Lune

EMNLP Findings2024

Characterizing LLM Abstention Behavior in Science QA with Context Perturbations

Bingbing Wen, Bill Howe, Lucy Lu Wang

Abstract

The correct model response in the face of uncertainty is to abstain from answering a question so as not to mislead the user. In this work, we study the ability of LLMs to abstain from answering context-dependent science questions when provided insufficient or incorrect context. We probe model sensitivity in several settings: removing gold context, replacing gold context with irrelevant context, and providing additional context beyond what is given. In experiments on four QA datasets with six LLMs, we show that performance varies greatly across models, across the type of context provided, and also by question type; in particular, many LLMs seem unable to abstain from answering boolean questions using standard QA prompts. Our analysis also highlights the unexpected impact of abstention performance on QA task accuracy. Counter-intuitively, in some settings, replacing gold context with irrelevant context or adding irrelevant context to gold context can improve abstention performance in a way that results in improvements in task performance. Our results imply that changes are needed in QA dataset design and evaluation to more effectively assess the correctness and downstream impacts of model abstention. 1

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

6 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Large Language Models Can Be Easily Distracted by Irrelevant Context cites Self-Consistency Improves Chain of Thought Reasoning in Language Models
  • Large Language Models Can Be Easily Distracted by Irrelevant Context cites Large Language Models are Zero-Shot Reasoners
  • Large Language Models Can Be Easily Distracted by Irrelevant Context cites Finetuned Language Models are Zero-Shot Learners
  • Large Language Models Can Be Easily Distracted by Irrelevant Context cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
  • Self-Consistency Improves Chain of Thought Reasoning in Language Models cites Large Language Models are Zero-Shot Reasoners
  • Large Language Models are Zero-Shot Reasoners cites Self-Consistency Improves Chain of Thought Reasoning in Language Models
  • Self-Consistency Improves Chain of Thought Reasoning in Language Models cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
  • Chain-of-Thought Prompting Elicits Reasoning in Large Language Models cites Self-Consistency Improves Chain of Thought Reasoning in Language Models
  • Large Language Models are Zero-Shot Reasoners cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
  • Textual Bayes: Quantifying Prompt Uncertainty in LLM-Based Systems cites Self-Consistency Improves Chain of Thought Reasoning in Language Models
  • Textual Bayes: Quantifying Prompt Uncertainty in LLM-Based Systems cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
  • Hallucinate Less by Thinking More: Aspect-Based Causal Abstention for Large Language Models cites Self-Consistency Improves Chain of Thought Reasoning in Language Models
  • Hallucinate Less by Thinking More: Aspect-Based Causal Abstention for Large Language Models cites Large Language Models are Zero-Shot Reasoners
  • Finetuned Language Models are Zero-Shot Learners cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
  • Bridging the Knowledge-Prediction Gap in LLMs on Multiple-Choice Questions cites Large Language Models are Zero-Shot Reasoners
  • Bridging the Knowledge-Prediction Gap in LLMs on Multiple-Choice Questions cites Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers6

  1. Textual Bayes: Quantifying Prompt Uncertainty in LLM-Based SystemsBrendan Leigh Ross, Noël Vouitsis, Atiyeh Ashari Ghomi, Rasa Hosseinzadeh et al.ICLR 2026 · 8 citations
  2. Variational Uncertainty Decomposition for In-Context LearningI. Shavindra Jayasekera, Jacob Si, Filippo Valdettaro, Wenlong Chen et al.NeurIPS 2025 · 7 citations
  3. Hallucinate Less by Thinking More: Aspect-Based Causal Abstention for Large Language ModelsVy Nguyen, Ziqi Xu, Jeffrey Chan, Estrid He et al.AAAI 2026 · 2 citations
  4. Bridging the Knowledge-Prediction Gap in LLMs on Multiple-Choice QuestionsYoonah Park, Haesung Pyun, Yohan JoICML 2026 · 1 citation
  5. Can LLMs replace Neil deGrasse Tyson? Evaluating the Reliability of LLMs as Science CommunicatorsPrasoon Bajpai, Niladri Chatterjee, Subhabrata Dutta, Tanmoy ChakrabortyEMNLP 2024

Builds on17

  1. Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
  2. Large Language Models are Zero-Shot ReasonersTakeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo et al.NeurIPS 2022 · 8,168 citations
  3. Finetuned Language Models are Zero-Shot LearnersJason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu et al.ICLR 2022 · 4,966 citations
  4. Large Language Models Can Be Easily Distracted by Irrelevant ContextFreda Shi, Xinyun Chen, Kanishka Misra, Nathan Scales et al.ICML 2023 · 970 citations
  5. Self-Consistency Improves Chain of Thought Reasoning in Language ModelsXuezhi Wang, Jason Wei, Dale Schuurmans, Quoc V. Le et al.ICLR 2023 · 681 citations

Related papers

  1. LLMs (Almost) Never Abstain Under Medical UncertaintyAlessio Cocchieri, Luca Ragazzi, Giuseppe Tagliavini, Gianluca MoroACL 2026
  2. Sufficient Context: A New Lens on Retrieval Augmented Generation SystemsHailey Joren, Jianyi Zhang, Chun-Sung Ferng, Da-Cheng Juan et al.ICLR 2025
  3. MASH: Modeling Abstention via Selective Help-SeekingMustafa Omer Gul, Claire Cardie, Tanya GoyalICML 2026 · 2 citations
  4. Beyond "I Don't Know": Evaluating LLM Self-Awareness in Discriminating Data and Model UncertaintyJingyi Ren, Ante Wang, Yunghwei Lai, Xiaolong Wang et al.ACL 2026 · 1 citation
  5. MediQ: Question-Asking LLMs and a Benchmark for Reliable Interactive Clinical ReasoningShuyue Stella Li, Vidhisha Balachandran, Shangbin Feng, Jonathan Ilgen et al.NeurIPS 2024 · 215 citations
Dusk over the sea between two cliffs drawn in fine vertical lines