To what extent does object knowledge bias the perception of goal-directed actions?

Predictive processing accounts of action understanding suggest that inferred goals generate top-down predictions that bias perception towards expected goals. These predictions are thought to be derived, in part, from the affordances of available objects. This thesis had three aims: (1) to test whether high-level action goals based on object knowledge can bias action perception, (2) to investigate the degree to which this perceptual bias can be influenced by high-level person knowledge, or by expertise in particular objects, (3) to explore the low-level mechanisms underlying the anticipatory representation of action goals associated with objects. Experiments used a modified representational momentum paradigm, as well as RT-based measures. In Chapter 2, we found that the presentation of a prime object led to a predictive bias in the perception of a subsequent action towards a functionally related target object. This bias was present for reaching actions, but not withdrawing actions (Experiment 1a) and persisted even when the functionally related target was simultaneously presented with an unrelated distractor (Experiment 1b). Crucially, this effect was specific to intentional actions, but was eliminated when the hand was replaced by a non-biological object following the same trajectory. This finding supports predictive processing views that action perception is guided by goal predictions, based on prior knowledge about the context in which the action occurs. We found no evidence that this perceptual bias could be influenced by prior knowledge about the gender of the actor (Chapter 3) or by participants' expertise in particular objects (Chapter 4). Chapter 5 tested for motor biases resulting from object-based goal predictions. Originally designed as a TMS study (Experiment 4a), this was tested online using RT measures as an index for motor preparation. We found no evidence that object affordances can be reliably measured using online RTs. Taken together these findings highlight the important role of object knowledge in action perception, while showing the limits to which this might be modulated by person knowledge and expertise. The final chapter highlights the challenges of developing robust behavioural measures for online testing of object affordances.