Contribution
A review of the evaluative space model (ESM) by two of its authors and two colleagues. It restates the postulates, answers critics (Russell and Carroll’s bipolarity) with the postulates of oscillation and calibration, reviews behavioural and neural evidence, and draws clinical implications.
Key insight
The output of the affect system can be bipolar while its mechanism is not. Behaviour “may be constrained to a single bipolar dimension”, and the response surface is many-to-one over a bivariate evaluative plane. So states that the output cannot tell apart, neutrality and ambivalence, are different states of the mechanism (§Abstract; Figure 2).
Assumptions
- Positivity goes with appetitive motivation and negativity with aversive. Anger is treated as defensive in its “distal motivation” (§The Evaluative Space Model).
- The affect system is the integrated set of mechanisms for telling hostile from hospitable stimuli. Reportable feeling is not the target.
- The ESM “is not a theory of what elicits affective states or emotions”.
Key results
- Behavioural and neural evidence for separable positive and negative channels: uncorrelated positive and negative affect, distinct striatal and orbitofrontal regions, and distinct transmitter systems.
- Coactivation in the field (graduation day, Larsen et al. 2001), in the laboratory (disappointing wins, Larsen et al. 2004), and by oscillation between opposite images (Norris, Eichenold & Cacioppo 2009).
- A negativity bias across reflexes, attention, ERPs and fMRI, and a positivity offset in mere exposure and person perception.
- Individual differences in both asymmetries are stable and predictive (Ito & Cacioppo 2005).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Positivity and negativity are separable at early stages of processing | moderate | behavioural independence; partly distinct neural substrates and transmitters |
| C2 | They can be coactivated, and coactivation is ambivalence | moderate | field and laboratory reports; oscillation study; Grabenhorst et al. 2007 |
| C3 | Coactivation is rare and highly arousing, and its circumstances tend to be avoided | weak | asserted as postulate; field data show bipolar structure on ordinary days |
| C4 | Negativity has higher gain, positivity a higher intercept | strong | many converging reviews and measures |
| C5 | Depression and anxiety differ in which channel is dysregulated | weak | consistent with attention and imaging findings; not directly tested |
Concepts
- modes of evaluative activation — reciprocal, uncoupled (independent) and nonreciprocal (coactivation).
- objective ambivalence — coactivation of positivity and negativity. Subjective ambivalence is its experience, including the blur that fast oscillation produces.
- response surface — the net predisposition toward (+) or away from (−) the target, over the evaluative plane. Its isocontours are “many-to-one mappings between the affective response surface and the underlying evaluative space”.
- negativity bias / positivity offset — higher gain for negativity, higher intercept for positivity.
- heterarchical organization — evaluation is represented at every level of the neuraxis, and higher levels can override lower.
Connections
Summarizes Cacioppo & Berntson 1994 (LIT-1453) and Cacioppo, Gardner & Berntson 1997, 1999 (LIT-1456), all unread. Argues against Russell and Carroll’s bipolarity. That is also the shape of Barrett’s core affect (LIT-1353). Cites Gray and Lang’s two-system theories, which it calls reciprocal, Carver & Harmon-Jones on anger, and Kahneman and Tversky on loss aversion as a negativity bias.
Bearing on the record
The record’s proposal (CLAIM-440, TERM-076) makes hedonic valence one ordering serving as the negotiation’s potential of a system’s own states. It reads ambivalence as “either no ordering at all … or split potential”, and neutrality as “potential with no movement”. The ESM bears on this in five ways.
- The distinction is anticipated. That ambivalence is not neutrality, and that a single bipolar measure collapses them, is the ESM’s founding point (Figure 2: “neutral to ambivalent … yields the same middling expression”). The proposal should cite the ESM for it.
- The ESM puts the difference in a different place. For the ESM both states are positions in a two-channel activation space, low-low and high-high. For the proposal they are properties of the potential structure among perspectives. The ESM’s ambivalence still produces an ordering: the net predisposition is a single number, “middling”. So “no ordering at all” is not the ESM’s ambivalence. Split potential is closer. Two channels each strongly active, pulling toward approach and toward avoidance, can be read as two rankings each prevailing part of the system. But the ESM combines them into one output and does not describe them as perspectives that win conflicts. The proposal’s “split potential” is therefore a reinterpretation of coactivation, not a restatement of it. It owes an argument that the two channels are perspectives in CLAIM-347’s sense.
- The same-output result is a warning for measurement. If neutral and ambivalent states give the same net predisposition, a record of which option is chosen cannot tell them apart. QUESTION-268’s “conflict record” must record which perspective’s ranking each decision followed, not only the decision. The ESM supplies marks that a choice record lacks. Ambivalence is “highly arousing”. At high coactivation “energy expenditure is taxing over long periods of time”. Its eliciting circumstances “tend to be avoided”. Each is a candidate independent marker for the proposal’s defeater (“ambivalent and neutral states that a record … cannot tell apart”).
- The asymmetries cut against a symmetric sign clause. The proposal gives sign as the direction of movement along the ordering, with nothing to choose between up and down. The ESM’s negativity bias (“The gain for the negative activation function is higher”) and positivity offset are asymmetries the proposal does not predict. If intensity is the potential’s strength, the negativity bias says downward movement prevails more strongly than equal upward movement. That is compatible with the proposal, but must be added to it.
- Heterarchy bears on CLAIM-435’s defeater. The ESM says “different modes of activation” can “be dominant or prevail depending on the level of representation”. Reflexes are reciprocally constrained and rigid, and “higher levels of the neuraxis have the ability to override these dominant responses”. CLAIM-435 reads Veit’s currency (THEORY-101) and ALARM’s override (THEORY-090) as two regimes of one ordering. It is defeated if “an override and a currency” are “shown to coexist in one animal as separate mechanisms”. The ESM’s heterarchy, with rigid low-level dispositions and flexible higher-level weighing that can override them, looks like such coexistence. It is not a demonstration, since the ESM does not show that the low level is lexicographic or the high level compensatory. But it is the sort of architecture the defeater names, and the record should look at it before promoting CLAIM-435.
Limitations
A review, partly a defence of its authors’ model. Many claims are postulates rather than results: the rarity and cost of coactivation, calibration, and the clinical predictions. The figures were not seen, only their captions. The ESM is explicitly not a theory of what elicits affect, so it is silent on the proposal’s “kind” of ordering, by bearing on maintenance.
Open questions
Whether coactivation is two potential rankings or one ranking with high input on both sides. Whether a positivity offset means the potential ordering has a positive default slope at rest, which would make the proposal’s neutral point, no movement, mildly positive.