Dynamic systems frameworks allow for examination of the fascinating interaction of stimuli, brain, and behaviour though model development; and provide promising avenues into clinical research. Interpreting model output, however, remains challenging, especially when transitioning from traditional experimental paradigms to more naturalistic behaviours. With the rising interest in computational methods for data analysis from multiple clinically-oriented scientific fields, it is imperative to consider the strengths, limitations, and practicalities of model outputs. This poster will detail methods and findings from a set of music listening studies. Using a combination of Hidden Markov modelling and Partial Least Squares, we observed distinct patterns of between-network brain activity related to behavioural and stimulus features. We will present the workflow used and discuss how these methods can be applied to naturalistic paradigms with an emphasis on interpretation and research with clinical populations.