This thesis reports the results of three novel studies that used effective connectivity and region of interest methods on fMRI data to provide insights into how familiarity shapes action observation and execution. Experiment 1 examined how observed movement familiarity modulates AON activity using dynamic causal modelling, a type of effective connectivity analysis. Participants viewed whole-body dance movements during scanning, and after scanning rated each movement on a measure of visual familiarity. These ratings were then applied as parametric modulators to the fMRI data, which revealed an attenuation of effective connectivity bidirectionally between parietal and temporal AON nodes when participants observed videos they rated as increasingly familiar. As such, the findings provide partial support for a predictive coding model of the AON, as well as illuminate how action familiarity manipulations can be used to explore simulation-based accounts of action understanding. Experiment 2 examined the relationship between AON response amplitude and participants’ familiarity with observed or executed actions. Specifically, this study examined whether increasing familiarity impacts AON engagement in a linear or quadratic manner. Using an elaborate guitar training intervention to probe the relationship between familiarity and AON engagement during action execution and action observation tasks, participants underwent fMRI scanning while executing one set of guitar sequences and observing a second set of sequences. Via region of interest analyses, linear, cubic and quadratic regression models were fitted to the data to match varying levels of familiarity. The data from the observation and execution conditions show mixed evidence for all types of models, suggesting that the response profile within key sensorimotor brain regions associated with the AON is not solely linear in nature in response to increasing familiarity. Moreover, by probing the objective and subjective nature of the prediction error signal, we show results that are consistent with a predictive coding account of AON engagement during action observation and execution. The final study of this thesis aimed to test the ssumptions of the predictive coding account using effective connectivity, when participants observed or executed familiar compared to unfamiliar actions. To test these predictions, we re-evaluated the same data collected for Study 2 wherein participants took part in an intensive guitartraining paradigm. Identifying core AON nodes from pre- and post-training scanning sessions, we then applied effective connectivity analyses to test whether changes in effective connectivity fit those hypothesised under the predictive coding account. We demonstrate that hypotheses derived from predictive coding that predict distinct patterns of modulation based on perceived or performed actions’ familiarity are generally supported by the empirical data. These findings contribute valuable insights toward understanding the complex role played by familiarity in modulating action cognition. The main empirical findings of this thesis show: 1) attenuation in connectivity within the AON when an action is perceived as more familiar; 2) the response profile of core AON regions to increasing familiarity (either objectively or subjectively defined) when performing or observing an action is not solely linear in nature; and 3) hypotheses derived from the predictive coding account concerning effective connectivity between core AON regions are largely supported when an intensive training paradigm is used to create a distinction between familiar and unfamiliar actions.