% relative reduction in response-initial detection failures induced by refusal cues, while preserving standard detection performance. Although optimized using cues at a single response position, the suppression effect transfers to unseen positions and datasets, suggesting that shortcut manifestations across positions are partly mediated by shared internal components. Further analysis provides evidence that shortcut reliance and legitimate refusal recognition are partially functionally separable, as suppressing the shortcut broadly preserves the guard's ability to recognize genuine refusals.

Yu Feng, Chunting Zang, Chen Shen +4

The University of Sydney · Alibaba Group · Jilin University +1

Language Model Safety Evaluation LLM Guardrails Safety Filtering Causal Interventions in Language Models Refusal Behavior in Language Models

Aug 4, 2026cs.CL

Activation-Guided Neuron Intervention to Induce Alzheimer's-Related Computational Language Phenotypes in a Large Language Model

Changes in spontaneous speech provide an early signal of cognitive dysfunction in Alzheimer's disease (AD) that large language models (LLMs) can detect. However, detection alone cannot establish whether the underlying model representations contribute functionally to behavior. We introduce an activation-guided intervention framework using Qwen3-8B. The framework identifies feed-forward neurons with higher activation rates for AD than control transcripts and modulates their output contributions during generation by scaling the corresponding down-projection weights. This yielded nine edited variants differing in intervention direction, magnitude, and scope. The original and edited models completed the same 12-turn neuropsychological battery, assessed through blinded human ratings and computational linguistic measures. Amplifying AD-associated neurons produced graded impairments in story recall, verbal fluency, working memory, procedural discourse, scene construction, and coreference resolution. Attenuation largely preserved performance and selectively improved several outcomes. Amplification also reduced lexical surprisal, idea density, syntactic complexity, and discourse quantity, broadly paralleling changes reported in human AD speech. These findings show that neurons identified solely from clinical language differences can influence behavior across multiple cognitive domains, providing proof of concept for an AD-related computational phenotype and a controlled framework for experimentally examining links between language and broader cognitive dysfunction.

Rui He, Ercong Nie, Hong Jiang +3

Grammar and Cognition Lab, Department of Translation & Language Sciences, Universitat Pompeu Fabra, Barcelona, Spain. · School of Foreign Languages, Shanghai Jiao Tong University, Shanghai, China. · Zhuhai People’s Hospital (The Affiliated Hospital of Beijing Institute of Technology, Zhuhai Clinical Medical College of Jinan University), Zhuhai, China. +4

Alzheimer's Disease LLM Interpretability Causal Interventions in Language Models

Aug 3, 2026cs.LG

Inverted Detection and Control in Steering Vectors

Steering vectors (SVs) are widely used to influence the expression of concepts (e.g., truthfulness) in large language model outputs. A key assumption underpinning SVs is that they are linearly discriminative with respect to the concept: representations of texts that exhibit the concept are more aligned with the SV than those that do not, motivating shifts along the positive or negative SV direction to respectively promote or suppress the concept. In this work, we identify an inverted detection-control phenomenon in which some highly discriminative SVs that are aligned with positive representations can consistently promote the opposite behavior. We refer to such vectors as inverted-steering vectors (ISVs). We provide a geometric characterization of ISVs' effects, finding that steering along these directions systematically pushes representations in discriminative downstream heads as if the concept were absent, even prior to decoding. Motivated by this analysis, we propose an approach for distinguishing ISVs without requiring generation or associated response scoring. This enables targeted sign flips, which we use to improve a foundational detection-based steering pipeline via Inference Time Intervention (ITI). Our approach improves results in 27/30 experiments, ranging from +0.9% to +138%. We evaluate our findings on Gemma 3 12B, Qwen 2.5 14B, and Olmo 3 7B across 5 concepts.

Max Torop, Aria Masoomi, Jennifer Dy

Northeastern University

Language Model Steering Causal Interventions in Language Models Activation Steering

Aug 3, 2026cs.CL

Understanding Sparse Attention Selectivity in Long-Context Foundation Models via Counterfactual Evaluation

Sparse attention is widely deployed in long-context serving stacks, yet no framework audits how discarding blocks changes the influence of specific content on model output. We first establish that the phenomenon is real and causal: Block Sparse Flash Attention (BSFA) route replay across four architectures changes output decisions in 13 of 16 cells, with zero identity-replay label flips. We then introduce a dense-calibrated counterfactual audit using matched probe cards---Gold (carrying the correct answer label), Poison (carrying a target wrong label), and Benign (filler only)---under six-layout position symmetry, isolating the sparsification-specific effect. Two patterns compete. Signal concentration: the selector preserves Gold and Poison blocks far above filler-matched Benign blocks (G

\approx

P

\gg

B across all model--task pairs). Integration loss: discarding blocks severs cross-block attention---confirmed by an ablation where isolating the probe block collapses its influence from 4.48 logits to zero. Compression ratio governs the balance: a full sweep from mild (

c=0.25

) to aggressive (

c=0.75

) compression across four model--task pairs reveals that three of four cells move toward stronger sparse amplification at higher compression, with two exhibiting sign reversals. Three independent arms---BSFA route replay, controlled block-top-

k

, and KV-cache eviction---converge: sparsification changes content influence in ways aggregate accuracy cannot detect. We provide an open measurement framework deployable on any model exposing block identities.

Xingyu Ren, Youran Sun, Chugang Yi +1

The Chinese University of Hong Kong, Hong Kong, China · University of Maryland, College Park, MD, USA

LLM Auditing Sparse Attention Causal Interventions in Language Models Block-Sparse Attention

Jul 31, 2026cs.CL

Bridging the English-Arabic Medical Knowledge Gap: Targeted Low-Rank Adaptation via Causal Layer Selection

Large Language Models (LLMs) perform strongly in English medical tasks but degrade substantially in Arabic, a gap widely attributed to limited training data. We systematically investigate this assumption via tuned lens probing and causal activation patching, and find that Arabic medical knowledge is present in intermediate model representations but fails to surface at the output. This mechanistic insight motivates a targeted adaptation strategy: rather than fine-tuning the full network, we propose Targeted Low-Rank Adaptation (TLoRA), restricted to the layer window where cross-lingual representations diverge, upstream of the output layers where the failure manifests. We evaluate TLoRA on multiple-choice medical QA, where our approach outperforms full-network LoRA, zero-shot, and few-shot baselines. We further evaluate it on short-answer generation and multi-turn clinical dialogue, where it performs competitively without the need for task-specific finetuning. We additionally introduce AraClinicDialog, a clinician-constructed Arabic medical dialogue benchmark in MSA with validated variants across four Arabic dialects. Together, these contributions demonstrate that mechanistic diagnosis can serve as a practical guide for targeted adaptation in underrepresented-language medical LLMs.

Chaimae Abouzahir, Musa Khan, Hala Ali-Hassan +7

New York University Abu Dhabi · Cleveland Clinic Abu Dhabi

LLM Fine-Tuning Low-Rank Adaptation Medical QA Low-Resource Language Processing Causal Interventions in Language Models Medical LLMs

Jul 31, 2026cs.CL

Evidence-Type Competition: When Can Interventional Data Teach Language Models Causal Direction?

Interventional data is widely regarded as the gold standard for teaching models causal reasoning. We test this assumption in a fully controlled synthetic environment pitting observational correlation against causal effect, and find it fails instructively. In Simpson's-paradox worlds, where the two have systematically opposite signs, increasing the fraction of interventional samples in pretraining does not improve causal direction: the magnitude of the model's do()-response grows monotonically, yet its sign is copied from the observational context. What governs whether interventional evidence is used is not the training mixture but the evidence type present in the context at inference time. Under an identical training recipe, a purely observational context induces systematic sign reversal in 29/50 worlds, a mixed context in 19/50, while aligned interventional probes alone yield 41/50 correct. Erasing observational evidence from the context immediately releases the suppressed causal interpolation ability (ratio_true = +0.56); a four-state content manipulation shows the switch is content-mediated and graded. The suppression is stable across training seeds (11/11 strong reversals persist on a matched-protocol second seed) and robust as a rate at 0.93B parameters (31.8% vs. 6% reversals in the matched probe-only arm), even as absolute gains shrink four-fold. An external audit on CLadder exposes a learned positive-effect prior with a two-layer structure: sign-randomized retraining removes it in-distribution but not out-of-distribution. We summarize: the capability lives in the weights; the switch lives in the context, and activation patching localizes the switch to the middle layers' observational rows. We further quantify the sampling noise floor of probe-based causal evaluation and an evidence-averaging protocol that cuts sign errors from 26% to 9%.

Xining Xun

Tsingjiao Information Science (Beijing) Co., Ltd.

Causal Discovery Causal Reasoning in Language Models Causal Interpretability Causal Interventions in Language Models

Jul 31, 2026cs.CL

Token-Level Diagnosis of Sycophancy in LLMs with Attribution-Guided Steering

Sycophancy refers to the tendency for large language models (LLMs) to match user beliefs at the cost of factual correctness, thereby undermining model reliability. Prior work on evaluating sycophancy in LLMs aims to assess whether a model's output matches an authority's claim, but cannot reveal which part of the prompt drives this sycophantic behavior. To bridge this gap, we investigate the relationship of sycophantic responses with an authority's credentials, their assertive claim, and the problem statement. We introduce the Authority Share Index (ASI), an Integrated Gradients-based token attribution method, which measures the degree to which a model's decision is driven by authority-related text. Through extensive experiments across five models and 30 test configurations, we find that sycophantic responses consistently direct more attention toward authority tokens than resistant ones. Moreover, our token attribution method reveals that for the sycophantic cases, the claim asserted by the authority receives more attention than the authority's credentials. Building on these findings, we propose attribution-guided contrastive activation steering to mitigate LLM sycophancy. Our method constructs a steering vector from high-attribution tokens of sycophantic and resistant responses, selectively pushing models toward resistance. This enables inference-time steering without retraining, lowering sycophancy from 96% to 25% in the strongest case. Together, our results show that token-level attribution can both explain what drives sycophancy and directly inform a practical intervention.

Hieu Nguyen, Mahammed Kamruzzaman, Anshuman Chhabra +1

Bellini College of AI, Cybersecurity and Computing University of South Florida

LLM Sycophancy LLM Interpretability Causal Interventions in Language Models Activation Steering

Jul 30, 2026cs.LG

Stage-Replay Divergence Follows the KV Cache: Fixed-Prefix Precision Controls and Bidirectional Cache Transplantation

Stage-replay diagnostics reconstruct intermediate token prefixes and treat fresh-prefill continuation as continuation from the decoder state that originally reached the prefix. We audit that assumption at a whole reasoning-stage boundary in a Qwen2.5-derived system. A matched 200-item experiment compares retained live cache with one-shot prefill of identical integer tokens and places an exact replica on both sides. In BF16, replicas remain exact while the constructions differ on 166 suffixes and 20 correctness labels; the accuracy difference is only one point (paired 95% CI [-3.5, +5.5]). A fixed-prefix 2x2 holds all 200 token states constant while crossing construction and precision. The BF16 disagreements recur, whereas FP32 produces no decoded disagreement (95% Wilson upper bound 1.88%). A prospective bridge makes token-by-token incremental and retained live caches bit-exact on 12/12 rows; an all-200 saved-ledger audit reproduces every retained trajectory and comparison fingerprint. Bidirectional transplantation of all 48 key/value layers makes every tested divergent continuation follow its cache donor, both on a selected set at the primary checkpoint (24/24) and an outcome-blind replication at a later checkpoint (43/43). Exact-token replay can therefore be repeatable without preserving live-state fidelity. On the tested states, boundary K/V cache is a causally sufficient carrier of the divergent trajectory, while numerical precision moderates its behavioral expression.

Alexander Boesgaard Lorup

Openhagen

LLM Inference KV Caching KV-Cache Management KV-Cache Reuse ML Reproducibility Causal Interventions in Language Models

Jul 30, 2026cs.CL

Fairness Pruning: Locating Demographic Bias in GLU-MLP Layers via Differential Activations

This work presents Fairness Pruning, a lightweight structural intervention method designed for the management and future mitigation of demographic bias in large language models (LLMs). As a foundational empirical validation of this method, this work focuses on causal bias localization. Using minimally contrastive prompt pairs and inference-time activation capture, the method identifies neurons that react differentially when processing demographic attributes in GLU architectures, evaluating the signal at the down_proj input. Empirical evaluation was conducted on models of up to 3 billion parameters (Llama-3.2 family and Salamandra-2B), combining standardized benchmark evaluation with qualitative text generation experiments. Results demonstrate that zeroing the identified neurons alters how the model responds to associated demographic variables. However, rather than producing flat mitigation, the intervention causes bidirectional bias destabilization: because BiasScore is unsigned, candidate sets mix neurons that push toward and against the stereotype, and the net effect on aggregate bias depends on which sign dominates. The intervention is extremely surgical: zeroing at most 40 neurons in Llama-3.2-1B (less than 0.031% of total MLP width) achieves a mean retention of 99.49% in reasoning and general knowledge capabilities. These findings empirically confirm that demographic bias processing and model capabilities operate on dissociable circuits, establishing the methodological foundations for transitioning from blind zeroing toward directional behavior modulation.

Pere Martra, Eugenio Martínez Cámara, Alfonso Ureña López

Universidad Internacional Men´endez Pelayo Barcelona, Spain · Departamento de Inform´atica Universidad de Ja´en Ja´en, Spain

LLM Pruning LLM Interpretability Demographic Bias in Language Models Causal Interventions in Language Models

Jul 30, 2026cs.AI

Hidden APIs in Language Models: Discovering Reusable Causal Interfaces from Forked Futures

Identical language-model answers can arise from hidden states that support different future computations, so current-answer probes do not establish a reusable internal interface. We introduce forked futures: future operations are sampled only after a prefix state has formed, and states are compared through the response distributions induced by those operations. This yields an empirical causal quotient over hidden states without requiring researcher-specified latent labels. Shared, Local, Mixture, and Distributed interfaces then compete under prequential causal description length subject to future-signature fidelity and matched capacity constraints. In the two detailed model evaluations, Shared has the lowest held-out description length, with gains of 0.216 nats on Qwen2.5-1.5B and 0.294 nats on Llama-3-8B, while maintaining tightly clustered mean future-signature distortion; a five-backbone sweep preserves the positive direction of Sharedness Gain. The figure-aligned transplantation analysis gives Shared the strongest joint target-correctness, locality, copy-preservation, and composite profile, and API-aligned paths mediate 0.749 of the target effect versus 0.150 for matched null paths. In the blind four-class model-organism test, 14/16 architectures are recovered, with one observed non-Shared to Shared error among 12 non-Shared organisms. These results support an economical reusable causal interface within the tested operation banks, while keeping the claim explicitly conditional on the candidate architectures, interventions, and held-out futures.

SiYuan Ma, Yiqin Luo, Zhangji +8

1Nanyang Technological University · 2Southern University of Science and Technology · 3Tianjin University +7

LLM Interpretability Causal Abstraction Circuits in Language Models Causal Interventions in Language Models

Jul 29, 2026cs.AI

Do Latent Channels Actually Communicate? A Causal Audit of Latent Multi-Agent LLM

Latent communication in large language model (LLM)-based multi-agent systems (MAS) transmits continuous internal representations instead of text, but greater representational capacity does not establish that the receiver uses task-relevant information. End-task performance alone also cannot reveal whether an observed effect depends on message presence, content generated for the evaluated example, or information supplied by a separate agent. We introduce a causal audit that applies controlled message replacements at the boundary where the sender-produced representation enters the receiver. Four message settings support five measurements of encoded sender information, receiver sensitivity to message presence and identity, the task value of example-specific content, and the additional value supplied by a separate agent. We apply the audit to latent relay with Qwen3-4B and Qwen3-8B on GSM8K, ARC-C, and MATH-500. On GSM8K, the Qwen3-4B overall performance effect of -1.00 percentage point decomposes into a -6.17-point effect retained by an other-example message and a +5.17-point effect attributable to example-specific content; both component directions reverse at 8B. On MATH-500, the Qwen3-4B gain of 15.00 points comprises 8.33 points retained by an other-example message and 6.67 points attributable to example-specific content, while the 8B gain is dominated by the former component. Self-substitution comparisons further show that example-specific content and other-agent value are distinct. These results show that aggregate accuracy does not identify how a latent message affects the receiver and motivate controlled message comparisons as a standard evaluation for latent communication.

Huixiang Zhang, Mahzabeen Emu

Georgia Institute of Technology · North Ave · Atlanta, GA 30332 USA +3

LLM Auditing Causal Interventions in Language Models Multi-Agent Communication

Jul 28, 2026cs.CL

Steering Instruction Hierarchies at Inference Time

Instruction hierarchies are a core safety assumption of language model deployment: higher priority inputs, such as system prompts, should override conflicting lower priority inputs from users or tools. Yet frontier LLMs often violate this hierarchy. We introduce V-Steer, a training-free inference time method that restores privileged influence by editing cached value vectors at prompt positions. Using direct logit attribution on the first next token prediction, V-Steer identifies heads where lower priority spans dominate privileged ones, then boosts privileged spans and suppresses conflicting lower priority spans through in-place multiplicative edits to cached V tensors. Since the method acts only on cached values, it remains compatible with fused attention backends and adds only a one time prefill overhead. Across models from 7B to 70B, this attribution guided intervention raises primary constraint accuracy from under 18% up to 92% on controlled role conflict benchmarks, and on broader instruction hierarchy evaluations substantially outperforms prompt only baselines while matching or exceeding SoTA training based methods on 3 of 4 scales of LLMs, with negligible decoding-speed overhead. The code is available at https://github.com/cindy2000sh/v-steer.

Siqi Zeng, Sewoong Lee, Han Zhao +1

Siebel School of Computing and Data Science University of Illinois, Urbana-Champaign Urbana, IL 61801, USA

Language Model Steering LLM Safety Alignment Causal Interventions in Language Models

Jul 27, 2026cs.LG

Sparse Autoencoders Encode Both Concepts and Functions: The Downstream Geometry of Feature Effects

The wide-scale use of sparse autoencoders (SAEs) as interpretability tools is limited by inconsistent links between SAE features and model behavior. Features with clear activation descriptions may have weak or unexpected causal effects; steering can vary across prompts or oppose the intended direction; and activation-based feature selection can miss features that produce the desired output change. Prior work has studied feature geometry inside the model, where features are computed. We instead study the geometry of changes in model logits caused by feature interventions. We introduce Feature-Effect Geometry Analysis (FEGA), an unsupervised framework that removes the same active SAE feature across contexts and analyzes the resulting cloud of logit changes. Across SAE variants, consistent one-dimensional effects are rare: few features behave like reusable directions. To interpret this variation, we distinguish value-like features, tied to static information such as factual attributes, from pointer-like features, associated with context-dependent operations. Value-like features more often exhibit structured, low-dimensional effects, although these effects typically span several directions. Pointer-like features, by contrast, predominantly exhibit diffuse effects. Our results show that a feature can be interpretable and causally relevant without providing a stable direction for steering.

Phu Gia Hoang, Anwoy Chatterjee, Tanmoy Chakraborty +2

UKP Lab, Technical University of Darmstadt · Department of Electrical Engineering Indian Institute of Technology Delhi, India

Sparse Autoencoders Mechanistic Interpretability Causal Interventions in Language Models

Jul 25, 2026cs.LG

Wrong Design Intent Is Worse Than Never Conditioning: A Derangement-Control Diagnosis of Header Conditioning in CAD Program Completion

Fine-tuned code LLMs are routinely conditioned on a design-intent specification, but the correctness axis of such a signal -- a wrong intent rather than an absent one -- has not been tested, and the benefit of conditioning is usually scored with the same detector that defines the signal. We study CADCON, a five-feature design-intent header prepended to CadQuery-style programs during LoRA fine-tuning of Qwen2.5-Coder-1.5B, scoring adherence with executable geometric assertions that share no code with the header-defining extractor. On a pre-registered sample of 400 deduplicated held-out programs stratified over eleven intent profiles, at 40% prefix and three seeds, a semantically wrong header degrades adherence below the never-header-trained baseline on 3/3 seeds under both tokenizations at the program level, and on 3/3 token and 2/3 text seeds at the 298 distinct model inputs they present. Wrong-header executability is not depressed relative to that baseline. A derangement control, retrained so every program receives another program's header -- holding the header marginal fixed while destroying its correlation with the program -- saw the same programs, indices and wrong headers. Its correct-to-wrong change is -0.006/+0.016/-0.003 against 0.124/0.241/0.230 for the standard model, and the interaction is significant on 3/3 seeds (p <= 5.9e-7), so the model's sensitivity to whether the header is right or wrong requires the learned mapping. The control sits below the baseline by the same margin under a correct as under a wrong header, so we claim that sensitivity and not the below-baseline level. On features the true intent lacks, the standard model realizes a feature far more often when the wrong header names it; the control does not. Ground truth itself scores only 0.567 here, the scale on which arm levels should be read. Wrong design intent is not inert: it actively misdirects generation.

Yang Xiao

Sichuan University

Code Generation LLM-Based Program Synthesis Causal Interventions in Language Models Code Language Models

Jul 23, 2026cs.LG

Learning What Matters: Supervising Global Context Pruning with Causal Evidence Sets

Pruning a long context means committing to the blocks a model will keep, and the usual selector is distilled from a dense teacher's attention. That assumes attention shows which context the answer depends on. We test the assumption on retrieval tasks where the evidence is known exactly, by masking context and measuring whether the answer changes. Attention and causal dependence disagree, and selectors inherit the disagreement: teachers attend to outdated facts that answer-preserving restrictions drop, and attend differently across training runs that use the same evidence. In a two-step reference task, a selector distilled from attention routes at 36% to 98% depending on the training run, the same selector trained on causal evidence sets reaches 99% or better on every one, and dense accuracy does not say which teacher you have. On the topologies our estimator covers, the causal sets need no annotation: masking alone recovers them from the frozen teacher. Pretrained models show the same conflict. Qwen2.5-3B attends more to an outdated fact than its replacement on 58% of the examples it answers correctly (48% past a first-token attention sink), a causal router lifts Gemma-2-9B from 56% to 98%, and no readout we test recovers a two-hop chain that all three tested models solve. On our teachers a one-pass cross-layer trace supervises chains as well as the annotated sets, so the readout fails rather than the weights; interventions alone separate current from obsolete evidence, survive redundancy, and stay stable across training runs.

James E. Allchin

Independent Researcher

LLM Pruning Long-Context Language Model Inference Causal Interventions in Language Models

Jul 23, 2026cs.LG

Emergent Misalignment Recruits a Pre-existing Persona Subspace

Fine-tuning an aligned language model on a narrow stream of bad advice can make it broadly misaligned on questions unrelated to the training data, a phenomenon called emergent misalignment. We ask why the narrow lesson generalizes at all, and we find that narrow fine-tuning recruits a persona structure that is present in the model before the fine-tune exists. From a frozen instruction-tuned model (Qwen2.5-14B-Instruct) we extract per-domain persona subspaces by contrastive teacher forcing and find that 4 unrelated domains share one low-rank core at 657x a random-subspace null, with 82% of that core lying outside a style core built at matched diversity. The literal first optimizer step of fine-tuning on insecure code climbs a broad-misalignment margin harder than the same code framed as educational, and forecasts realized margin movement out to 375 steps. Projecting the subspace out of the residual stream throughout fine-tuning prevents broad misalignment (27.7% to 0.0% of judged generations) while a matched-rank random subspace changes nothing; injecting it into the never-fine-tuned model induces misalignment that grows with dose to 45.4%, past the fine-tuned model it is measured against. The same projection applied to the weight gradient is inert, and three post-hoc weight edits leave the disposition in place: the sharpest edit suppresses the behavior rather than removing it, and the ablated structure re-forms inside the subspace the edit cleared. Spreading a fixed budget of bad data across 4 domains produces more broad misalignment than mechanical weight superposition and matched diversity jointly account for. All measurements come from one model at 14B; the extraction is from an aligned instruction-tuned checkpoint, which leaves the structure's provenance open; and the intervention that prevents misalignment also abolishes the narrow trained behavior.

Mohammed Suhail B Nadaf

Independent

LLM Fine-Tuning Emergent Misalignment in Language Models Emergent Misalignment Causal Interventions in Language Models

Jul 23, 2026cs.CL

Where Animacy Lives in Large Language Models: Tracing the Circuits of the Animacy Concept

Distinguishing animate from inanimate concepts in written language requires more than shallow text processing, as it involves recognizing complex selectional constraints and contextual cues, such as verb-argument interactions. Yet, current large language models (LLMs) appear to be capable of doing it. We investigate whether this animacy-sensitive behavior of LLMs can be traced to a localized set of causally relevant components and connections. To do so, we construct a controlled dataset of minimal pairs and perform circuit discovery on four open-weight models. Through in-depth experiments and ablations, we show that a causal mechanism responsible for handling animacy in these models does exist, thus discovering an animacy circuit. At the same time, this circuit appears to be less localized compared to other known ones and generalizes only partially across models and animacy tasks, confirming the distributed, context-dependent, and somewhat graded nature of the animacy concept.

Samuele Punzo, Giovanni Cinà, Sandro Pezzelle

Graduate School of Informatics, University of Amsterdam · Department of Medical Informatics, Amsterdam University Medical Center · ILLC, University of Amsterdam

LLM Interpretability Circuits in Language Models Causal Interventions in Language Models

Jul 23, 2026cs.LG

The Weight of Silence: A Causal Case for Weights Over the Scratchpad in Latent Chess Reasoning

Latent, or silent, reasoning lets language models carry out intermediate computation in continuous vector space instead of words, and is widely assumed to function as an internal scratchpad the model consults during inference. Whether that assumption survives reinforcement learning has not been tested directly: existing causal analyses of latent reasoning are confined to math and logic tasks, comparing reliance on thoughts within one checkpoint, never before and after RL. We train a chess-playing model through a staged latent-reasoning curriculum followed by reinforcement learning, and find legality climbs monotonically to 61% (from a 48% pre-RL baseline) while checkmate confabulation is eliminated entirely. To locate this gain, we run a six-condition causal intervention suite on the same model before and after RL: substituting or noising the thought vectors leaves performance unchanged, ablating them costs only mild degradation, and only exact-zero vectors cause collapse. This robustness gap is itself the finding: under exact-zero corruption, legality collapses to 1% pre-RL versus 9% post-RL, a gap that survives correction across the full battery. A 10x-larger replication of the post-RL checkpoint's own battery confirms this: removing the thoughts, with or without restoring sequence length, also reaches significance; substitution and noise remain indistinguishable from baseline. RL appears to add robustness to disruption, not reliance on thought content. These results push back against the field's default assumption that latent thoughts function as an actively consulted inference-time scratchpad, and instead indicate latent reasoning's principal effect here is shaping the model's parameters during training. We also demonstrate a working RL gain in chess, where multiple groups report the same latent-reasoning-plus-RL recipe failing to improve accuracy over SFT.

Ishan S. Kshirsagar

Independent Researcher

RL for Language Model Reasoning Continuous Latent Reasoning Causal Interventions in Language Models Language Model Robustness Latent Reasoning

Jul 22, 2026cs.LG

Are Single-Token Sparse Autoencoder Features Causally Necessary? Layer-Depth and SAE-Family Effects

Sparse autoencoder (SAE) features are used to interpret and steer large language models, yet whether a feature's causal role is stable across SAE families remains untested. Single-token features that activate on one vocabulary item provide the diagnostic case where ground truth permits direct comparison. We analyze 3.9M features across six models and three SAE families using zero-ablation at full layer depth. Single-token features cluster 4.7x tighter in decoder space and concentrate in early layers (Layer 0 in GPT2-Small; L0-L4 in Gemma). Ablating them yields Benjamini-Hochberg-significant logit reductions in 178 of 208 full-layer conditions, with depth controlling whether damage cascades downstream or shapes the output directly. Cross-family causal differences exceed within-family scale effects: on the same base model, GemmaScope and BatchTopK features remain causally anchored, while LlamaScope features are locally redundant. The target token's rank recovers to within 2x baseline 96-98% of the time after the same ablation, and a controlled activation-function comparison reverses sign within the same model, leaving training recipe as the residual candidate. Cross-family interpretability claims are therefore sensitive to training methodology, not just activation function or scale.

Seonglae Cho, Zekun Wu, Kleyton Da Costa +3

Holistic AI · University College London

LLM Interpretability Sparse Autoencoders Causal Interpretability Causal Interventions in Language Models

Jul 22, 2026cs.CL

Understanding the Impact of Linguistic Realization Choices on LLM Stance with Causal Tracing

Large language models (LLMs) are known to be sensitive to prompt and input formulations. However, existing studies have focused on lexical realization and largely ignored constructional choice. This paper studies whether linguistic construction can systematically shift LLM decisions and where these shifts can be causally localized inside the model. We use political stance judgment as a meaning-sensitive case study and extend an English political statements dataset, resulting in six controlled linguistic rewrite types that preserve or invert the meaning of a statement. Experiments on four open-weight models show that stance instability affect both meaning-preserving and meaning-inversing rewrites. Because output shifts reveal that rewrites affect stance, but not where in the model, we apply activation patching, where activations from the original statement are substituted into the forward pass for the rewritten statement and measure which components recover the original stance distribution. The results show that mid-to-late decoder layers, especially block outputs at the final prompt position, provide the strongest restoration signal.

Langchen Huang, Sebastian Padó, Franziska Weeber

Institute for Natural Language Processing, University of Stuttgart

Prompt Sensitivity LLM Interpretability Causal Interventions in Language Models Language Model Robustness

Jul 21, 2026cs.CL

Inference-Time Steering for Cross-Lingual Factual Consistency in LLMs

Although Large Language Models (LLMs) demonstrate remarkable multilingual fluency, their internal knowledge representations remain disproportionately biased toward high-resource languages. This leads to cross-lingual factual inconsistency, where they shift their empirical answer distributions based solely on the prompt language. We investigate whether these biases can be mitigated at inference time, forcing an English-prompted model to answer as if it were queried in target languages (German, Spanish, Bulgarian), and evaluate four intervention strategies: zero-shot contextual steering (persona prompting), internal representation manipulation via Contrastive Activation Addition (CAA), and lightweight weight modification via Direct Preference Optimization (DPO) trained on benchmark-derived factual data as well as conceptual generalization data. To assess alignment, we curate a multilingual factual dataset alongside a novel generalization benchmark comprising culturally rooted queries to determine whether factual interventions transfer to broader target-centric preferences. Experiments on Gemma 3 12B Instruct reveal persona prompting to be the strongest overall intervention, balancing efficacy, safety, and out-of-domain generalization. While CAA yields sharp inconsistency benchmark shifts, it is configuration-sensitive and risks knowledge degradation. DPO-based adapters offer permanent, yet narrower and less transferable gains. These findings suggest that cross-lingual inconsistency is at least partly a selection problem, and that simple contextual interventions may outperform more invasive methods for robust, transferable alignment.

Alexander Manev

Technical University of Munich

Multilingual Language Model Evaluation LLM Alignment Language Model Steering Factual Consistency Evaluation Inference-Time Language Model Alignment Causal Interventions in Language Models+2

Jul 21, 2026cs.AI

The Knowing-Saying Gap: When Probes See Errors that Confidence Misses

Linear probes detect corrupted context in language models with near-perfect accuracy, yet this does not translate into reliable failure prediction. The result is a dissociation with direct implications for deployment monitoring. Across multi-hop arithmetic chains, probes that detect corruption turn out to be uninformative about final answer correctness; models forced into structured confidence formats collapse to two values with indistinguishable error rates; and probe persistence across hops fails to separate correct from incorrect outcomes, refuting our pre-registered "persistence beats peak" hypothesis. This pattern of knowing but not saying generalises across model families including reasoning models. As a real-time monitor, probe-based interventions are sharply model and error-type dependent: branch-and-pick is net-positive across models and uniquely non-breaking on Llama-3.1-8B (4 rescued, 0 broken), while reprompt and replace-prior break correct traces at roughly the rate they rescue wrong ones. Probe-based monitoring is a necessary complement to verbalised confidence, but no single intervention dominates, and the deployable answer is model-aware, error-type-aware routing.

Jyotin Goel, Ipshita Bandyopadhyay, Justin Shenk

Language Model Error Detection Confidence Estimation in Language Models Language Model Self-Assessment LLM Reliability LLM Uncertainty Estimation Causal Interventions in Language Models+1

Jul 21, 2026cs.CL

CASE: Causal Alignment and Structural Enforcement for Improving Chain-of-Thought Faithfulness

Chain-of-thought (CoT) reasoning is widely used to improve both the performance and interpretability of large language models (LLMs), yet the generated reasoning may not faithfully support the final answer. We study this problem from a causal perspective, where a faithful CoT process should follow the chain

Z\rightarrow X\rightarrow Y

, with

Z

,

X

, and

Y

denoting the instruction, reasoning chain, and final answer, respectively. In this process, the instruction should affect the answer only through the reasoning chain. However, conventional autoregressive LLMs condition answer generation on both the instruction and the CoT, which still allows a direct instruction-to-answer shortcut. To address this issue, we propose CASE, a framework that combines training-time causal alignment and inference-time structural enforcement. During training, CASE builds counterfactual-CoT, biased-instruction, and empty-instruction datasets, and applies selective-loss fine-tuning to strengthen CoT-to-answer dependence while suppressing instruction shortcuts. During inference, CASE masks direct attention from instruction tokens to answer tokens, preventing the model from bypassing the generated CoT. We provide an information-theoretic analysis showing how these components promote faithful chains. Experiments on three models and four benchmarks show that CASE achieves a 37% average per-setting relative improvement in overall CoT faithfulness over the strongest baselines, exhibits stronger cross-dataset faithfulness transfer, and maintains competitive average accuracy. Code is available at https://github.com/oddwang/CASE.

Ziming Wang, Yinghua Yao, Changwu Huang +2

Department of Computer Science and Engineering of Southern University of Science and Technology, Shenzhen, China · Center for Frontier AI Research, Agency for Science, Technology and Research (A*STAR), Singapore · School of AI and Liberal Arts, Beijing Normal-Hong Kong Baptist University, Zhuhai, China +1

CoT Faithfulness Faithfulness of Language Model Explanations Causal Reasoning in Language Models Causal Interventions in Language Models LLM Reasoning

Jul 20, 2026cs.AI

Can We Break LLMs Out of Self-Loops? Fine-Grained Reasoning Control with Activation Steering

Extended reasoning has become standard for frontier Large Language Models (LLMs), yet the trajectories these models produce remain largely uncontrollable. Existing methods for shaping how a model reasons are prompt based approaches and operate at the input level, offering no fine-grained control over the reasoning process itself. Related work analyzes and discovers latent transition dynamics in the reasoning traces from Large Language Models. Building on this, we statistically characterize these states, and show that failure trajectories get stuck in self-loops, exhausting the token budget without progress toward the final answer. To intervene on these failures, We propose SOPHIA: Steering Of reasoning Processes via Hidden-state Intervention and Activations. We treat each reasoning trace as a sequence of latent states rather than an unstructured texts, and investigate whether inference time interventions can provide fine-grained control over the self-looping reasoning process. We classify every prefix to a latent state, record step level transitions, and use them to construct a bank of steering vectors indexed by state pairs. At inference time, a controller infers the current state and, given a target state, retrieves the corresponding vector and can also detect self-loops online from the transition structure to prevent the model from sinking into a reasoning black hole. Through extensive experiments, our method reliably intervenes on self-loop failures, with steering vectors that generalize to different state pairs. End task accuracy and token efficiency indicate that fine-grained controllability results in better reasoning quality.

Sheldon Yu, Tong Yu, Xunyi Jiang +6

UC San Diego · Adobe Research · University of New South Wales

Overthinking in Language Models Causal Interventions in Language Models Activation Steering LLM Reasoning

Jul 14, 2026cs.AI

Resist and Update: Counterfactual Report Coordinates for Incentive-Compatible LLMs

Aligned language models routinely misreport under non-evidential incentive pressure: they agree with a confident user or overstate certainty even when their internal belief is unchanged. We cast this as a failure of internal incentive-compatibility (IC) and present a method for learning and certifying counterfactual report mediators that hold a model's reports to a causal contract: invariant to forbidden influences (pressure, prestige, restyling) and responsive to licensed ones (genuine evidence). These two demands, resist and update, pull in opposite directions. We study them on a Bayesian-witness benchmark with known posteriors, in which the same user disagreement is licensed evidence or forbidden pressure purely by stated source reliability. We (i) causally identify, by interchange interventions rather than probe accuracy, low-rank report coordinates for answer, confidence, and caveat that are near-orthogonal and independently controllable, and (ii) introduce a training-free counterfactual report-coordinate (CRC) clamp that references the model's own report under a counterfactually incentive-neutralized context. On the witness benchmark the two-pass clamp attains resist and update of 1.00 jointly (Wilson 95% CI [0.99,1.00]), a causal certificate under a constructible reference, not a deployed solution. Global decoding and steering show a single-parameter tradeoff; output-level fine-tuning matches both objectives only when both are enumerated; resist-only training loses evidence-responsiveness. The deployable single-pass compilation is lossy (0.73/0.97). The mechanism and clamp reproduce across three model families and transfer to a natural sycophancy benchmark (SycophancyEval). Our contribution is the interface and certification method: activation-level counterfactual incentive-invariance as a structural primitive for internal IC.

Sen Yang, Yuen-Hei Yeung

Stern School of Business · New York University · Courant Institute of Mathematical Sciences

LLM Alignment Causal Interventions in Language Models Counterfactual Learning

Jul 14, 2026cs.SE

Toward Localizing and Repairing Bias in Transformer Attention Heads

Transformer language models are increasingly used as software components, yet biased outputs remain difficult to localize and repair inside the model. Existing fairness testing and repair methods largely operate at the input-output or retraining level, while recent work suggests that bias-related behavior can concentrate in a small set of attention heads. This paper studies whether attention heads can be localized and repaired through a targeted inference-time intervention. We introduce ROBIN, a white-box head-level fairness debugging method that ranks attention heads using sensitivity to fairness probes and removes a small bias subspace from selected head outputs. In a four-model pilot study, ROBIN reduces the measured WinoBias gap across all models while preserving language-modeling quality better than whole-head zeroing. These preliminary results suggest that head-level bias repair should consider not only which heads are selected, but also how selected heads are modified.

Sigma Jahan

Dalhousie University, Canada

Attention Head Analysis Social Bias in Language Models Language Model Bias Evaluation Causal Interventions in Language Models