engleski

Sensorimotor forward models accommodate different levels of prediction specificity

Prediction of external events relies on our motor system which is able to simulate the dynamics of our environment by developing internal representations of those events (i.e., forward models) and using them for prediction of incoming stimuli. Results from previous studies suggest that this process is subserved by a basic premotor-parietal network not only when subjects predict the exact identity of the upcoming stimulus, but also when they predict only one of its features. The present study investigates whether the involvement of this network is restricted only to the level of individual stimulus features or if it can generalize to a higher class of abstraction, namely a categorical one based on arbitrary rule-based conjunctions of those features. We directly compared the processing of external events of different specificity, namely perceptual sequences in which prediction of incoming stimuli could be made on the level of individual stimuli (“token”) with those in which predictions were restricted to the level of stimulus categories (“type”). The obtained results confirm that prediction across different levels of abstraction is supported by the same premotor-parietal network without significant modulation of activation within any of its key regions. Furthermore, the processing of two sequence classes could be differentiated by the engagement of additional lateral prefrontal, occipital and posterior temporal regions supporting categorization in “type” sequences. These findings suggest that forward models in perception can be defined on a higher level of abstraction in contrast to those within the motor domain which require high degree of precision and accuracy.

categorization; deviant detection; prediction; premotor cortex; sequencing

nije evidentirano