The Deeper the Love, the Deeper the Hate

The study examines the relationship between romantic love and hate, a link that has been understudied relative to extensive work on love (e.g., Sternberg’s triangular theory; Fisher’s models). Love has been conceptualized variously as action, attitude, experience, and a prototypical emotion, highlighting its multifaceted nature. Romantic hate is often tied to relational betrayal and overlaps conceptually with romantic jealousy. Prior methodologies have struggled to capture the complexity of the love–hate dynamic, and mechanisms underlying their interrelation remain unclear. This study tests whether similarity between partners increases love in non-betrayal contexts and whether higher prior love predicts greater hate following betrayal, using experimentally controlled vignettes.

Prior work emphasizes romantic love’s structure and stages (Sternberg, 1986; Fisher, 1989; Fisher et al., 2006) and views love as multifaceted (Ekman, 1972; Izard, 1977; Tomkins, 1984). Romantic hate is linked to betrayal and jealousy (Yoshimura, 2004). Similarity, rather than complementarity, typically predicts attraction, mate preference, and relationship quality (Berscheid & Reis, 1998; Luo & Klohnen, 2005; Hudson et al., 2014). Equity theory suggests perceived imbalance between inputs and outcomes can shift affect within relationships (Walster et al., 1973; Hatfield et al., 1979). Romantic jealousy correlates positively with romantic love and relationship threat sensitivity (Mathes & Severa, 1981; Orosz et al., 2015). These strands motivate examining how similarity shapes love and how betrayal transforms love into hate.

Design: Two-part experimental vignette paradigm inducing romantic love (Part 1) and hate following betrayal (Part 2). Participants imagined themselves as the protagonist interacting with three targets (A, B, C) differing in excellence and similarity. Participants: N=59 (30 men, 29 women; M_age=20.2, SD=1.5), recruited from colleges; screened for no psychiatric/neurological history. Relationship statuses varied. Ethics approval obtained; written informed consent; small incentives provided. Materials/Vignettes: Adapted from Takahashi et al. (2009) and modified for romantic context per prior love definitions. Three targets: A (equal excellence, high similarity), B (equal excellence, low similarity), C (low excellence, low similarity). Participants studied materials, recalled self-traits ("I am...") and target information, and imagined being in a relationship with each target. Negative events (betrayals; e.g., having an affair) were later presented for each target in Part 2. Measures:

• Trial-level ratings: In Part 1, love toward each target's traits (1=none to 6=extreme love). In Part 2, hate toward each target after each negative event (1=none to 6=extreme hate).
• Passionate Love Scale (PLS; 15 items; 1–9 per item) administered after Part 1 (baseline love per target) and after Part 2 (post-betrayal love). Cronbach’s alpha=0.94. Procedure: E-Prime 2.0 presented randomized items. Part 1: fixation (1000 ms), stimuli (target label, comparison domain, specific traits) up to 2000 ms or until response; then PLS per target. Part 2: similar presentation of traits followed by negative events (betrayals) with 1000 ms ISIs; hate ratings collected, then PLS again. Analysis: Within-subjects averaging of love ratings, hate ratings, and PLS items. One-way repeated-measures ANOVAs tested differences across targets (A, B, C) for love ratings, hate ratings, and PLS (pre- and post-betrayal). A 3 (target) × 2 (time: before vs after) repeated-measures ANOVA on PLS assessed betrayal effects. A 3 (target) × 2 (affect: love vs hate) ANOVA examined love–hate relationships. Simple effects tested when interactions significant. Pearson and partial correlations (controlling age, gender) assessed love–hate associations per target.

• Love (Part 1): Significant differences across targets in love ratings, F(2,116)=985.710, p<0.001, η²=0.944. Means (±SD): A=5.53±0.48 > B=4.52±0.54, F(1,58)=177.796, p<0.001, η²=0.754; B > C=1.66±0.45, F(1,58)=977.526, p<0.001, η²=0.944. PLS also differed by target, F(2,116)=450.352, p<0.001, η²=0.886; A=109.73±11.80 > B=93.46±14.59, F(1,58)=60.263, p<0.001, η²=0.510; B > C=38.39±20.40, F(1,58)=519.537, p<0.001, η²=0.900.
• Hate after betrayal (Part 2): Hate ratings differed across targets, F(2,116)=229.640, p<0.001, η²=0.798; A=5.25±0.57 > B=4.84±0.55, F(1,58)=34.768, p<0.001, η²=0.375; B > C=3.02±0.98, F(1,58)=216.921, p<0.001, η²=0.789. Post-betrayal PLS differed, F(2,116)=316.544, p<0.001, η²=0.845; A=88.95±22.00 > B=71.97±21.83, F(1,58)=63.119, p<0.001, η²=0.521; B > C=27.81±14.39, F(1,58)=333.357, p<0.001, η²=0.852.
• Betrayal effect on love (PLS before vs after): Target × time interaction, F(2,116)=10.432, p<0.001, η²=0.152. Love decreased after betrayal for all targets: A: 109.73→88.95, F(1,58)=74.822, p<0.001, η²=0.560; B: 93.46→71.97, F(1,58)=68.179, p<0.001, η²=0.540; C: 38.39→27.81, F(1,58)=27.842, p<0.001, η²=0.324.
• Love vs hate relationship: Target × affect interaction, F(2,116)=95.357, p<0.001, η²=0.622. For A, love > hate even after betrayal (Love 5.53 vs Hate 5.25), F(1,58)=17.889, p<0.001, η²=0.236. For B, love < hate (4.52 vs 4.84), F(1,58)=14.652, p<0.001, η²=0.202. For C, love < hate (1.66 vs 3.02), F(1,58)=102.933, p<0.001, η²=0.640.
• Correlations between love and hate: A: r=0.55, p<0.001 (partial r=0.48, p<0.001); B: r=0.29, p<0.05 (partial r=0.27, p<0.05); C: r=0.12, p>0.05 (partial r=0.12, p>0.05). Interpretation: Greater similarity increased love; higher prior love predicted greater hate following betrayal, with love remaining dominant for the highly similar, highly valued target even after betrayal.

Findings support a nuanced link between romantic love and hate. Similarity increased initial love toward potential partners, aligning with attraction and assortative mating literature. Following betrayal, hate was strongest toward the previously most-loved target, indicating that love and hate are interrelated; betrayal can transform strong romantic love into intense hate. Yet, for the highly similar, equally excellent target (A), love remained higher than hate even after betrayal, whereas for less-similar or lower-excellence targets (B, C), hate exceeded love. This pattern highlights that preexisting attachment strength and perceived similarity modulate emotional responses to betrayal. Potential mechanisms include romantic jealousy (emotional and cognitive components) and perceived equity imbalance: greater investment in a loved partner magnifies negative affect when outcomes violate expectations. The results elucidate differential emotional reactions to different targets and the complexity of love–hate dynamics in romantic contexts.

The study supports the adage “the deeper the love, the deeper the hate,” demonstrating that similarity is a crucial factor enhancing love and that stronger prior love predicts greater hate after betrayal. Emotional reactions vary by partner characteristics; when initial love is strongest (high similarity, equal excellence), love can remain dominant even after betrayal, whereas for less-loved targets, hate predominates. The work underscores an interrelation between romantic love and hate and the moderating role of similarity, suggesting avenues for future research on status/relationship context moderators and mechanisms such as jealousy and perceived equity.

• Independence of induced relationships: Despite instructions that the protagonist was in three distinct relationships at different life periods, independence of feelings across targets could not be guaranteed.
• Uncontrolled relationship status: Participants’ current relationship status was not controlled, which may affect love/hate responses; future work should test status as a predictor.
• Similarity manipulation via vignettes: Reliance on imagined scenarios may have limited the effectiveness of the similarity manipulation, depending on participants’ ability to adopt the protagonist perspective.