Fig. 2: Summary of results of in-person experiments.

a Schematic of trial structure for fusion (left) and preference (right) experiments. Participants heard a stimulus, and judged whether it contained one or two sounds, or rated its pleasantness. Line segments denote individual notes, as were presented in the main experiments with musical intervals. b Results of first fusion control experiment, in which participants heard one or two concurrent talkers. Not all Tsimane’ participants completed the experiment (hence the smaller N compared to other panels). Here and in (c), (f), and (g), graph plots proportion of trials on which participants reported hearing one sound, plotted separately for Boston non-musicians and Tsimane’. Here and in (c)–(f) and (h), plots show the mean ± SEM. Results for individual participants are shown in Supplementary Fig. 1. c Results of second fusion control experiment, in which a separate set of participants heard one or two concurrent sung vowels, resynthesized to be either harmonic or inharmonic. F0 difference between vowels was chosen to avoid fusion in Western listeners when the notes were harmonic. Participants for this experiment were different from those for other experiments (hence different sample size). d, e Results of preference control experiments, in which participants rated the pleasantness of recorded laughs and gasps, and of smooth and rough synthetic tones, respectively. In the latter case, tones consisted of pairs of frequencies presented either dichotically, to avoid beating, or diotically, to produce beating (roughness). f Results of fusion experiment with musical intervals composed of synthetic notes. Fusion judgments were pooled across canonically consonant and dissonant musical intervals (and across tuning systems, which gave indistinguishable results). g Results of fusion experiment with sung notes, pooled across consonant and dissonant intervals. Here and in (i), plots show the mean ± within-participant SEM. h Results of preference experiments with musical intervals composed of synthetic notes (averaged within consonant and dissonant subsets, and tuning systems). i Results of preference experiments with musical intervals composed of sung notes (averaged within consonant and dissonant subsets). Across all results graphs, asterisks denote statistical significance of pairwise comparisons: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and n.s. not significant. For subplots (b), (c), (f), and (g), two-sided Wilcoxon signed-rank tests were used to test for differences between conditions. For subplots (d), (e), (h), and (i), two-tailed paired t-tests were used to test for differences between conditions.