Emotion is an optimizer's control plane, not an irrational advisor
mimi10v3 · x · 2026-09-04
The author argues that describing rational choice as evaluation over alternatives hides the harder question: where alternatives come from. A mind needs a proposal distribution over thoughts to make some memories retrievable and some futures imaginable.
Emotion, on this account, is a temporally extended meta-control configuration that integrates bodily state, opportunity/danger, learned value, uncertainty, agency and social meaning, then retunes several systems at once:
- Salience: what gets attention
- Search: which memories and futures get sampled
- Precision: which signals are trusted
- Learning: which prediction errors update the model, and how much
- Memory: what gets consolidated
- Action readiness: approach, avoid, freeze, persist, explore, or seek help
- Computation: cached habits vs. model-based deliberation
Conclusion: emotion is neither the utility function nor an irrational advisor — it is closer to an optimizer's control plane, changing the effective objective, sampling process, learning rate, and compute allocation.
More from Research
- Researcher: the best papers answer nothing and raise more questions — yacinelearning · 2026-09-04
- IBM releases DRACO: dynamic rubrics give per-step credit assignment for verifier-free agent RL — ibm-research · 2026-09-04
- The Last Translation Benchmark debuts with multimodal examples that break leading translation models — Vilém Zouhar · 2026-09-04
- Reed-Solomon List Decoding Breakthrough Resolves Major Open Problem in Coding Theory — AlexKontorovich · 2026-09-04
- AI autoresearchers help crack 30-year-old coding theory problem, soundness up to 68.02 bits — BenBlaiszik · 2026-09-04
- Redditor fine-tunes SDXL on 60 childhood photos to simulate memory recall — uisato · 2026-09-04