Presentation Information

[C15-03]Exploration–exploitation trade-off in cultural evolution of technology: An experiment in a two-generation framework

*Yutaka Kobayashi1, Yo Nakawake2 (1. Kochi Univ. of Tech. (Japan), 2. JAIST (Japan))

Keywords:

cultural evolution,human behavior,learning,technology,laboratory experiment,innovation,cultural transmission,agent-based simulation,prosocial behavior,decision making

Exploration of novel technologies often comes at the expense of opportunities to gain immediate benefits by exploiting existing technologies. The relative importance of exploration to exploitation depends on how the current generation values future beneficiaries based on factors such as genetic relatedness. Evolutionary theory and classical economics both predict that the transmission of technologies to non-kin does not promote innovation or even inhibits it. However, human behaviors may be neither optimal nor rational, as recent experiments suggest. In the present study, therefore, we test the above theoretical prediction with a laboratory experiment, which simulates the cultural evolution of technologies with an exploration-exploitation trade-off in a two-generation framework. Participants were randomly paired to play the role as either the 1st or 2nd generation, and engaged in the “virtual arrowhead task”; they designed an arrowhead on the screen by changing the values of three attributes (i.e., height, width, and depth). In each of 50 trials, participants either explored a new design or went hunting to get points using the arrowhead on hand. Better designs yielded higher scores, but the underlying fitness landscape was concealed. We compared three conditions: (1) “asocial”: no transmission occurred, (2) “unrepaid”: the final arrowhead design of the 1st generation was transmitted to the 2nd generation, and (3) “repaid”: transmission occurred and the 1st generation received an extra reward equivalent to the performance of the 2nd generation. We tested two hypotheses: (i) the frequency of exploration is higher in the unrepaid than asocial condition (human prosocial tendency is assumed), and (ii) it is higher in the repaid than asocial condition. We analyzed a formal model and conducted agent-based simulations to confirm that hypothesis (ii) holds true for rational agents, whereas hypothesis (i) not. The experimental results indeed supported hypothesis (ii) but not (i), agreeing with the theoretical prediction. These results imply that humans do not increase investment in exploration of novel technologies to benefit anonymous successors; instead, incentive mechanisms, genetic relatedness, or reduced anonymity, is required. This presentation is based on a published work by Nakawake and Kobayashi (2024, Royal Society Open Science 11, 231108. DOI: 10.1098/rsos.231108).