Multiple weak biases support adaptive choices without prior experience: a self-supervised strategy.

From social interaction to foraging, the survival of newborn animals depends on their ability to make adaptive decisions without prior experience. While ethological research focused on strong innate biases, recent empirical evidence shows that early preferences are often weak, transient and variable. After reviewing these findings, we introduce a novel model that explains how multiple weak biases can support adaptive choices in naive, inexperienced animals. We propose a self-supervised strategy in which naive animals combine internal predispositions with environmental structure to guide behaviour. First, predispositions target features that are rare in the background but common in relevant stimuli (e.g. red colour, upward movement), thus reducing false positives. Second, the evolution of multiple weak biases allows animals to increase accuracy of decisions using co-occurring cues (e.g. colour, sound and movement). Weak biases are advantageous as they prevent strong responses to misleading partial matches. Requiring the co-occurrence of multiple rare cues supports effective choices in complex environments with many misleading stimuli. This simple self-supervised strategy offers a new framework for understanding early decision-making, with implications that span across behavioural ecology, developmental psychology, comparative cognition and artificial intelligence, where decision-making with sparse evidence is a crucial challenge.