# Spatial Reasoning in LLM Game Agents: Impact of Causal Context and Multi-Step Planning

arXiv: https://arxiv.org/abs/2607.22732 · HTML: https://mlsanity.com/paper/2607.22732/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Spatial and embodied reasoning, one of 17 directions in LLM reasoning.
- Builds on 22 of its 33 references that are in the map.
- First posted on arXiv 22 Jul 2026.

## Placed in

- LLM reasoning / [Spatial and embodied reasoning](https://mlsanity.com/lr/spatial-and-embodied-reasoning/)
- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 22 of 22)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games](https://mlsanity.com/paper/2411.13543/) (2024)
- [SmartPlay: A Benchmark for LLMs as Intelligent Agents](https://mlsanity.com/paper/2310.01557/) (2023)
- [GameBench: Evaluating Strategic Reasoning Abilities of LLM Agents](https://mlsanity.com/paper/2406.06613/) (2024)
- [General Video Game AI: a Multi-Track Framework for Evaluating Agents, Games and Content Generation Algorithms](https://mlsanity.com/paper/1802.10363/) (2018)
- [OptimalThinkingBench: Evaluating Over and Underthinking in LLMs](https://mlsanity.com/paper/2508.13141/) (2025)
- [Wait, Wait, Wait... Why Do Reasoning Models Loop?](https://mlsanity.com/paper/2512.12895/) (2025)
- [Deep Reinforcement Learning for General Video Game AI](https://mlsanity.com/paper/1806.02448/) (2018)
- [Measuring General Intelligence with Generated Games](https://mlsanity.com/paper/2505.07215/) (2025)
- [Representation Learning for Grounded Spatial Reasoning](https://mlsanity.com/paper/1707.03938/) (2017)
- [Passive learning of active causal strategies in agents and language models](https://mlsanity.com/paper/2305.16183/) (2023)
- [The SIFo Benchmark: Investigating the Sequential Instruction Following Ability of Large Language Models](https://mlsanity.com/paper/2406.19999/) (2024)
- [ADAM: An Embodied Causal Agent in Open-World Environments](https://mlsanity.com/paper/2410.22194/) (2024)
- [GVGAI-LLM: Evaluating Large Language Model Agents with Infinite Games](https://mlsanity.com/paper/2508.08501/) (2025)
- [Causal-aware Large Language Models: Enhancing Decision-Making Through Learning, Adapting and Acting](https://mlsanity.com/paper/2505.24710/) (2025)
- [SpaRC and SpaRP: Spatial Reasoning Characterization and Path Generation for Understanding Spatial Reasoning Capability of Large Language Models](https://mlsanity.com/paper/2406.04566/) (2024)
- [From Code to Play: Benchmarking Program Search for Games Using Large Language Models](https://mlsanity.com/paper/2412.04057/) (2024)
- [Curious Causality-Seeking Agents Learn Meta Causal World](https://mlsanity.com/paper/2506.23068/) (2025)
- [From Gameplay Traces to Game Mechanics: Causal Induction with Large Language Models](https://mlsanity.com/paper/2602.00190/) (2026)

## Built on by (most cited first, 1 of 1)

- [Compiling VGDL into Causal Models](https://mlsanity.com/paper/2609.05459/) (2026)

## Abstract (arXiv)

LLM-based game agents often perform poorly on more complex tasks. This work examines whether these failures are linked to limited spatial reasoning and evaluates whether causal prompt augmentation and multi-step planning can improve win-rates while managing response latency. Using the open-source Qwen3 model family, we conduct experiments across varying model scales, reasoning modes, and planning horizons. We further introduce a focused GVGAI benchmark consisting of three custom games with five difficulty levels to isolate spatial navigation. The evaluation follows two paradigms: an initial ``positioning experiment'' to test an agent's ability to find its exact coordinates, and a study of game-play success. Our results show that while larger models with an enabled thinking mode identify their positions more accurately, overall performance in coordinate matching remains limited for smaller models. Win rates decrease as game levels and layout complexity increase, validating the benchmark's difficulty scaling. Integrating causal context into the prompts tends to improve the agents' success rates, particularly for bigger models. While enabling thinking mode and longer planning horizons significantly improve performance, multi-step planning further reduces mean per-step response times, offering a practical trade-off between reasoning depth and execution speed.