Rational heuristics shape neurodevelopmental variation

Although bidirectional feedback between brain and behaviour is a cornerstone of probabilistic epigenesis, contemporary theories of neurodevelopmental variation characteristically describe only unidirectional cascades from genes to brain to behaviour. To complement this prevailing view, we present a new theory of the behavioural origins of individual differences and clinical conditions in children. Our thesis is that bounded rational decision-making involving heuristics – cognitive shortcuts that simplify decision-making under uncertainty – contributes unique variance to neurodevelopmental differences over and above fundamental genetic and neurobiological factors. We demonstrate the strength of this account in computational simulations involving artificial agents that when navigating their environments can either (i) engage directly with an information source like speech to meet their aims and learn or (ii) use a fast-and-frugal heuristic like peer imitation to meet their aims. The only difference between agents was learning rate: a proxy for a child’s capacity to acquire new information. When learning rate was low, agents’ reliance on heuristics increased and this – although rational given their constraints – led to substantially reduced learning compared to a theoretical baseline involving an agent with matched learning rate that was programmed to consistently pursue direct engagement. Simulation supports the idea that, perhaps paradoxically, bounded rational decision-making can amplify learning delays over time. The recommendation of this paper is that behavioural feedback should be embedded more widely into causal process models of the origins of neurodevelopmental variation. This account has the capacity to inform how we think about, assess, and respond to individual differences in child development.