An empirical investigation of emotion and the criminal law: towards a "criminalization bias"?

The paper examines whether emotion—specifically disgust—relates to decisions about what behavior should fall under criminal law. Traditional criminalization theory emphasizes reflective, deliberative reasoning (e.g., harm principle) to justify criminalization. However, moral psychology and cognitive neuroscience indicate that affective processes heavily influence moral judgments, often unconsciously and with subsequent rationalization. The authors focus on virtual child pornography as a theoretically low-harm but high-disgust case and contrast it with a high-risk financial behavior scenario deemed high harm but low disgust. They hypothesize that (1) vignettes high in disgust (virtual and actual child pornography) would be more strongly criminalized than a high-harm, low-disgust financial vignette; (2) higher trait disgust sensitivity would be associated with stronger criminalization tendencies, especially for virtual child pornography; and (3) they explore whether legal expertise mitigates these emotion-related effects.

The authors review work in moral psychology demonstrating that moral judgment involves both affective and reflective processes (Haidt, Greene, Cushman). Disgust is highlighted as a potent negative emotion linked to pathogen avoidance and moral condemnation, particularly within the purity domain. Experimental findings on incidental disgust manipulations are mixed overall, but robust associations exist between trait disgust sensitivity and harsher condemnation across domains, especially purity-related violations. Disgust sensitivity correlates with punitive judgments, deontological preferences, and negative attitudes toward groups or practices perceived as impure. The review situates virtual child pornography within debates on criminalization: harm-based arguments are contested with mixed empirical support for downstream harms, while legal moralism offers abstract justifications often deemed insufficient alone. The authors propose that affective reactions (disgust) may help explain support for criminalizing virtually simulated offenses despite uncertain harm, motivating the notion of a potential "criminalization bias" wherein affect may overweight perceived harm for modern, emotionally provocative but low-harm behaviors, and underweight harm for abstract, impersonal, low-disgust behaviors (e.g., complex financial risks).

Design: A within-participant 2 (harm: high vs low) × 2 (disgust: high vs low) vignette design, with trait disgust sensitivity as a continuous between-participant moderator. Two dependent measures per vignette: a 7-point agreement rating (1=strongly disagree to 7=strongly agree) and a dichotomous decision (agree/disagree) regarding whether the behavior should be criminalized. Participants: 1725 began the survey; exclusions included participants <16 years (N=38) and participants who did not rate any vignette (N=285), yielding N=1402 (517 males, 876 females, 9 not reported; mean age 33.57 years, SD 13.02; 4 missing age). Legal experts (N=103; legislative lawyers, legal policymakers, lawyers, judges) were compared to laypersons (N=1297). Groups were matched on age and gender distribution. Materials/Vignettes: Four vignettes characterized by objective (literature-based) harm and disgust levels: (1) Wearing a sweater with clashing bright colors (low harm, low disgust); (2) Contingent convertible bonds (CoCos), explained as a high-risk financial instrument (high harm, low disgust); (3) Virtual child pornography (low harm, high disgust); (4) Actual child pornography (high harm, high disgust). Each vignette asked whether the behavior should be criminalized (binary yes/no and 7-point Likert). Trait Measure: Dutch Disgust Scale-Revised (DS-R-NL; 25 items; 0–4 scale). Cronbach’s alpha α=0.78 (N=1175). Disgust sensitivity scores were standardized (z-scores) for analysis. Procedure: Online Qualtrics survey (Feb 2017–Feb 2018). Recruitment via university networks, public outreach (magazine), and social media; experts via law firms and the Dutch Academy for Legislation. Vignettes were presented in randomized order. Demographics and legal expertise collected. After vignettes and DS-R-NL (order counterbalanced), participants were debriefed. Ethics and Preregistration: Approved by Leiden University Psychology Research Ethics Committee (CEP17-0803_260). No formal preregistration at the time; the time-stamped ethics protocol with hypotheses, raw data, and analysis code is available on OSF (https://osf.io/a49qe). Analytic Strategy: Linear mixed-effects models (for continuous ratings) and mixed-effects logistic regression (for dichotomous decisions) using R (lme4, lmerTest), with random intercepts (and random slopes when possible). Confidence intervals via Monte Carlo bootstrap. Models tested main effects of vignette disgust and harm and moderation by trait disgust sensitivity, with exploratory moderation by legal expertise. Simulation-based sensitivity power analyses (simr) assessed power for key interaction terms.

Main Effects (Continuous Ratings): Both vignette disgust and harm robustly increased criminalization ratings. Disgust: b=3.63, SE=0.05, t=67.17, p<0.001, 95% CI [3.52, 3.72]; Harm: b=2.51, SE=0.05, t=46.25, p<0.001, 95% CI [2.42, 2.63]. There was a disgust × harm interaction (b=−0.56, SE=0.08, t=−7.26, p<0.001), indicating harm raised ratings more for low-disgust vignettes (M low harm=1.22 vs high harm=3.73) than for high-disgust vignettes (M low harm=4.84 vs high harm=6.80). Main Effects (Dichotomous Decisions): Higher disgust and higher harm increased odds of deciding to criminalize: Disgust b=4.92, SE=0.26, z=17.71, p<0.001; Harm b=3.80, SE=0.25, z=15.09, p<0.001. No significant disgust × harm interaction. Decision Frequencies: Low harm/low disgust (sweater): ~1% yes; Low harm/high disgust (virtual child pornography): ~63% yes; High harm/low disgust (financial instrument): ~36% yes; High harm/high disgust (actual child pornography): ~99% yes. Moderation by Disgust Sensitivity (Continuous): Significant interactions showed that higher trait disgust sensitivity was associated with higher criminalization ratings especially for low-harm/high-disgust vignettes (virtual child pornography) and, to a lesser extent, for high-harm/low-disgust vignettes (financial behavior). Disgust × DS: b=0.47, SE=0.06, t=8.53, p<0.001; Harm × DS: b=0.18, SE=0.06, t=3.28, p=0.002; Disgust × Harm × DS: b=−0.66, SE=0.08, t=−8.74, p<0.001. Simple slopes: negligible DS effect for low/low and high/high cells; significant for high harm/low disgust (b=0.19, SE=0.04) and stronger for low harm/high disgust (b=0.48, SE=0.04). Moderation by Disgust Sensitivity (Dichotomous): Overall main effect of DS (b=0.59, SE=0.26, z=2.29, p=0.02). Interaction patterns were clearest when focusing on the two mixed cells: vignette main effect (b=1.18, SE=0.09, z=12.71, p<0.001) and DS × vignette interaction (b=0.49, SE=0.09, z=5.30, p<0.001), indicating a stronger association of DS with criminalizing the low harm/high disgust vignette relative to the high harm/low disgust vignette. Moderation by Legal Expertise: For continuous ratings, a three-way interaction (Disgust × Harm × Expertise) was significant (b=0.94, SE=0.29, t=3.19, p=0.001). Laypeople rated the high harm/low disgust financial vignette more criminalizable than experts (lay M≈4.87 vs expert M≈3.23). The four-way interaction including DS was not significant, suggesting experts and laypeople were similarly influenced by trait disgust sensitivity. No additional significant effects emerged for dichotomous decisions. Power Sensitivity Analyses: The continuous three-way interaction (estimating b=0.15) had simulated power ≈100% (95% CI ~[99.6, 100]); the dichotomous focused DS × vignette interaction (estimating b=0.15) had simulated power ≈91% (95% CI ~[88.8, 93.9]).

Findings support the central hypothesis that affective processes, indexed by trait disgust sensitivity and triggered by highly disgusting content, are associated with decisions to criminalize behavior. Virtual child pornography (low harm/high disgust) was more strongly criminalized than a high-harm/low-disgust financial scenario, contrary to expectations from a strict harm-principle lens. The association between disgust sensitivity and criminalization was especially pronounced for virtual child pornography, aligning with theories that purity-related, concrete and personal violations elicit stronger affective responses. Legal expertise did not attenuate these effects; experts and laypersons showed similar patterns regarding the influence of disgust sensitivity, and experts were, if anything, less inclined than laypersons to criminalize the abstract financial vignette. The authors interpret this as preliminary evidence for an affective contribution to criminalization judgments and propose a theoretical framework of a potential "criminalization bias," wherein emotions may overweight perceived harm in certain modern, emotionally evocative but low-harm contexts, and underweight harm in abstract, low-disgust contexts. The results raise broader implications about the role of emotion in legal theorizing and policy, suggesting criminalization decisions may not be purely deliberative and may involve rationalization of affect-driven intuitions.

This study offers preliminary empirical evidence that emotion—particularly disgust—relates to criminalization decisions among both laypeople and legal professionals. Participants were more inclined to criminalize virtual child pornography (low harm/high disgust) than a high-risk financial behavior (high harm/low disgust), and higher trait disgust sensitivity predicted stronger criminalization tendencies, especially for the virtual child pornography scenario. Theoretically, the authors propose a dual-process perspective on criminalization and a potential "criminalization bias" for modern offenses where affective responses may misalign with actual harm. They encourage further interdisciplinary research—combining rigorous psychological methods and legal theory—to validate, generalize, and refine these insights, and to inform better-grounded criminalization policies.

- Vignette design: To maximize ecological validity, vignettes were drawn from existing legal-theoretical debates, leading to disparities in orthogonality, length, and complexity—especially the high harm/low disgust financial scenario—which may have introduced confounds. Future work should standardize vignette complexity and include multiple vignettes per harm × disgust cell. - Single vignette per cell: Using one vignette for each condition limits generalizability; the relative effect size pattern (strongest DS association in low harm/high disgust vs high harm/low disgust) should be interpreted cautiously. - Emotion scope: The study focused solely on disgust as a proxy for emotion; other affective states (e.g., anxiety, anger) may also contribute to criminalization judgments and should be examined. - Expert sample: The expert group was relatively small (N=103) and heterogeneous (various legal roles). Power to detect expertise moderation may have been limited. Incentives and tighter expertise definitions may improve measurement precision in future research. - Measurement limits: Ceiling and floor effects in the high/high and low/low cells constrained detection of DS interactions for dichotomous outcomes; future designs should aim to increase variability and validate scenarios more extensively.