Distinct vocal flexibility encodes food identity in marmoset monkeys

Abstract

Combinatorial vocal behavior, in which discrete vocal elements are recombined to expand the vocal repertoire, is a hallmark of human language and has been observed in some nonhuman primates. Here, we show that marmoset monkeys produce distinct vocalizations that discriminate food categories and items by producing long sequences dominated by a single call type whose spectral and temporal features vary. These sequences carried reliable information about food type that could be accurately predicted by trained classifiers. Moreover, rather than combining distinct call types, marmosets rely on sequencing and graded acoustic variation within a single call class to generate functionally diverse signals. These results suggest that precursors of linguistic combinatoriality and multiparameter coding are not unique to human speech but are present in nonhuman primate communication.