Modeling Human-Inspired Behavioral Strategies in the El Farol Bar Problem

Abstract

Individuals deciding whether to attend popular social venues often seek to maximize their satisfaction by avoiding overcrowding. The El Farol Bar Problem models how individuals repeatedly decide whether to attend a bar while attempting to achieve this goal. While agents in the original El Farol model use prediction-based attendance strategies, less is known about how human-inspired behavioral strategies influence crowding and population-level happiness. We modified the NetLogo implementation of the El Farol problem by replacing the prediction-based attendance mechanism with four decision heuristics: Routine (fixed attendance patterns), Repeat Success (win-stay/lose-shift), Explore (random exploration), and Recent Experience (recent-outcome adaptation). Simulations of 100-agent populations compared balanced populations with populations in which 70 agents shared the same behavioral strategy and the remaining 30 agents were evenly distributed among the other three strategies. We evaluated outcomes using attendance variability, overcrowding frequency, and normalized happiness, a measure of satisfaction relative to the maximum achievable level. Among the dominant strategy populations, Explore produced the strongest overall performance, achieving 76.5% normalized happiness, the lowest attendance variability (SD = 8.88), and overcrowding in only 29.9% of weeks. Routine also performed well, maintaining 74.6% normalized happiness, low attendance variability (SD = 9.27) and 32.3% overcrowding frequency, the second lowest. In contrast, Repeat Success performed worst overall, with normalized happiness falling to 27.7%, attendance variability increasing substantially (SD = 33.66), and overcrowding occurring in 50.0% of weeks. Recent Experience produced intermediate normalized happiness (45.4%) and attendance variability (SD = 22.10), but the highest overcrowding frequency (58.1%). Balanced populations maintained 62.5% normalized happiness, indicating that behavioral diversity can offset the effects of less effective decision strategies. Overall, these findings suggest that simple behavioral rules can outperform more sophisticated adaptive strategies, while greater behavioral diversity may enhance population-level happiness in agent-based models.