Marc D. Lewis (2005), Behavioral and Brain Sciences 28(2):169–194 — DOI-10.1017/S0140525X0500004X
Key takeaways
- Against linear appraisal (§2.3, p. 172). Sequential appraisal models, Scherer’s among them, keep emotion out of the causal chain that forms the appraisal. On Lewis’s reading, emotion components act from the start.
- Self-organization (§3.2, pp. 174–175). Positive feedback grows new patterns. Negative feedback stabilizes them, and through entrainment “individual elements or groups of elements lose their independence and become embedded in a larger regime” (p. 174). “The outcomes of self-organization are fixed or periodic (or chaotic) organizational patterns, called attractors” (p. 175), and for an individual “only one of these converges at a time” (p. 175). Jumps between them are “phase transitions or bifurcations” (p. 175).
- The model (§3.3, pp. 175–177). An emotional interpretation (EI) is triggered by a perturbation, followed by nucleation. “For an EI to “take over” the psychological system, a phase of self-amplification must follow” (p. 176). It then self-stabilizes: “Despite the ambiguous nature of the world, and even in the presence of conflicting cues, appraisal–emotion configurations resolve into stable patterns” (p. 177). A superordinate appraisal may “enslave” its constituents (p. 177), after Haken.
- Neural integration (§§5–6, pp. 185–193). Nested loops, neuromodulation and vertical integration. “Vertical integration assures that EIs have some degree of psychological momentum or imperturbability” and “not only maintains but also selects viable EIs” (p. 192). The novel prediction is that “the degree of phase synchrony in the theta range will correlate with measures of emotional relevance and subjective emotional feelings” (p. 186).
- Timescales (§3.3.4, §6.3). Stable EIs drive synaptic change, so “the accumulation of order at two very different time scales is considered interdependent” (p. 177).
- Conclusion (p. 193). “coherent appraisals are not antecedents of emotions, but emerging outcomes of interactions among constituent systems underlying appraisal and emotion.”
Standing in the record
Filed on 2026-10-23 for the record’s proposal that hedonic valence is potential of a negotiation by one ordering (CLAIM-440, TERM-076). Lewis’s account already contains the dynamics of one organization coming to prevail. It has a regime that takes over, persists and is selected over competitors. What prevails, though, is an interpretation, not an ordering, and nothing in it ranks the system’s own states or gives a sign. The bearing is set out in NOTE-1119.
No instruction for machine-learning practice; nothing here belongs in the anthology.
The copy. It was read from a posting by the Geneva E3 Lab, not by the publisher. The owner accepted that posting as a lawful copy on 2026-10-23.