Executive functions and the generation of “Random” sequential responses: a computational account

When asked to generate sequences of random responses, people exhibit strong and reliable biases in their behaviour. The origins of these biases have been linked to the operation of so-called executive functions through empirical studies varying, e.g., rate of production, modality of response, and (in dual task conditions) secondary task. We present a computational process model of random generation that accounts for a broad range of these empirical effects. The model, which operationalises a previous verbal account of random generation, is grounded in both the cognitive architectures and the executive functions literatures. As such, it instantiates a hypothesis concerning the interaction of multiple distinct executive functions in the generation of complex behaviour. In particular, it is argued on the basis of simulations of empirical findings that three cognitive factors play separable roles in random generation behaviour: cognitive load, which when high exacerbates underlying biases in a generation stage, monitoring, which when impaired results in greater inequality of response usage, and set-shifting, which when impaired results in less frequent switching between response schemas