MAP: A Map-then-Act Paradigm for Long-Horizon Interactive Agent Reasoning

TL;DR

The paper introduces MAP, a paradigm that shifts environment understanding before action execution in interactive agents, leading to improved performance across benchmarks.

Contribution

It proposes a novel Map-then-Act framework inspired by human cognition, with three stages: exploration, mapping, and knowledge-augmented execution.

Findings

MAP surpasses baseline performance in 22 of 25 game environments.

Training on MAP-2K dataset outperforms expert traces.

MAP achieves consistent gains across different benchmarks and LLMs.

Abstract

Current interactive LLM agents rely on goal-conditioned stepwise planning, where environmental understanding is acquired reactively during execution rather than established beforehand. This temporal inversion leads to Delayed Environmental Perception: agents must infer environmental constraints through trial-and-error, resulting in an Epistemic Bottleneck that traps them in inefficient failure cycles. Inspired by human affordance perception and cognitive map theory, we propose the Map-then-Act Paradigm (MAP), a plug-and-play framework that shifts environment understanding before execution. MAP consists of three stages: (1) Global Exploration, acquiring environment-general priors; (2) Task-Specific Mapping, constructing a structured cognitive map; and (3) Knowledge-Augmented Execution, solving tasks grounded on the map. Experiments show consistent gains across benchmarks and LLMs. On ARC-AGI-3, MAP enables frontier models to surpass near-zero baseline performance in 22 of 25 game environments. We further introduce MAP-2K, a dataset of map-then-act trajectories, and show that training on it outperforms expert execution traces, suggesting that understanding environments is more fundamental than imitation.