Educational Robotics, Problem-Solving and Computational Thinking in Elementary-Aged Autistic Learners: A Critical Narrative Review
Rebeca Preethi *
Gwinnett County Public Schools, Suwanee, Georgia, United States.
*Author to whom correspondence should be addressed.
Abstract
Educational robotics is increasingly used to teach programming and problem-solving in primary education, yet evidence concerning elementary-aged autistic learners is fragmented across special education, computer science education and socially assistive robotics. This critical narrative review evaluates what existing research can establish about robotics-related computational thinking, task-specific problem-solving and the transfer of these capabilities beyond trained activities. Literature published from 2006 to 23 August 2026 was considered, supplemented by earlier methodological work where relevant. Research was examined through an evidence hierarchy distinguishing direct studies of autistic primary-aged learners from studies involving older autistic learners, general-population primary pupils and social-robot interventions targeting non-academic outcomes. The strongest direct evidence demonstrates that some autistic pupils can acquire explicitly taught robot-programming sequences under structured instruction; it does not establish equivalent gains in independently assessed computational thinking or general problem-solving. Elementary after-school programmes support feasibility and engagement, but perceived improvement and participation are not interchangeable with cognitive learning effects. Research involving non-autistic primary pupils provides a more substantial, although heterogeneous, basis for expecting benefits in particular programming concepts. Its findings include positive controlled effects, null results and evidence that adding a physical robot can increase extraneous cognitive load relative to screen-only programming. Socially assistive robot studies contribute insight into interaction and accessibility, but their social outcomes cannot be pooled with coding or reasoning outcomes. Important methodological limitations include small and selected samples, bundled interventions, variable baseline ability, inconsistently defined constructs, narrow assessment tasks and rare testing of maintenance or cross-context transfer. An integrated framework is advanced in which robotics can create opportunities for accessible feedback, guided debugging and collaborative reasoning, provided instruction, environmental conditions and assessment are individually adapted. Priority is given to adequately powered, ethically inclusive comparative studies with accessible independent measures, documented implementation fidelity and meaningful longitudinal follow-up. Educational robotics is a plausible, adaptable pathway to computational participation, not an independently established method for improving general problem-solving among autistic elementary pupils.
Keywords: Autism spectrum, primary education, algorithmic reasoning, coding instruction, accessible assessment, transfer of learning, inclusive STEM, embodied learning