# Can Centaur truly simulate human cognition? The fundamental limitation of instruction understanding

*Source* · `src-liu-2025-centaur-instruction-understanding` · status: **n/a**

| field | value |
|---|---|
| kind | Peer-reviewed paper |
| authors | Wei Liu, Nai Ding |
| venue | National Science Open 5(1) |
| date | 2025-12-11 |
| url | https://www.nso-journal.org/10.1360/nso/20250053 |
| doi | 10.1360/nso/20250053 |
| accessed | 2026-10-24 |

## Notes

With task instructions removed or replaced by misleading ones, Centaur still generally outperformed domain-specific cognitive models; authors conclude it relies on superficial statistical cues rather than understanding instructions.

## Backlinks (computed)

**cited by:** [LLM models of individual human choice](../technique/human-behavior-simulation-models.md)

## Citation

Wei Liu, Nai Ding (2025-12-11). *Can Centaur truly simulate human cognition? The fundamental limitation of instruction understanding*. National Science Open 5(1) https://www.nso-journal.org/10.1360/nso/20250053

---
Provenance: created in run `2026-10-24-techniques-llm-agents`.