Giorgio Papitto, Angela D. Friederici and Emiliano Zaccarella
Action and language have been proposed to rely on similar hierarchical principles of structural complexity. Support to this proposal is the fact that the processing of both action and language sequences appears to involve analogous neural populations of Brodmann Area (BA) 44, in the left Inferior Frontal Gyrus (IFG). Both levels of comparison (i.e., structural and neural) seem however to suffer from various limitations. At the structural level, we claim that language and action share features of pivotal importance but exclusively when focusing on non-syntactic properties. We suggest that sequencing in the action domain is defined by processing mechanisms mapping sensory inputs and statistical regularities of action-goal relationships, while linearization in the language domain is necessarily dependent upon the underlying syntactic structure. At the neural level, we showed in a recent large functional meta-analysis that left BA44 can be differentiated into two functionally distinct sub-regions, depending on the domain under analysis. More specifically, we observed that while anterior BA44 (aBA44) is consistently recruited during language processing, posterior BA44 (pBA44) is involved during action execution, imitation and imagery, thus in tasks in which mental representations of actions are accessed for linking the input to the instruction to be performed or imagined. At this stage, two questions remain: (i) what type of information is encoded by a mental representation, and (ii) what kind of neural modulation can be seen in BA44 during information encoding? Here we begin to approach (i), by reporting data from a behavioral study (N = 20) pilot for an upcoming event-related functional magnetic resonance imaging (fMRI) experiment. Participants were required to perform finger-tapping sequences with their right hand. In the task, four types of stimuli were presented, each one coding a specific sequence or one of the following classes of sequences: (a) sequences sharing similar motor features; (b) sequences sharing an abstract rule; (c) sequences sharing no specific motor or abstract information. Participants were shown first a class or sequence cue (contextual signal) and then a sequence cue (execution signal), which stands for the sequence to be performed. Results show that contextual signals guide behavior providing help in selecting premotor representations and that these signals can be modulated in terms of the amount of content provided as this modulation is reflected in behavior. In the light of these data, we hypothesize that activity within pBA44 can be modulated by the specificity of information carried by a certain stimulus. Such a finding would be a confirmation concerning the type of processes the sub-region is performing: i.e., retrieving and maintaining mental representations of actions until execution. This would bring evidence to the assumption that action hierarchies are to be found in different representational levels and not in structural ones. Thus, actions might be considered as grounded on causal relationships and statistical regularities that are only ancillary in language. Coherently, the cortical networks involved in language and action could be deemed to have developed independently from one another, even if interactions across the two are still possible.