Study Reveals the Rhythmic Roots of Human Laughter in Our Great Ape Cousins
New research demonstrates that the rhythmic structure of human laughter, specifically its isochronous or "metronome-like" beat, is shared across all great apes, suggesting a deep evolutionary origin for this fundamental vocalization.
Methodology: Capturing the Sound of Joy
Audio recordings were collected from 13 non-human great apes—four orangutans, two gorillas, three bonobos, and four chimpanzees—across seven institutions between 2004 and 2006. These were compared with recordings of laughter from four human children in their homes. Vocalizations were elicited during playful interactions with familiar humans, including tickling. Recordings were made using high-fidelity Sony WM-D6C (with Sennheiser ME 60 microphone) and Nagra IV-SJ (with Sennheiser MKH 816 microphone) equipment.
Statistical Analyses of Laughter Patterns
Audio files were converted and processed at 22,050 Hz with a 60 Hz high-pass filter. Calls were annotated for start time and duration. The analysis focused on bouts (sequences of calls with intervals <8 ms or same mode) containing at least three calls, yielding a total of 140 bouts (42 bonobo, 35 chimpanzee, 34 gorilla, 13 human, 16 orangutan). Interval durations (t_k) were calculated between consecutive calls within bouts, producing 458 values.
Laughter Tempo: What Drives the Beat?
Three linear mixed-effects models (LMMs) tested how tempo (log-transformed) varies with key factors:
- LMM1 assessed phylogenetic distance with age as a control variable.
- LMM2 included the interaction between phylogenetic distance and context (play vs. tickling).
- LMM3 tested tempo variation by species and context with pairwise comparisons.
Tempo Variability: Measuring Consistency
The coefficient of variation (CV) of laughter tempo was calculated per recording (N=32). LMM4 tested phylogenetic distance and age as fixed factors, with the individual as a random intercept.
Testing Isochrony: The Metronome of Laughter
Ratios r_k (the duration of t_k relative to the combined duration of consecutive intervals) were computed, resulting in 316 values. Three generalized linear mixed models (GLMMs) tested whether r_k values fell within the on-isochrony range (0.440 < r_k < 0.555) versus off-isochrony ranges, using a Poisson distribution with offset:
- GLMM1 analyzed all data.
- GLMM2 analyzed the play context only.
- GLMM3 analyzed the tickling context only.
Statistics and Reproducibility
Analyses were performed in R. Sample sizes correspond to bouts, intervals, ratios, or recordings. Replicates were independent bouts or recordings. Random effects accounted for individual and file identity. No data were excluded except recordings not meeting inclusion criteria. No statistical methods were used to predetermine sample size.