# Frame Problem
The frame problem, identified by McCarthy and Hayes (1969), asks how an AI system can efficiently determine what *doesn't* change when an action is taken. If a robot moves a box, the box's location changes—but millions of other facts (the color of the walls, the time of day, gravity) remain the same. How does the system avoid re-checking everything?
Originally a technical AI problem, [[Hubert Dreyfus]] and others argued it reveals a deeper issue: classical AI struggles with relevance—knowing what matters in a situation. Humans effortlessly filter irrelevant information; symbol-manipulating systems cannot. The frame problem became a key argument for [[Embodied Cognition]] and against [[Computational Theory of Mind]]. Modern AI addresses it through heuristics, neural networks, and learned representations.
## The Problem
| Aspect | Challenge |
|--------|-----------|
| Logical | How to represent non-change efficiently |
| Computational | Avoiding combinatorial explosion |
| Philosophical | How do minds determine relevance? |
## References
- McCarthy & Hayes. "Some Philosophical Problems from the Standpoint of AI" (1969)
- Dennett, Daniel. "Cognitive Wheels: The Frame Problem of AI" (1984)
## Related
- [[Hubert Dreyfus]]
- [[Artificial Intelligence (AI)]]
- [[Computational Theory of Mind]]
- [[Symbol Grounding Problem]]