A neurocomputational model of developmental trajectories of gifted children under a polygenic model: When are gifted children held back by poor environments?

From the genetic side, giftedness in cognitive development is the result of contribution of many common genetic variants of small effect size, so called polygenicity (Spain et al., 2016). From the environmental side, educationalists have argued for the importance of the environment for sustaining early potential in children, showing that bright poor children are held back in their subsequent development (Feinstein, 2003a). Such correlational data need to be complemented by mechanistic models showing how gifted development results from the respective genetic and environmental influences. A neurocomputational model of cognitive development is presented, using artificial neural networks to simulate the development of a population of children. Variability was produced by many small differences in neurocomputational parameters each influenced by multiple artificial genes, instantiating a polygenic model, and by variations in the level of stimulation from the environment. The simulations captured several key empirical phenomena, including the non-linearity of developmental trajectories, asymmetries in the characteristics of the upper and lower tails of the population distribution, and the potential of poor environments to hold back bright children. At a computational level, ‘gifted’ networks tended to have higher capacity, higher plasticity, less noisy neural processing, a lower impact of regressive events, and a richer environment. However, individual instances presented heterogeneous contributions of these neurocomputational factors, suggesting giftedness has diverse causes.