Unravelling the cognition of coding in 3-to-6-year olds

There is growing interest in teaching children computer programming ("coding") to prepare them for the demands of our increasingly digital society. However, we do not yet understand what cognitive skills children need in order to learn to code. The aim of our research program is to identify the requisite skills, with the goal of building a cognitive model of coding. The present research used a wooden robot ("Cubetto", www.primotoys.com) to investigate coding ability in young children. Exp. 1 describes the development and evaluation of the assessment instrument, which was tested with 18 3-to-5-year-old children. The instrument ("Coding Development (CODE) Test 3-6") was used in Exp. 2 to investigate the relationship between coding skill and "cognitive compiling" - the ability to formulate mental action plans in natural language. Thirty 5-to-6-year-olds participated in Exp. 2. Using Bayesian statistics, we found evidence that cognitive compiling predicts coding performance above and beyond age and nonverbal intelligence. We evaluate the outcomes and reflect on whether cognitive compiling depends solely on maturation or might be a skill that can be trained, and if so, how this could be done.