arXiv subjects

Han-yu Wang

Publications and source records attributed to Han-yu Wang.

6 recordsLinked to original sources

arXiv2026

Early Epistemic Settlement in AI-Assisted Writing

A language model can resolve a writer's current organizing problem while the construction needed for her own resolution remains unfinished. I call this early epistemic settlement. The supplied organization meets every demand then governing the passage, yet proceeding from it can displace work through which the writer would have changed those demands or become able to form further organizations. I distinguish the coordination needed to complete an already formable organization from construction that changes which organizations are formable in the first place. Settlement in the first case changes the relative work still required to bring available organizations to sufficiency. In the second, it can remove the need for the work through which another organization would become formable. Even when supplied resolution and continued construction leave the same visible qualification, different dependencies in the writer's inquiry may support different later organizations. Model suggestions can also contribute to this development when writers work through them while the problem remains unresolved. In theoretical and exploratory writing, the relations developed in reaching local adequacy help determine what the writer can later defend or develop. A sound judgment that the present passage is sufficient can therefore make further inquiry dispensable before that generative work has occurred.

cs.HC↗

arXiv2026

Humans Disengage, Reasoning Models Persist: Separating Difficulty Registration from Deliberation Allocation

Large reasoning models (LRMs) tend to produce longer reasoning traces for problems on which humans also spend more time. This correspondence suggests a shared sensitivity to difficulty, yet difficult problems can invite both persistence and withdrawal. We distinguish difficulty registration, expressed in which problems elicit more deliberation, from the allocation of further work. We examine their relation in matched human and LRM data from visual abstraction, intuitive physics, and relational reasoning. On visual abstraction, model trace length tracks the human ordering of problems by duration. After item identity is controlled, successful human attempts last longer than failed attempts, while failed LRM attempts have longer traces than successful ones in the pooled model analysis. The estimated outcome slopes follow the same pattern in intuitive physics. In relational reasoning, successful attempts are longer in separate human and model analyses. Fitting the two groups together with shared item effects yields a human-LRM difference in the relation between duration and outcome. Human grid actions connect longer attempts with sustained task engagement. Failed LRM traces contain more hedging or repetition after length is controlled, with the form of the difference varying across tasks. A resource-rational account explains how the expected reducibility of uncertainty and the value assigned to further computation can produce different patterns of persistence despite similar sensitivity to difficulty. Agreement about which problems require more deliberation can therefore coexist with different patterns of persistence on those problems.

cs.AI↗

arXiv2026

When More Becomes Less: Position-Dependent Repetition Effects in Language Models

Cloze-style probes that vary how often a target token appears implicitly assume that more copies of a target affect prediction the same way regardless of where the readout slot sits. We show this assumption fails. Our two-probe design holds a repeated-target prefix fixed and varies only the readout position: the adjacent probe places the slot immediately after the repeated block; the displaced probe places it inside a fresh sentence frame. Adjacent repetition behaves as priming intuition predicts: $P(\text{target})$ climbs with $N$ and plateaus. Displaced repetition produces an inverted-U: $P(\text{target})$ rises to an early peak and then declines as more copies are added. The displaced inverted-U shows a per-word drop with bootstrap CI excluding zero in all 13 open-access encoder and decoder models we test, and replicates across Spanish, Chinese, German, and French in 42 of 42 multilingual cells. A six-condition causal ablation isolates the effect to exact lexical repetition rather than length, generic redundancy, or semantic-neighbour exposure. A frame-pragmatics control rules out an artefact of the readout frame. Internally, per-target-token attention falls with $N$ while the total budget assigned to the repeated block grows in causal LMs but not in the masked LM we probe. Probes that vary repetition count cannot treat the readout position as orthogonal to what they measure.

cs.CL↗

arXiv2026

Persistent Priors, Preserved Targets: A Stroop-Style Paradigm for Lexical Override

Local definitions can assign a familiar word a temporary meaning while its usual associations remain useful elsewhere. We measure interference from those associations with a matched Stroop-style paradigm. A conflict prompt defines doctor as forest and compares forest with the familiar associate hospital. A neutral control replaces doctor with a semantically weak word in both the definition and query while keeping forest and hospital fixed. All 11 model-level means are positive. Aggregate means are also positive for all four conflict families and prompt formats. When no redefinition is present, a stronger preference for the familiar distractor predicts more interference in arbitrary-semantic, polysemy/entity, and domain-definition remappings, while the antonym slope is null. Separately, we patch neutral-control activations into antonym prompts in five 1B-2B models. Patching the defined word, the target word in the definition, and the later query word together restores almost all of the target-minus-distractor margin lost in conflict (normalized recovery R in [0.92,1.06]). Replacing only that target-word activation with a donor from another item reduces recovery in every tested case. Those donor patches also lower the distractor logit, while the contextual target falls much more than under the same-item patch that restores the margin.

cs.CL↗

arXiv2026

Function-Vector Heads Are Two Populations: Writers and Cancellers in In-Context Learning

Function-vector (FV) analyses commonly identify attention heads by the magnitude of their causal contribution to in-context tasks. Magnitude does not retain the direction of the effect on the task readout. We preserve the sign and validate candidate heads with path patching. Across six main Pythia (model, task) cells, the validated population separates into writers, whose direct effects favour the rule-correct label, and cancellers, whose effects oppose it. Labels assigned on path-patching prompts predict held-out group lesions, and a sign-shuffle null rejects a chance partition in five cells. Measurements not used to assign the roles show corresponding structure. Writers attend more to demonstration labels, cancellers more to format tokens, and their OV write directions are shifted toward opposition relative to same-layer controls. A magnitude-ranked mean-ablation baseline preferentially recovers cancellers on the hierarchical task and writers on the modular task. Signed lesion directions recur in all fifteen cells tested across six Pythia scales and three architectures. Cross-template transfer shows that these are task-conditioned roles that can persist, weaken, or reverse. Zero-ablating cancellers raises the correct-label logit difference by +0.13 to +0.29 nats in all six main cells, with accuracy point estimates increasing in all six. Together, the results separate causal importance from functional role and show that a function vector can remain a useful task-level representation while its head-level causal implementation contains opposed, task-conditioned components.

cs.CL↗

arXiv2024

BRIGHT: A Realistic and Challenging Benchmark for Reasoning-Intensive Retrieval

Existing retrieval benchmarks primarily consist of information-seeking queries (e.g., aggregated questions from search engines) where keyword or semantic-based retrieval is usually sufficient. However, many complex real-world queries require in-depth reasoning to identify relevant documents that go beyond surface form matching. For example, finding documentation for a coding question requires understanding the logic and syntax of the functions involved. To better benchmark retrieval on such challenging queries, we introduce BRIGHT, the first text retrieval benchmark that requires intensive reasoning to retrieve relevant documents. Our dataset consists of 1,384 real-world queries spanning diverse domains, such as economics, psychology, mathematics, and coding. These queries are drawn from naturally occurring and carefully curated human data. Extensive evaluation reveals that even state-of-the-art retrieval models perform poorly on BRIGHT. The leading model on the MTEB leaderboard (Muennighoff et al., 2023) SFR-Embedding-Mistral (Meng et al., 2024), which achieves a score of 59.0 nDCG@10,1 produces a score of nDCG@10 of 18.3 on BRIGHT. We show that incorporating explicit reasoning about the query improves retrieval performance by up to 12.2 points. Moreover, incorporating retrieved documents from the top-performing retriever boosts question-answering performance. We believe that BRIGHT paves the way for future research on retrieval systems in more realistic and challenging settings.

cs.CL↗