Contribution
A proposal to replace linear appraisal-then-emotion models with a dynamic-systems (DS) account. It is stated first psychologically, then neurally, and joined in a single model. On this account “coherent appraisals are not antecedents of emotions, but emerging outcomes of interactions among constituent systems underlying appraisal and emotion” (p. 193).
Key insight
The unit is the emotion–appraisal amalgam, the “emotional interpretation”, and it is selected, not computed. Many potential configurations compete, and feedback lets one take over the system and hold it.
Assumptions
- That cognition is self-organization in a complex dynamic system: “cognitive systems reconfigure themselves by achieving states of higher order through the synchronization of their constituents” (p. 173).
- A componential view of emotion: “(1) arousal, (2) action tendencies, (3) attentional orientation, and (4) affective feeling” (p. 178), with appraisal components of perception, evaluation, attention, memory and planning.
- That psychological wholes can be mapped onto neural loops whose coupling and synchrony can in principle be measured.
Key results
Argument and model, with neural evidence reviewed:
- Eight DS principles (§3.2): emergent order, positive feedback, negative feedback and entrainment, circular causality, increasing complexity, multistability, phase transitions, interacting timescales.
- Three phases (§3.3): trigger (perturbation, then nucleation), self-amplification (“Positive feedback thus “grows” appraisal–emotion processes that capture the psychological system”, p. 176), and self-stabilization (negative feedback, constraint satisfaction, a superordinate appraisal that may “enslave” its constituents, p. 177).
- Competition and selection: “the consolidation of a particular “solution” from competing ensembles is rapid and nonlinear” (p. 177). Vertical integration gives an EI “psychological momentum or imperturbability” and “selects viable EIs” (p. 192).
- Alternation: “the most realistic portrait of a self-stabilizing EI includes several well-worn interpretations that fluctuate intermittently, as familiar images and emotions drive the system between available attractors” (p. 177). Mr. Smart may oscillate between “angry empowerment versus shameful submissiveness” (p. 192).
- Thresholds set by state: “Baseline states not only determine thresholds for what is meaningful” but constrain what emerges (p. 191).
- Timescales: stable EIs drive synaptic change, so learning is long-term order produced by enduring EIs (pp. 177, 192–193).
- Measure: phase synchrony, “a fixed temporal relation between the oscillations of independent regions” (p. 185). The 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).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Appraisal and emotion constituents cause one another, and whole appraisals emerge from that | moderate | review of emotion-on-cognition effects; DS argument |
| C2 | EIs are attractors, one converging at a time, with phase transitions between them | weak–moderate | argued by analogy with DS systems; appraisal theory’s few canonical profiles |
| C3 | Corticolimbic loops, neuromodulation and vertical integration implement the phases | moderate | neural review (§§4–5), read only in part here |
| C4 | Theta-band phase synchrony indexes emotional relevance and feeling | weak | stated as a novel prediction |
| C5 | Stable EIs are the precondition of emotional learning, so traits grow from states | moderate | LTP and neuromodulation evidence, as reviewed |
Concepts
- emotional interpretation (EI) — an emotion–appraisal amalgam that emerges and stabilizes as a whole.
- self-amplification / self-stabilization — the positive-feedback growth phase and the negative-feedback coherence phase of an EI.
- circular causality — after Haken: a higher-order form constrains the coupling of its constituents, which in turn sustain it.
- vertical integration — coupling across levels of the neuraxis, which gives an EI momentum and restricts which EIs are viable.
Connections
Opposes sequential appraisal models (Scherer, Lazarus, Smith and Kirby), and agrees with Frijda and Zeelenberg that appraisal can be an outcome. Draws on Haken’s synergetics and on Kelso, Thelen and Smith, and Freeman. Cites Carver and Scheier (1990, LIT-1441), the account the record’s sign clause takes (TERM-076), among the few emotion theories with negative feedback. Sander and Scherer’s commentary (pp. 216–217) objects that the account cannot say when an episode begins or ends, and offers Scherer’s criterion, “a certain degree of synchronization of emotion components driven by specific appraisal outcomes” (p. 216). Lewis replies that “Scherer’s model and mine share an emphasis on synchronization in emotion episodes” (p. 225). Scherer’s model is read in LIT-1451.
Bearing on the record
Does Lewis’s attractor account already contain “an ordering coming to dominate”? It contains the dynamics of potential almost exactly, and not the ordering.
- The dynamics. The record’s term reads potential through Frijda’s three marks (THEORY-055). Lewis’s account has all three. Priority: an EI must “take over” the psychological system (p. 176), and its constituents “lose their independence and become embedded in a larger regime” (p. 174). Persistence: vertical integration gives it “momentum or imperturbability” (p. 192). Resumption: an attractor is a state the system returns to after small perturbations, and the “well-worn interpretations that fluctuate intermittently” are returned to (p. 177). It also has what CLAIM-440 needs and Frijda’s marks do not supply. One wins “from competing ensembles” (p. 177), “only one of these converges at a time” (p. 175), and losing is real: Lewis’s oscillating Mr. Smart is the record’s split or alternating potential, its reading of ambivalence. And it has no regulator above the competitors, which is THEORY-056’s point.
- What it lacks. What prevails in Lewis is an interpretation, a configuration of appraisal and emotion constituents. It is not a ranking of options, and so it is not the kind of thing that can be a currency or fail to be one. Nothing in the account ranks the system’s own states by their bearing on its maintenance. Valence enters only as something the orbitofrontal cortex “attends to” in anticipated events (p. 179). There is no sign clause and no account of intensity beyond the degree of synchrony. So the proposal’s two distinctive moves are not in Lewis. One is that what prevails is an ordering, which makes CLAIM-436’s question of currency askable. The other is that the ordering is of a particular kind, which makes the regime valence. Lewis supplies a dynamical mechanism by which such an ordering could come to prevail, and a precedent the record should cite rather than rediscover.
On the regimes of CLAIM-435. Lewis’s dynamics are threshold-rich by design. Effects “may be exponential, subject to threshold effects” (§3.1), transitions are abrupt, and baseline states “determine thresholds for what is meaningful” (p. 191). If the earliest valence were an EI-like attractor, the prediction would be ALARM’s discontinuity, with a threshold that moves with the animal’s state rather than staying fixed. That is a third case beside the claim’s fixed level threshold and its continuous curve: a state-dependent, possibly hysteretic, switch. Like Scherer’s hysteresis (NOTE-1113), it means the trade-off test must control or record the animal’s prior state.
On measurement. Lewis’s phase synchrony is an operational measure, EEG or LFP phase locking, but of coherence among regions, not of which ranking governs action. For QUESTION-268 it is a candidate marker of the potential’s strength, and needs pairing with a behavioural record of which conflicts are won.
On CLAIM-283. In Lewis’s account, the authority a configuration is granted is not a quantity separate from the precedence it claims. It is the degree to which constituents become entrained to it. If that reading holds, the claimed–granted gap is the transient before stabilization, and chronic conflict is oscillation between attractors.
Limitations
The model is a framework, not a fitted dynamical system: no equations, no estimated parameters. Sander and Scherer’s charge that it gives no operational boundaries for an episode is fair for the psychological model. Lewis answers it with the phase model, not with a measure. Only part of the neural review was read here.
Open questions
Whether an EI’s selection among “competing ensembles” can be described as one ranking winning over others, which is the bridge the record needs. And whether theta-band synchrony tracks which of two conflicting tendencies governs behaviour, or only how much is going on.