Unfolding Spatial Cognition: Evaluating Multimodal Models on Visual Simulations

TL;DR

This paper introduces STARE, a benchmark for evaluating multimodal models on complex visual spatial reasoning tasks, revealing current models' limitations in multi-step visual simulation compared to humans.

Contribution

The paper presents STARE, a comprehensive benchmark for assessing multimodal models on visual spatial reasoning, highlighting their struggles with multi-step visual simulations.

Findings

Models perform well on simple 2D transformations.

Models struggle with complex 3D and multi-step tasks.

Humans outperform models significantly in accuracy and speed.

Abstract

Spatial cognition is essential for human intelligence, enabling problem-solving through visual simulations rather than solely relying on verbal reasoning. However, existing AI benchmarks primarily assess verbal reasoning, neglecting the complexities of non-verbal, multi-step visual simulation. We introduce STARE(Spatial Transformations and Reasoning Evaluation), a benchmark designed to rigorously evaluate multimodal large language models on tasks better solved through multi-step visual simulation. STARE features 4K tasks spanning foundational geometric transformations (2D and 3D), integrated spatial reasoning (cube net folding and tangram puzzles), and real-world spatial reasoning (perspective and temporal reasoning), reflecting practical cognitive challenges like object assembly, mechanical diagram interpretation, and everyday spatial navigation. Our evaluations show that models excel at reasoning over simpler 2D transformations, but perform close to random chance on more complex tasks like 3D cube net folding and tangram puzzles that require multi-step visual simulations. Humans achieve near-perfect accuracy but take considerable time (up to 28.9s) on complex tasks, significantly speeding up (down by 7.5 seconds on average) with intermediate visual simulations. In contrast, models exhibit inconsistent performance gains from visual simulations, improving on most tasks but declining in specific cases like tangram puzzles (GPT-4o, o1) and cube net folding (Claude-3.5, Gemini-2.0 Flash), indicating that models may not know how to effectively leverage intermediate visual information.