Characterizing the Interaction of Cultural Evolution Mechanisms in Experimental Social Networks

TL;DR

This work investigates how cultural artifacts evolve under the joint influence of social topology, selection, and reproduction by implementing networked, iterative singing experiments. Using three distinct network topologies and ablation conditions, the study shows that topology combined with selection yields melodies that are more complex and pleasant than in linear transmission, with topology-specific effects that largely vanish when selection or reproduction is removed. The findings demonstrate a critical interaction among mechanisms, showing that network structure can shape cultural evolution only when combined with selection and reproduction biases. Methodologically, the approach blends scalable online experimentation with naturalistic music tasks, offering a framework to study the interplay of cognitive and social processes in cultural evolution.

Abstract

Understanding how cognitive and social mechanisms shape the evolution of complex artifacts such as songs is central to cultural evolution research. Social network topology (what artifacts are available?), selection (which are chosen?), and reproduction (how are they copied?) have all been proposed as key influencing factors. However, prior research has rarely studied them together due to methodological challenges. We address this gap through a controlled naturalistic paradigm whereby participants (N=2,404) are placed in networks and are asked to iteratively choose and sing back melodies from their neighbors. We show that this setting yields melodies that are more complex and more pleasant than those found in the more-studied linear transmission setting, and exhibits robust differences across topologies. Crucially, these differences are diminished when selection or reproduction bias are eliminated, suggesting an interaction between mechanisms. These findings shed light on the interplay of mechanisms underlying the evolution of cultural artifacts.

Figures (5)

• Figure 1: Schematic of the paradigm. A. The three components involved in the cultural process. B. The three topologies considered, a square lattice (periodic boundary conditions not depicted), a random regular graph, and a modular network.
• Figure 2: Emergent melodic prototypes. A. Melodic contours derived from jointly clustering all data (deviation from mean is given in semitones). B. Mean z-scored pleasantness ratings for each melodic cluster. C. Example evolution of prototypes as a function of time from one experimental batch. Nodes are colored based on the cluster of their melody, and edges are highlighted when the neighboring nodes share the same cluster (see Methods). Error bars indicate 95% confidence intervals (CIs).
• Figure 3: Comparison between linear and non-linear (i.e., with topology) evolution. A. Cluster prevalence as a function of iterations. B. Clustered melodies in PCA space across last three iterations (see Methods). C. Average population pleasantness after a three-iteration burn in period (all conditions are initialized randomly and identically). Error bars indicate 95% CIs.
• Figure 4: No selection condition. A. Melodies are sampled uniformly from the local environment for imitation. B. Average population pleasantness after a three iteration burn-in. C. Evolution of neighbor similarity and entropy. Markers indicate different experimental batches (per condition). More transparent colors indicate earlier iterations. D. Average deviation from mean for neighbor similarity and entropy after a three iteration burn-in. Error bars indicate 95% CIs.
• Figure 5: No reproduction condition. A. Selected melodies are mutated through Gaussian noise rather than singing. B. Average population pleasantness. C. Evolution of neighbor similarity and entropy. D. Average deviation from mean for neighbor similarity and entropy. See Figure \ref{['fig:ablation-selection']} for details.