paper

Artificial Emotions

A proposed agent architecture uses learned desirability markers to guide responses.

Rodrigo Ventura, Luís Custódio, and Carlos Pinto-Ferreira ask whether emotion-inspired mechanisms can improve machine decision-making. The original subtitle is “Good Bye Mr. Spock!”[1]

Method and contribution

The proposed architecture combines elaborate cognitive representations with simpler perceptual assessments. A simplified implementation attaches learned desirability markers to recalled images and updates them with environmental input. The reported examples illustrate recall, adaptation, and fading associations.[1]

Alignment relevance and limits

The paper connects Machine Intelligence with the problem of selecting useful action under time constraints. Calling an output a “feeling” describes the architecture’s inspiration; the authors explicitly avoid equating it with human subjective experience. The implementation omits the proposed indexing mechanism and provides illustrative behavior rather than a broad comparative safety evaluation. It neither establishes that emotions are necessary for all intelligence nor that emotion-like agents are aligned. The URL’s 2015 upload path does not establish original publication year.

Sources

  1. Artificial Emotions · Source record src-018 · Back to claim ↑1 ↑2

Pages that link here

Last updated 2026-10-07