Appraisal overlap
The central construct of Tong & Jia (2017): an account of why emotions co-occur, framed in the vocabulary of appraisal theory. Positive emotions are notoriously entangled — they correlate far more with one another than negative emotions do (pairwise r = .15–.56 in the study) — and the usual explanation is that they simply share positive valence. Appraisal overlap proposes something more structured: two emotions co-occur to the extent they share their whole appraisal profile, not just their pleasantness.
The measure
For each pair of emotions and each participant, the study first computes how each emotion correlates with each of 13 appraisals across that person’s momentary reports. The overlap between two emotions is then the (inverse) Euclidean distance between their two appraisal-correlation vectors — lower distance d means the two emotions “line up” on the appraisals in the same way. The appraisals are partitioned a priori:
- POA — predicted overlapping appraisals: appraisals both emotions were theoretically predicted to relate to in the same direction (e.g. amusement and joy both predicted positive on pleasantness, understandability, moral congruence and negative on problems, effort).
- NOA — nonpredicted overlapping appraisals: everything else — appraisals predicted in opposite directions, or for which no prediction could be made.
The claim is that co-occurrence should track POA overlap specifically, not overlap in general — a stronger, falsifiable version of “similar emotions co-occur.”
What Tong & Jia found
Four postulates, all supported in aggregate (paired t tests across ~63 emotion pairs, Fisher-normalised, FDR-corrected):
- Emotion pairs overlap more on POA than NOA — t(63) = 3.02, p = .004.
- Co-occurrence correlates more strongly with POA overlap than NOA overlap — t(62) = 3.62, p < .001.
- Even with pleasantness removed, POA overlap still beats NOA overlap — t(44) = 6.28, p < .001.
- Even with pleasantness removed, co-occurrence still tracks POA overlap over NOA — t(44) = 2.41, p = .012.
Postulates 3 and 4 are the load-bearing ones. They show positive emotions do not co-occur merely because they are all pleasant — they share specific nonpleasant appraisals, and those predict which emotions travel together. (Repeating the analysis with problems also excluded, since problems can carry valence, left the result standing.)
The valence sub-result, and its place in the wiki
The same analysis that isolates nonpleasantness structure also quantifies just how dominant pleasantness is. Overlap on pleasantness alone is far tighter than overlap on all other appraisals — dPleasantness smaller than dNOA at η² = .94, and smaller than the rest of the POA combined at η² = .92. Pleasantness is ‘the central unifying denominator underlying positive emotions.’
So appraisal overlap delivers the wiki’s recurring dual verdict on valence in one construct: pleasantness is the single strongest organiser of positive emotion and insufficient to explain their co-occurrence structure. This is the co-occurrence-side echo of the differentiation-side point on core-affect and tong-2015-positive-emotion-appraisals — valence is the floor, not the building.
The causal caveat travels with the construct
Appraisal overlap is a correlational construct built on correlational data. That two emotions share an appraisal profile, and co-occur, is consistent with appraisals producing both emotions, with the emotions shaping the appraisals, or with a common categorization producing appraisal-describable content and emotion together (Barrett’s reading). The construct describes a real regularity in how appraisals and emotions co-vary; it does not, on its own, name a mechanism. See are-appraisals-causes-or-descriptions.
Antecedents
The overlap-distance method is borrowed from Kuppens, Van Mechelen, Smits & De Boeck (2004) and Pe & Kuppens (2012), and the nearest prior result is Erbas, Ceulemans, Koval & Kuppens (2015), who found that high emotion differentiation (low co-occurrence) went with low appraisal overlap — but treated emotions and appraisals as global constructs, so could not say which appraisals’ overlap drove which emotions’ co-occurrence. Tong & Jia’s contribution is the hypothesis-driven, appraisal-specific, emotion-pair-specific version. None of these antecedent papers are held in raw/.