Exploration for gains and loss-avoidance across the human lifespan

Navigating an uncertain world, we constantly face the explore-exploit dilemma: should we stick with what we know, or explore options that may or may not prove better? An influential theory proposes that humans solve this dilemma through a division of labor in time: explore in childhood, exploit in old age. Our work adds critical empirical nuance to this theory. Exploration can manifest as deliberate information-seeking (“directed exploration”) or choice randomness (“random exploration”), and these strategies are employed differently for gain-seeking and loss-avoidance. Here, we ask how these modes of exploration relate to one another and jointly shape decision-making across the lifespan. A cross-sectional sample of 254 participants aged 8–79 performed a bandit task dissociating directed and random exploration in gain-seeking and loss-avoidance contexts. Factor analysis revealed three factors driving behavior: two domain-specific factors reflecting information-seeking for gains and loss-avoidance, respectively, and a domain-general factor reflecting choice randomness. Directed exploration for gains increased steadily and well into young adulthood before declining in older adulthood, whereas directed exploration for loss-avoidance changed little across the lifespan. Children and older adults, relative to young adults, were characterized by greater choice randomness across contexts, even when exploration was disadvantageous. Though working memory and processing speed accounted for some of these effects, independent effects of age remained. Our findings demonstrate that exploration is not a unitary phenomenon: it encompasses dissociable subprocesses – some domain-specific and others domain-general – with distinct developmental timelines that are shaped by cognitive maturation and decline as well as factors beyond cognitive ability.