Using Hidden Markov Models to identify cognitive strategies during block construction

This secondary data analysis aims to investigate how children and adults integrate perceptual information from different sources and incorporate it into their planning. To that end, we will use computational modelling on data already collected at the Physical Cognition lab at the Centre for Brain and Cognitive Development in which adults and children constructed models based on different viewpoints. Our analyses focus on examining the order of their construction. The rationale: A construction composed of eight blocks possesses dozens of correct states, and hundreds of correct actions to reach those states. Although the state space for building such a construction is theoretically huge, the paths that adults and children adopt are highly constrained. Despite researchers acknowledging that the construction paths adults and children take are not random, the fundamental question remains unanswered. What determines these constrained construction paths? Construction paths are sequences of Duplo blocks being combined together, with each combination possessing a transitional probability. In Hidden Markov Models (HMMs), these sequences of observable data are known as emissions. HMMs assume that these observable emissions are determined by a sequence of underlying variables, known as hidden states. Since previous research has not been able to explain the highly constrained construction paths through the observable data, HMMs are a strong candidate for testing whether the constrained construction paths can be explained by hidden states. This study will be the first to investigate whether HMMs can explain the constrained construction paths taken by adults and children, computationally.