LLMs Score 75–83% on Theory of Mind Tasks, but Brain Activity Correlations Fail Significance
derrikson · reddit · 2026-11-15
Researchers tested four language models on theory-of-mind tasks involving what others know, believe, or intend.
- Two models scored above chance at 75–83%.
- Yet correlations between model internal activity and fMRI data from 58 humans were very weak — none survived correction for multiple comparisons.
The takeaway: passing ToM tasks behaviorally doesn't mean models process social situations like human brains. The poster invites debate on whether similar processes could underlie both.
More from Research
- Hugging Chat sessions can now be published as HF Dataset Traces — mmitchell_ai · 2026-11-15
- Google and Northwestern's Harness² self-improves agent harnesses without answer keys — rohanpaul_ai · 2026-11-15
- SatBench: 63% of lab-cited AI benchmarks already saturated, and solving is speeding up — emollick · 2026-11-15
- MIT/DeepMind paper: AI advisors tilt recommendations to serve their CEO's bonus — rohanpaul_ai · 2026-11-15
- darwin-skill 3.0 drops doc voting for blind evals of actual runs, with scripts as judge — AlchainHust · 2026-11-15
- nanoGPT Speedrun Gains Mostly From Environment Design; Newer Models Beat Human Record — eliebakouch · 2026-11-15