Contrast Sensitivity in Multimodal Large Language Models: A Psychophysics-Inspired Evaluation
On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Abstract (arXiv)
Understanding how Multimodal Large Language Models (MLLMs) process low-level visual features is critical for evaluating their perceptual abilities and has not been systematically characterized. Inspired by human psychophysics, we introduce a behavioural method for estimating the Contrast Sensitivity Function (CSF) in MLLMs by treating them as end-to-end observers. Models are queried with structured prompts while viewing noise-based stimuli filtered at specific spatial frequencies. Psychometric functions are derived from the binary verbal responses, and contrast thresholds (and CSFs) are obtained without relying on internal activations or classifier-based proxies. Our results reveal that some models resemble human CSFs in shape or scale, but none capture both. We also find that CSF estimates are highly sensitive to prompt phrasing, indicating limited linguistic robustness. Finally, we show that CSFs predict model performance under frequency-filtered and adversarial conditions. These findings highlight systematic differences in frequency tuning across MLLMs and establish CSF estimation as a scalable diagnostic tool for multimodal perception.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 15 of its 43 references that are in the map. First posted on arXiv 14 Aug 2025.
Builds on
15 of 43 references.
Closest ancestors · 1
Same direction, newest first
Other references · 14
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
- 2023Visual Instruction Tuningcited by 7,792
- 2025Qwen2.5-VL Technical Reportcited by 5,418
- 2024Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preferencecited by 1,151
- 2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,127
- 2025Magma: A Foundation Model for Multimodal AI Agentscited by 138
- 2024What Did I Do Wrong? Quantifying LLMs' Sensitivity and Consistency to Prompt Engineeringcited by 88
- 2023Spatial-frequency channels, shape bias, and adversarial robustnesscited by 28
- 2021Contrast Sensitivity Functions in Autoencoderscited by 13
- 2025Do computer vision foundation models learn the low-level characteristics of the human visual system?cited by 12
- 2024The Art of Deception: Color Visual Illusions and Diffusion Modelscited by 9
- 2024MindSet: Vision. A toolbox for testing DNNs on key psychological experimentscited by 8
Built on by
Citing papers per year: 2026: 1.
1 paper.
Similar, not cited
Close in wording, no citation either way
- 2024A Comprehensive Study of Multimodal Large Language Models for Image Quality Assessment
- 2025Exploring Multimodal Perception in Large Language Models Through Perceptual Strength Ratings
- 2026Seeing or Knowing? Visual Context Sensitivity in Multimodal Large Language Models
- 2024Evaluating and Advancing Multimodal Large Language Models in Perception Ability Lens
- 2025Evaluating Graphical Perception with Multimodal LLMs
- 2025Emergent Bayesian Behaviour and Optimal Cue Combination in LLMs
- 2025Are Multimodal Large Language Models Pragmatically Competent Listeners in Simple Reference Resolution Tasks?
- 2025The Perceptual Observatory Characterizing Robustness and Grounding in MLLMs