Motor imagery of hand actions: Decoding the content of motor imagery from brain activity in frontal and parietal motor areas

How motor maps are organized while imagining actions is an intensely debated issue. It is particularly unclear whether motor imagery relies on action-specific representations in premotor and posterior parietal cortices. This study tackled this issue by attempting to decode the content of motor imagery from spatial patterns of Blood Oxygen Level Dependent (BOLD) signals recorded in the frontoparietal motor imagery network. During fMRI-scanning, 20 right-handed volunteers worked on three experimental conditions and one baseline condition. In the experimental conditions, they had to imagine three different types of right-hand actions: an aiming movement, an extension-flexion movement, and a squeezing movement. The identity of imagined actions was decoded from the spatial patterns of BOLD signals they evoked in premotor and posterior parietal cortices using multivoxel pattern analysis. Results showed that the content of motor imagery (i.e., the action type) could be decoded significantly above chance level from the spatial patterns of BOLD signals in both frontal (PMC, M1) and parietal areas (SPL, IPL, IPS). An exploratory searchlight analysis revealed significant clusters motor- and motor-associated cortices, as well as in visual cortices. Hence, the data provide evidence that patterns of activity within premotor and posterior parietal cortex vary systematically with the specific type of hand action being imagined.