Children’s recognition of slapstick humor is linked to their Theory of Mind

The study investigates whether preschoolers (4–5 years) recognize slapstick humor in misfortune scenarios only when the victim shows a bewildered, comic facial expression rather than angry or painful expressions; whether this ability increases with age; whether it is associated with Theory of Mind (ToM); and whether it relates to facial emotion recognition. The work is motivated by the importance of humor in early development, cognitive accounts of humor (e.g., incongruity detection and resolution), developmental trajectories of humor and emotion recognition, and proposals that ToM supports interpreting social-emotional context and humorous intent. Building on adult evidence that facial expressions, especially bewilderment, trigger humorous appraisal in slapstick contexts, the authors test how age, ToM, and emotion recognition jointly relate to children’s humor recognition.

Prior theories describe humor as universal and tied to cognitive development. McGhee’s cognitive-stage theory outlines progression from simple incongruities in toddlerhood to appreciation of slapstick (ages 2–7) and later linguistic ambiguities (7–11). Theory of Mind (ToM) develops from infancy through childhood, with major advances around age 4 (false-belief understanding), and has been linked to humor comprehension in adults, potentially via overlapping neural networks. Adult studies show that in slapstick, a victim’s bewildered expression acts as a perceptual trigger for comic appraisal: ERP studies report enhanced N170/N220 to comic versus affective/no-face conditions and N400 to no-face, suggesting facial expressions resolve contextual ambiguity. STG stimulation via tDCS speeds comic categorization, implicating face/biological motion processing in humor recognition. The two-stage humor model posits initial incongruity detection followed by resolution and amusement; a bewildered face signals harmlessness, facilitating resolution. Developmentally, emotion recognition refines nonlinearly up to ~6 years; language and social experiences bolster ToM. Some gender and sibling-context differences in humor usage have been noted, though their impact on recognition is uncertain. The link between ToM and humor recognition in children, particularly for slapstick, remained understudied prior to this work.

Design: Pre-registered online original research using Gorilla Experiment Builder. Three main measures: (1) humor recognition task (slapstick misfortune images), (2) facial emotion recognition task (happy, sad, fearful, angry), and (3) parent-reported ToM (CSUS). Participants: N = 61 kindergarteners (32 female), ages 49–58 months (M = 53, SD = 2.6), recruited in Zurich, Switzerland. Exclusion: developmental disorders, impaired sight, preterm birth, bi-/multilingual. Parent respondents: 92% mothers, 8% fathers; high parental education typical. Ethics approval obtained; informed consent provided. Small gift provided post-participation. Materials and stimuli: Emotion recognition used 24 trials with 4 faces per trial (2 male, 2 female) from the Radboud Faces Database (RaFD), each displaying one of four emotions. Each actor served as target at least once. Humor recognition used paired images of misfortune scenarios derived from Manfredi et al. (2014, 2017), balanced on actor characteristics and body parts. Each pair contrasted a Comic (bewildered/funny facial expression) versus an Affective (angry/painful expression) version of the same scenario. Eye-tracking via Gorilla was attempted but not analyzed due to poor calibration/data quality. Procedure: Parents completed CSUS, consent, and demographics before testing. Children completed two tasks in counterbalanced order online with a present parent instructed not to influence responses. A brief cartoon separated tasks. Emotion recognition: child heard a prompt (e.g., “Which is the happy/sad/fearful/angry face? Please point out.”) while viewing four faces; the parent clicked the chosen face. Inter-trial fixation cross 250 ms with 100 ms pauses. Scoring: 1 = correct, 0 = incorrect. Humor recognition: one practice trial, then 32 test trials presented in blocks (four blocks of 6 trials and one of 8), randomized. Each trial showed a pair (comic vs affective) for 6,000 ms with fixation; an audio prompt asked “Which picture is funny? Please, point it out.” The parent pressed left/right to record the child’s choice. Scoring: 1 = comic chosen, 0 = affective. Data analysis: Preliminary multiple linear regression tested gender (female reference) and number of siblings predicting humor recognition proportion; neither significant, so excluded from main model. Main analysis: generalized linear mixed-effects model (binomial logit) predicting trial-level humor recognition (0/1) from standardized (z) predictors: age, ToM (CSUS), and emotion recognition accuracy for each emotion (happy, sad, fearful, angry). Random intercepts for Participant and Trial; no random slopes due to convergence considerations and lack of theoretical predictions for varying slopes. Model: glmer(humor_recognition ~ Age_scaled + ToM_scaled + happy_scaled + sad_scaled + fearful_scaled + angry_scaled + (1|Participant) + (1|trial), family = binomial). Model diagnostics indicated acceptable residuals and random effects distributions. OSF preregistration, anonymized data, and analysis scripts were provided by the authors.

- Descriptives (proportions correct, 0–1 scale): Humor recognition M = 0.67 (SD = 0.14; range 0.28–0.91). Overall emotion recognition M = 0.78 (SD = 0.14; range 0.38–1.00). By emotion: Happy M = 0.95 (SD = 0.13), Sad M = 0.79 (SD = 0.19), Angry M = 0.69 (SD = 0.24), Fearful M = 0.68 (SD = 0.24). ToM (CSUS) M = 3.2 (SD = 0.29; range 2.5–3.8). - Preliminary predictors: Gender (Beta = 0.02, SE = 0.04, t = 0.55, p = 0.588) and number of siblings (Beta = −0.01, SE = 0.05, t = −0.13, p = 0.895) were not significant predictors of humor recognition proportion. - Main GLMM results (odds ratios, 95% CI, p): • ToM (z): OR = 1.21, 95% CI [1.02, 1.42], p = 0.025 (significant). One SD higher ToM associated with 21% higher odds of identifying the comic picture. • Age (z): OR = 1.10, 95% CI [0.94, 1.29], p = 0.239 (ns). • Emotion recognition (z): Happy OR = 0.67, p = 0.542; Sad OR = 1.20, p = 0.716; Fearful OR = 0.73, p = 0.415; Angry OR = 1.89, p = 0.088 (trend only). None reached significance. • Random effects: ICC = 0.16; N participants = 61; N trials = 32; observations = 1,952. Marginal R^2 = 0.018; Conditional R^2 = 0.124. - Above-chance recognition: From the earliest tested age, the predicted probability of choosing the comic image exceeded chance (50%). This above-chance performance held across the observed ranges of ToM and emotion recognition scores.

Findings indicate that preschoolers can distinguish between misfortune scenarios that are humorous versus non-humorous based on the victim’s facial expression, selecting the bewildered/comic expression over angry/painful alternatives at above-chance levels. Contrary to expectations of a strong age effect within 4–5 years, age did not significantly predict humor recognition, suggesting either early emergence of this ability or developmental changes occurring outside this narrow window. ToM robustly predicted humor recognition, aligning with adult evidence linking mentalizing with humor processing, and supporting the idea that children leverage understanding of others’ mental states to interpret incongruity as playful and harmless, thereby resolving it as funny. Emotion recognition accuracy for basic expressions did not significantly explain humor recognition, despite the facial expression being the diagnostic cue in the task; this may reflect ceiling or limited variability in basic emotion recognition at this age, the importance of contextualized emotion processing rather than decontextualized faces, or insufficient sensitivity of the selected emotion set. The pattern dovetails with adult ERP findings that comic facial expressions act as triggers for humor processing and with recent child EEG work showing enhanced N170 for funny versus affective images, suggesting early, automatic face-expression processing contributes to humor recognition, while ToM helps interpret the social-emotional meaning.

The study shows that 4–5-year-old children recognize slapstick humor in misfortune contexts preferentially when the victim displays a bewildered (comic) expression rather than angry or painful expressions. Within this age range, recognition does not increase significantly with age, but is associated with higher parent-reported ToM. Basic facial emotion recognition performance was not a significant predictor, suggesting that contextualized interpretation and mental-state understanding are more relevant for recognizing slapstick humor at this stage. Future research should broaden the age range, include additional and context-embedded emotions (e.g., surprise, disgust), pair behavioral with neural measures, and incorporate direct behavioral ToM assessments and control conditions to disentangle facial-expression and context effects. The findings highlight potential applications of slapstick-based activities and ToM-focused interventions to support socio-cognitive development and wellbeing in early childhood.

- Response format: Forced-choice between two images may reflect relative rather than absolute funniness; no direct measures of amusement (e.g., smiles/laughter) were collected. - Emotion set: Only four basic emotions (happy, sad, fear, anger) were tested; other emotions like surprise or disgust might be more informative for humor contexts. - ToM measurement: Assessed solely via parent-report CSUS; no behavioral ToM tasks, leaving potential reporter bias. - Eye-tracking: Planned gaze data were unusable due to calibration/data quality issues, limiting implicit process insights. - Controls and design: No control condition presenting bewildered expressions in non-comic contexts; future designs should better isolate facial-expression versus contextual contributions.