Abstract

Nestedness is a pattern that is reliably encountered across diverse cultural datasets: libraries, movie collections, people's answers to trivia quizzes, language's phonological inventories, etc. Nestedness is the fact that items found in small inventories tend to belong to a strict subset of the items found in bigger inventories. For instance, small libraries tend to consist of books that are also found in most other libraries, while bigger libraries contain both common and uncommon books. This talk will discuss the causes and consequences of cultural nestedness. Nestedness mechanically results from the existence of variance in the prevalence of cultural items (like books) and in the size of cultural repertoires (like libraries); any mechanism that makes some items much more prevalent than others (e.g. content-based selection, conformity) can be expected to increase nestedness. Contrary to a common claim, nestedness is not, in itself, a sign of cumulative culture, but it can be increased when a cultural item is adopted based on possessing another item. Although its causes are quite general, nestedness carries interesting consequences that yield specific predictions. I tested two of them. First, if someone possesses a rare cultural item they are likely to possess a more common one: for instance, if you know the answer to a difficult trivia question, you are more likely to know the answer to an easy one. This prediction is verified, and more importantly people know it to be true and use it to infer the competence of others, with high precision. Second, if a cultural item breaks nestedness — for instance, an uncommon movie that is found only in small movie collections, it is likely to be more appealing than its lack of success indicates. We used this idea to detect "hidden gem" movies.