Introduction

Traditions are behaviors that persist over time and are shared among group members by virtue of social learning processes (see Ch. 1). The direct observation of animals using social cues to discover or learn a behavior is perhaps the most straightforward evidence of a tradition, and numerous longitudinal, naturalistic studies and controlled laboratory experiments have yielded such evidence (see, for instance, Chs. 7, 9, 13, and 14). However, efforts to collect direct evidence are sometimes deemed impractical; consequently investigators have sought ways to infer the existence of traditions on the basis of indirect evidence. Can we identify traditions when we lack direct observations of social learning?

I present a new approach for dealing with indirect evidence. This cue reliability approach (CRA) addresses a special category of potential traditions: behaviors that (a) reflect an individual's classification of a stimulus or tactic as either safe or harmful, and (b) are costly if the individual makes classification errors. Is hemlock a safe food or a dangerous toxin? Should garter snakes be dismissed as benign trespassers or avoided as lethal predators? Animals can answer these questions by consulting local traditions. However, traditional knowledge is not necessarily the only source of information available. The CRA is designed to help us to determine if animals need social cues to classify correctly potentially dangerous stimuli or bad tactics. It begins by identifying a decision-maker's options regarding an unfamiliar stimulus or untested tactic, and the possible outcomes associated with each option.