Inferential power in identifying frequency-dependent social learning strengthened by increasing behavioural options

Area of Science:

• Behavioral Ecology
• Evolutionary Biology
• Animal Behavior

Background:

• Two-option choice designs are standard for studying social learning in animals.
• Real-world animal behavior involves multiple simultaneous choices, unlike typical experimental setups.
• Small sample sizes in animal social learning studies limit the identification of specific learning strategies.

Purpose of the Study:

• To investigate how increasing the number of behavioral options (from 2 to >2) impacts the ability to identify social learning strategies.
• To assess the effect of sample size and the number of choices on inferential power in social learning research.
• To compare the effectiveness of different frequency-dependent social learning strategies (conformist, unbiased, anti-conformist) under varying experimental conditions.

Main Methods:

• Simulated 7200 datasets across 72 parameter combinations, varying the number of options (k=2-5), population size (n=5-250), and strength of frequency dependence.
• Applied a Bayesian social learning model to analyze simulated data.
• Evaluated model performance based on posterior probabilities, standard deviations, highest posterior density intervals, and median estimates of frequency dependence.

Main Results:

• Increasing the number of behavioral options significantly enhances the accuracy and certainty of identifying social learning strategies.
• These improvements are most notable in small to intermediate sample sizes, common in empirical research.
• The study highlights the importance of considering unchosen options in social learning analyses.

Conclusions:

• Expanding the number of choices in experimental designs can increase the inferential power for identifying social learning strategies without needing larger sample sizes.
• This approach offers a more efficient method for studying animal social learning, particularly in field settings.
• Understanding the full behavioral repertoire, including unchosen options, is crucial for robust social learning research.