Mechanisms of Prompt-Induced Hallucination in Vision-Language Models

• URL: http://arxiv.org/abs/2601.05201v1
• Date: Thu, 08 Jan 2026 18:23:03 GMT
• Title: Mechanisms of Prompt-Induced Hallucination in Vision-Language Models
• Authors: William Rudman, Michal Golovanevsky, Dana Arad, Yonatan Belinkov, Ritambhara Singh, Carsten Eickhoff, Kyle Mahowald,
• Abstract summary: We study the failure mode in a controlled object-counting setting, where the prompt overstates the number of objects in the image.<n>We identify a small set of attention heads whose ablation substantially reduces prompt-induced hallucinations (PIH) by at least 40% without additional training.<n>Our findings offer insights into the internal mechanisms driving prompt-induced hallucinations, revealing model-specific differences in how these behaviors are implemented.
• Score: 58.991412160253276
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large vision-language models (VLMs) are highly capable, yet often hallucinate by favoring textual prompts over visual evidence. We study this failure mode in a controlled object-counting setting, where the prompt overstates the number of objects in the image (e.g., asking a model to describe four waterlilies when only three are present). At low object counts, models often correct the overestimation, but as the number of objects increases, they increasingly conform to the prompt regardless of the discrepancy. Through mechanistic analysis of three VLMs, we identify a small set of attention heads whose ablation substantially reduces prompt-induced hallucinations (PIH) by at least 40% without additional training. Across models, PIH-heads mediate prompt copying in model-specific ways. We characterize these differences and show that PIH ablation increases correction toward visual evidence. Our findings offer insights into the internal mechanisms driving prompt-induced hallucinations, revealing model-specific differences in how these behaviors are implemented.

Related papers

• Two Causes, Not One: Rethinking Omission and Fabrication Hallucinations in MLLMs [31.601057368065877]
  Existing methods, based on the flawed assumption that omission and fabrication hallucinations share a common cause, often reduce omissions only to trigger more fabrications.<n>In this work, we overturn this view by demonstrating that omission hallucinations arise from insufficient confidence when mapping perceived visual features to linguistic expressions.<n>We propose the Visual-Semantic Attention Potential Field, a conceptual framework that reveals how visual evidence to infer the presence or absence of objects.
  arXiv  Detail & Related papers  (2025-08-30T05:47:41Z)
• HEAL: An Empirical Study on Hallucinations in Embodied Agents Driven by Large Language Models [27.72821031361892]
  We present the first systematic study of hallucinations in large language models performing long-horizon tasks under scene-task inconsistencies.<n>Our goal is to understand to what extent hallucinations occur, what types of inconsistencies trigger them, and how current models respond.
  arXiv  Detail & Related papers  (2025-06-18T02:13:41Z)
• Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling [78.78822033285938]
  Vision-Language Models (VLMs) excel at visual understanding but often suffer from visual hallucinations.<n>In this work, we introduce REVERSE, a unified framework that integrates hallucination-aware training with on-the-fly self-verification.
  arXiv  Detail & Related papers  (2025-04-17T17:59:22Z)
• Mitigating Low-Level Visual Hallucinations Requires Self-Awareness: Database, Model and Training Strategy [53.07517728420411]
  We introduce the first instruction database specifically focused on hallucinations in low-level vision tasks.<n>We propose the Self-Awareness Failure Elimination (SAFEQA) model to improve the perception and comprehension abilities of the model in low-level vision tasks.<n>We conduct comprehensive experiments on low-level vision tasks, with the results demonstrating that our proposed method significantly enhances self-awareness of the model in these tasks and reduces hallucinations.
  arXiv  Detail & Related papers  (2025-03-26T16:05:01Z)
• Towards a Systematic Evaluation of Hallucinations in Large-Vision Language Models [57.58426038241812]
  Large Vision-Language Models (LVLMs) have demonstrated remarkable performance in complex multimodal tasks.<n>These models still suffer from hallucinations when required to implicitly recognize or infer diverse visual entities from images.<n>We propose a novel visual question answering (VQA) benchmark that employs contextual reasoning prompts as hallucination attacks.
  arXiv  Detail & Related papers  (2024-12-29T23:56:01Z)
• H-POPE: Hierarchical Polling-based Probing Evaluation of Hallucinations in Large Vision-Language Models [0.0]
  We propose H-POPE, a coarse-to-fine-grained benchmark that assesses hallucinations in object existence and attributes. Our evaluation shows that models are prone to hallucinations on object existence, and even more so on fine-grained attributes.
  arXiv  Detail & Related papers  (2024-11-06T17:55:37Z)
• VideoHallucer: Evaluating Intrinsic and Extrinsic Hallucinations in Large Video-Language Models [59.05674402770661]
  This work introduces VideoHallucer, the first comprehensive benchmark for hallucination detection in large video-language models (LVLMs) VideoHallucer categorizes hallucinations into two main types: intrinsic and extrinsic, offering further subcategories for detailed analysis.
  arXiv  Detail & Related papers  (2024-06-24T06:21:59Z)
• Quantity Matters: Towards Assessing and Mitigating Number Hallucination in Large Vision-Language Models [57.42800112251644]
  We focus on a specific type of hallucination-number hallucination, referring to models incorrectly identifying the number of certain objects in pictures. We devise a training approach aimed at improving consistency to reduce number hallucinations, which leads to an 8% enhancement in performance over direct finetuning methods.
  arXiv  Detail & Related papers  (2024-03-03T02:31:11Z)
• Plausible May Not Be Faithful: Probing Object Hallucination in Vision-Language Pre-training [66.0036211069513]
  Large-scale vision-language pre-trained models are prone to hallucinate non-existent visual objects when generating text. We show that models achieving better scores on standard metrics could hallucinate objects more frequently. Surprisingly, we find that patch-based features perform the best and smaller patch resolution yields a non-trivial reduction in object hallucination.
  arXiv  Detail & Related papers  (2022-10-14T10:27:22Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.