Overhang Tower: Resource-Rational Adaptation in Sequential Physical Planning
Summary: arXiv:2604.09072v1 Announce Type: new
Abstract
Humans effortlessly navigate the physical world by predicting how objects behave under gravity and contact forces, yet how such judgments support sequential physical planning under resource constraints remains poorly understood. Research on intuitive physics debates whether prediction relies on the Intuitive Physics Engine (IPE) or fast, cue-based heuristics; separately, decision-making research debates deliberative lookahead versus myopic strategies. These debates have proceeded in isolation, leaving the cognitive architecture of sequential physical planning underspecified.
Introduction
How physical prediction mechanisms and planning strategies jointly adapt under limited cognitive resources remains an open question. In a recent study, we explored this dynamic interaction using a construction task known as Overhang Tower, where participants are required to maximize horizontal overhang while ensuring stability.
Key Findings
Our research revealed several critical insights:
- Dual Transition Under Resource Pressure: Participants exhibited a dual transition in response to resource constraints, which included shifts in both physical prediction mechanisms and planning strategies.
- Intuitive Physics Engine (IPE) Dominance: In the early stages of the task, participants relied heavily on IPE-based simulation techniques to predict the behavior of physical structures.
- CNN-Based Heuristics: As task complexity increased, participants transitioned to using convolutional neural network (CNN)-based visual heuristics, which allowed for quicker decision-making.
- Time Pressure Effects: Time constraints significantly truncated deliberative lookahead, prompting participants to adopt shallower planning horizons.
Discussion
The findings from this study challenge the prior notion that physical prediction and planning strategies operate independently or through a single mechanism. Instead, we propose a hierarchical and resource-rational architecture that dynamically adjusts to trade-off between computational cost and predictive fidelity.
Conclusion
This research provides a unified perspective on two long-standing debates in cognitive science: the simulation versus heuristics debate and the myopic versus deliberative planning debate. Our results suggest that humans possess a dynamic repertoire of cognitive strategies that are reconfigured based on the cognitive budget available during task execution.
Future Directions
Further research is necessary to explore the implications of these findings for understanding cognitive architecture and its application in real-world scenarios. Investigating how these adaptive strategies can be leveraged in artificial intelligence systems may also open new avenues for enhancing decision-making processes in autonomous agents.