Discovery-based learning designs incorporating active exploration are common within computer-based instructional simulations, supported by constructivist theories of learning focussing on active individual knowledge construction. On the other hand, researchers have highlighted empirical evidence showing that ‘pure’ discovery learning is of limited value and that combinations of explicit instruction and guided discovery learning are more effective. Little is known, however, about differences in the cognitive processing that occurs when a learner undertakes active discovery learning using a computer-based simulation compared to when they are guided through observation of simulation output. A better understanding of the cognitive processing occurring when learners interact with on-line materials in the context of specific learning designs is important both for networked learning researchers and for on-line teachers. This paper reports on a study in which the brain activations from two learning conditions using computer-based simulations were compared using functional magnetic resonance imaging (fMRI). One condition allowed exploration through manipulation of simulation parameters, while the other allowed observation of simulation output from preset parameters. Drawing on constructivist theories of learning, it was hypothesised that the active exploration condition would lead to greater activation of brain areas associated with working memory organisation and long term memory formation. The study also set out to explore the broader feasibility of using fMRI to explore learners’ cognitive processing while undertaking holistic learning activities using on-line learning materials. Results of the study were somewhat equivocal about differences in brain activation with no consistent differences in activation between the two conditions able to be measured. Consistent with our related research which suggests that discovery learning strategies vary substantially across individuals, results of this study suggest that the cognitive processing during the two conditions varied across participants. Integrated analysis of the exploration processes, learning outcomes and measured brain activations of individuals shows promise in better understanding the relationship between learning strategy, interaction and cognition when using instructional simulations. Approaches like this that draw on an analysis of data on learning process and outcome along with an analysis of physiological measures (in this case blood flow as an indicator of brain activation) are expected to be at the leading edge of learning analytics research in coming years. The study also highlighted challenges associated with the use of fMRI to explore learners' cognition while undertaking learning activities allowing significant learner control and involving extensive computer-based interaction.