# Project Affect

Status: exploratory. Integrated: 2026-10-14.

## Canonical research definition

Project Affect studies emotion and affect as regulatory mechanisms of cognition—how they shape **attention, salience, valuation, memory, motivation, risk perception, and action selection**—and attempts to reproduce those functions in artificial cognitive systems.

The target is **synthetic affective mechanisms**. The question is what computational role affect plays and what changes when that role is deliberately instantiated or perturbed. Affect, cognition, and motivation are interacting subsystems; this does not settle a universal emotion taxonomy.

Emotional self-description alone is anthropomorphic theater, not evidence of regulation or subjective feeling. A future mechanism requires operational measures, manipulable variables, alternatives, controls, failure criteria, and a next decision. No model, dataset, approved experiment, or result is established by this definition.

## Unresolved questions

Which regulatory function should be isolated first? How can its effect be distinguished from prompt wording, explicit reward/incentive changes, extra resources, or style? What would support a claim about human cognition from an artificial regulatory effect? Which interactions with memory, Fiction, Imagination, or Daydream deserve testing after individual functions are characterized?

Read [research status](research-status.md), [Hermes Method](../../lab/hermes-method.md), [project map](../../lab/project-registry.md), and [open questions](../../lab/open-questions-and-next-experiments.md). Affect has no selected execution priority over existing programs.

Source: [discussion S36–S39](../../docs/handovers/2026-10-14-hermes-discussion-source.md#s36); user named Project Affect in S37. [Prior art](../../shared/literature/synthetic-cognition-landscape.md) supports an established surrounding field, not a Hermes result.