Human-Like Attention? A Psychophysical Comparison of Visual Search in Humans and MLLMs

Abstract

Visual search is a fundamental cognitive ability. This study investigates whether Multimodal Large Language Models (MLLMs) exhibit human-like difficulty signatures in visual search tasks. We compared search performance of humans (n = 1,250) and MLLMs using identical 2D and 3D stimuli across different set sizes. Both groups showed efficient performance in feature searches, most clearly when the target had a unique color, but performance degradation in conjunction searches as set sizes increased. Additionally, we found strong correlations between human and MLLM error rates ($\rho =0.82$), which suggests that MLLMs are sensitive to similar objective complexities, such as stimulus heterogeneity. However, differences were found as well: whereas humans invested extra search time to respond accurately on target-absent trials, MLLMs exhibited extreme present/absent response biases in complex searches. We conclude that MLLMs replicate high-level human performance signatures, yet their underlying computations differ significantly.

References in corpus (13)

• A Survey on Multimodal Large Language Models
• Gemma 3 Technical Report
• Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting
• An Introduction to Vision-Language Modeling
• Blending Is All You Need: Cheaper, Better Alternative to Trillion-Parameters LLM
• Human Cognitive Benchmarks Reveal Foundational Visual Gaps in MLLMs
• Attention Alignment Between Humans and Vision-Language Models
• Do vision-language models search like humans? Reasoning tokens as a reaction-time analog in classic visual-search paradigms
• Exploring Perceptual Limitation of Multimodal Large Language Models
• From Cognition to Computation: A Comparative Review of Human Attention and Transformer Architectures
• I Spy With My Model's Eye: Visual Search as a Behavioural Test for MLLMs
• Visual Structures Helps Visual Reasoning: Addressing the Binding Problem in VLMs
• The Geometry of Representational Failures in Vision Language Models