Human Emotion Ontology (HEO) · Schema version 1.0.0
This registry documents the major active disagreements in emotion science and how the ontology represents each one. The ontology does not resolve these disputes; it encodes them (via multiple parents, low confidence, and notes) so that a knowledge-graph consumer can reason about the disagreement.
Dispute. Basic-emotion theorists hold that a small set of emotions are biologically hardwired, expressed and recognized similarly across cultures, and produced by dedicated neural circuits (Ekman 1999; Panksepp 1998; Keltner et al. 2019). Constructionists hold that emotion categories are learned concepts applied to a continuous core-affect substrate; "anger" and "fear" are no more natural kinds than "furniture" (Barrett 2006, 2017; Russell 2003; Lindquist et al. 2012).
Evidence on both sides. Meta-analytic reviews find modest, task-dependent autonomic and neural specificity (Lindquist et al. 2012 vs. Kreibig 2010); posed expression recognition is robust in literate cultures (Ekman & Friesen 1971; Cordaro et al. 2018, 2019) but weaker in cultures with little Western media (Gendron et al. 2014; Jack et al. 2012).
Ontology treatment. The basic-emotion candidates are grouped under grp.basic_emotion and the emotion category, with theoretical_frameworks listing both basic_emotion and constructionist where relevant; affect and abstract.core_affect provide the constructionist substrate. The dispute is recorded in 04 and in the node notes for contested members.
Dispute. Six (Ekman's core), seven (Panksepp's systems — a different construct: neurobehavioral systems, not facial-expression categories), eight (Plutchik), ten (Izard), up to seventeen (Ekman's expanded 1999 list).
Specific contested memberships:
blend of anger + disgust (Plutchik's dyad) vs. a culturally specific concept. Ontology: node exists with medium confidence; both framings recorded (blend_of edge with high confidence from Plutchik; basic_emotion grouper).
sentiment. Ontology: medium confidence; also parented to grp.social_emotion.
Ontology: medium confidence with an explicit note.
brief; some argue it is not an emotion at all (Ortony & Turner 1990). Ontology: kept, valence 0, note attached.
Ontology: kept as basic with a note.
sentiment. Ontology: node with cat.sentiment parent and blend_of edges.
Dispute. Russell's circumplex says valence + arousal suffice; PAD adds dominance; discrete theorists say categories are irreducible; Cowen & Keltner (2017) find 27 categories bridged by continuous gradients — evidence that both structures coexist.
Ontology treatment. Both representations are present by design: every node has dimensional estimates, and the graph's similarity/opposition edges reproduce circumplex structure (see the Valence × Arousal map view in the visualization), while discrete nodes preserve categorical identity.
Dispute. The classic bipolar model treats pleasant–unpleasant as a single continuum; the evaluative-space model (Cacioppo & Berntson 1994) and mixed-emotion research (Larsen et al. 2001; Berrios et al. 2015) show separable positivity and negativity that co-occur in bittersweet experiences. Critics argue apparent mixed states are rapid alternation rather than simultaneity (Larsen & McGraw 2014 debate).
Ontology treatment. co_occurs_with edges, grp.mixed_emotion, and meta.mixed_feelings / meta.ambivalence nodes encode the simultaneity position; the dispute is noted on meta.mixed_feelings.
Dispute. Plutchik's dyads (joy+trust=love, fear+surprise=awe) treat complex emotions as combinatorial products. Appraisal theorists treat the same states as distinct appraisal patterns (e.g., awe = vastness + need for accommodation; Keltner & Haidt 2003). Constructionists treat them as separate concepts, not mixtures. Several of Plutchik's dyads are idiosyncratic (guilt = joy+fear; shame = fear+disgust) and are not endorsed by other traditions.
Ontology treatment. blend_of edges exist for all dyads, with confidence matching support: high for love/submission/disapproval/remorse/contempt/ aggressiveness/optimism; medium for envy, hope, delight, pessimism; low for guilt, shame, pride, outrage. Node definitions follow the modern (non-dyad) literature.
Dispute. Tangney's program treats them as distinct with distinct action tendencies (guilt → repair, shame → hide). Lewis (1993) treats shame and guilt as differing only in focus (self vs. behavior). Some cross-cultural work finds the shame/guilt boundary does not map onto non-Western emotion lexicons (Wierzbicka 1999). Whether embarrassment is a third distinct emotion or a mild shame variant is also debated (Keltner 1995 vs. Miller 1996).
Ontology treatment. Three distinct nodes, grp.self_conscious_emotion, similar_to edges among them, and notes carrying the dispute.
Dispute. Basic-emotion accounts often treat anxiety as fear's cognitive variant; dimensional accounts place them on a continuum; clinical work treats anxiety as a distinct syndrome with different pharmacology and behavior (Barlow 2002; Öhman 2008; Cowen & Keltner 2017's anxiety→fear→horror gradient).
Ontology treatment. Distinct nodes with a similar_to (strong) edge, shared physiology, and a note documenting both positions.
Dispute. Lexically confusable, empirically distinct: envy is a two-person emotion (wanting what another has); jealousy is three-person (fear of losing a relationship to a rival) (Parrott & Smith 1993). Some languages lack a distinction; some scholars treat romantic jealousy as a blend.
Ontology treatment. Distinct nodes, similar_to (moderate, noting the conflation), blend_of edges for romantic jealousy.
Dispute. Classic view: anger → approach/aggression (Frijda 1986; Harmon-Jones 2003). Alternative: anger sometimes motivates avoidance (retreat to avoid consequences), and trait anger relates to both (Carver & Harmon-Jones 2009). Ontology: approach/avoidance dimension on anger marked approach with a note; both behavior.attack and (via frustration→anger chain) withdrawal are present.
Dispute. Classic universality (Ekman & Friesen 1971; Cordaro et al. 2018, 2019) vs. culture-specific decoding (Gendron et al. 2014 — Trobriand Islanders; Jack et al. 2012 — Western vs. Eastern reading). The strongest claim that holds is a small universal expressive core with strong cultural modulation.
Ontology treatment. cultural_scope field; basic emotions marked broad; the dispute recorded in the research report.
Dispute. Levenson's program found some autonomic differentiation (1992); meta-analytic review found little specificity (Lindquist et al. 2012); Kreibig's review (2010) found moderate specificity depending on method. Ontology: physiological_association edges carry medium confidence and the dispute is noted in the confidence model.
Dispute. The Kübler-Ross five stages (denial, anger, bargaining, depression, acceptance) are widely popularized but lack empirical support as a fixed sequence (Bonanno 2009; Stroebe et al. 2007). Grief is better modeled as heterogeneous, non-linear.
Ontology treatment. loss → sadness → grief → resignation transition chain is present with an explicit note: "'acceptance' as a stage is disputed"; resignation is used as the terminal node rather than a separate "acceptance" stage.
Dispute. Dimensional views place normal sadness and depressive mood on a continuum with major depression; categorical views treat the disorder as qualitatively distinct. The ontology includes emotion.sadness and mood.depressed (a state) but excludes clinical disorders by design (per the project brief), recording the continuity dispute in notes.
Dispute. Universalist vs. relativistic readings of concepts like amae, saudade, fago, and schadenfreude: are they distinct emotions, or lexicalizations of combinations English already distinguishes? Wierzbicka (1999) and Lutz (1988) argue the former; some cognitive scientists argue the latter.
Ontology treatment. Cultural concepts are real nodes with cultural_scope: culturally_specific, mapped to English-proximate states via culturally_related_to and near_synonym_of edges — so both readings are queryable. cultural.schadenfreude duplicates the concept of emotion.schadenfreude to document the lexicalization itself.
Dispute. Plutchik's dyad (fear+surprise) makes awe mixed/negative; modern research treats it as a distinct positive emotion (Keltner & Haidt 2003; Shiota et al. 2007); Cowen & Keltner (2017) find it adjacent to calmness and aesthetic appreciation; some factor analyses split awe into threat-awe and beauty-awe. Ekman's later list includes it as basic.
Ontology treatment. Node with high confidence, valence +0.7, blend_of edge to fear+surprise (medium confidence), and theoretical_frameworks covering aesthetic/moral/basic.
Dispute. Chapman et al. (2009): moral violations activate the disgust system. Pizarro et al. (2011): the link is metaphorical. Ontology: emotion.moral_disgust node with medium confidence, similar_to both disgust and indignation, and the dispute in its note.
Dispute. Both are repetitive negative thought; worry is future-oriented and verbal (Borkovec), rumination is past-focused and passive (Nolen-Hoeksema). Some argue they are the same process with different objects. Ontology: distinct nodes with similar_to and directional links.
| Mechanism | Example |
|---|---|
| Multiple groupers | emotion.contempt ∈ basic + moral + negative |
| Low confidence + dispute note | blend_of edges for Plutchik's secondary/tertiary dyads |
evidence_status: disputed | nodes where the literature is actively split |
| Dual edges | emotion.awe is blend_of fear+surprise and similar_to wonder |
| Structural duplication | cultural.schadenfreude vs. emotion.schadenfreude |
| Exclusion | clinical disorders (major depression, GAD) are not nodes |