Neural encoding and functional interactions underlying pantomimed movements

Abstract

Pantomimes are a unique movement category which can convey complex information about our intentions in the absence of any interaction with real objects. Indeed, we can pretend to use the same tool to perform different actions or to achieve the same goal adopting different tools. Nevertheless, how our brain implements pantomimed movements is still poorly understood. In our study, we explored the neural encoding and functional interactions underlying pantomimes adopting multivariate pattern analysis (MVPA) and connectivity analysis of fMRI data. Participants performed pantomimed movements, either grasp-to-move or grasp-to-use, as if they were interacting with two different tools (scissors or axe). These tools share the possibility to achieve the same goal. We adopted MVPA to investigate two levels of representation during the planning and execution of pantomimes: (1) distinguishing different actions performed with the same tool, (2) representing the same final goal irrespective of the adopted tool. We described widespread encoding of action information within regions of the so-called "tool" network. Several nodes of the network-comprising regions within the ventral and the dorsal stream-also represented goal information. The spatial distribution of goal information changed from planning-comprising posterior regions (i.e. parietal and temporal)-to execution-including also anterior regions (i.e. premotor cortex). Moreover, connectivity analysis provided evidence for task-specific bidirectional coupling between the ventral stream and parieto-frontal motor networks. Overall, we showed that pantomimes were characterized by specific patterns of action and goal encoding and by task-dependent cortical interactions.

Keywords: Action; Connectivity; Goal; MVPA; Motor system; Pantomime; Tool; fMRI.

Fig. 1

Timeline of the experimental trial. The trial started with a verbal cue instructing the subject about the type of action to pantomime (duration 1 s). After 9 s of delay (planning), the subject was instructed with an auditory cue (‘beep’) to perform the pantomime (execution) with the right dominant hand. After 2.5 s, another auditory cue (‘beep’) indicated the end of the trial. The participant waited for a new cue to start the following trial (inter-trial interval, ITI, duration 11.5 s)

Fig. 2

A MVPA pairwise comparisons. To test for action encoding (blue box), we performed an MVPA analysis, comparing trials for the grasp-to-move and grasp-to-use conditions for a specific tool. To test for goal encoding (red box), we adopted cross-decoding. We trained the classifier on the pairwise comparison between grasp-to-move vs grasp-to-use for one tool (e.g. axe) and then tested the classifier for the same comparison on the other tool (e.g. scissors). Then, we performed the same analysis but switching the trials adopted for training (e.g. scissors) and testing (e.g. axe) the classifier (B). Cross-validation approach. Each experimental run comprised 16 trials, 4 for each experimental condition (left part of the panel). Decoding accuracy was estimated with a leave-one-run-out cross-validation approach. We trained the classifier on single trials of seven runs (4 trials × 7 runs per condition) and then testing the classifier on the trials of the remaining run (4 trials × condition). For action representation (blue box in the upper part of p anel [B]), we trained the classifier on a specific pairwise comparison (e.g. grasp-to-use scissors vs. grasp-to-move scissors) and tested the classifier on the same pairwise comparison. We did this procedure separately for the two tools and then averaged the decoding accuracy values for these two comparisons. For goal representation (red box in the upper part of panel [B]), we trained the classifier on one pairwise comparison (e.g. grasp-to-move vs grasp-to-use with the axe) and then tested the classifier on the same comparison on the other tool (e.g. scissors). Then, we performed the same analysis but switching the trials adopted for training (e.g. scissors) and testing (e.g. axe) the classifier. Finally, we averaged the decoding accuracy values for these two decoding combinations

Fig. 3

Connectivity analysis: ROIs position and connections. Visual representation of the three ROIs considered for connectivity analysis: aIPS, PMv and pMTG. The black arrows represent the intrinsic (bidirectional) connections between the nodes considered in the DCM analysis. We assumed that information (auditory input) enters the system from pMTG

Fig. 4

Models tested for the modulatory effects. We tested the modulatory effect of the ‘grasp-to-use’ task considering all the possible meaningful combinations of forward and backward modulatory connections between the considered nodes. A total of fifteen models were tested. The connections considered in each model are schematically represented in the image. To allow a direct comparison, we adopted the same numbers for the models in Table 3

Fig. 5

Summary of ROI-based MVPA results. This panel depicts a schematic representation of the results for the two phases of the task. We indicated ROIs with a significant decoding for action information with blue circles (FDR-corrected, q < 0.05) and for goal information with red circles (FDR-corrected, q < 0.05). A black circle indicates no significant decoding for the considered comparisons (FDR-corrected, q < 0.05)

Fig. 6

Decoding results for ROI-based MVPA. The bar graphs show the average decoding accuracy for ‘concrete’ action encoding (blue) and for ‘abstract’ goal encoding (red), separately for the planning phase (left) and in the execution phase (right). SPOC is located on the medial part of the brain, so it is represented outside the rendering of the template. Significant decoding is indicated with asterisks (p < 0.05*, p < 0.005**, FDR-corrected, q < 0.05, red star)

Fig. 7

DCM Results. A. Intrinsic connectivity. The red arrows indicated positive coupling between two nodes, whereas the green arrows indicated negative between two nodes (q < 0.05 FDR-corrected). There was a bidirectional increased coupling between pMTG and PMv and unidirectional from pMTG to aIPS and from PMv to aIPS. Reduced communication was evident between aIPS and the other two functionally connected nodes B. Modulatory connectivity of the winning model. The ‘grasp-to-use’ task positively modulated the reciprocal connections between pMTG and PMV which showed enhanced coupling. The communication between aIPS and the other two nodes was instead reduced