Understanding Action Selection in the Tool Use Network

项目总结:

Project Summary: Skilled use of tools is a defining achievement of human cognition, and is enabled by the storage of tool-specific action memories. Many tools are associated with more than one action, and most everyday tasks are associated with more than one tool. Limb apraxia is a common, disabling, and puzzling left-hemisphere disorder characterized by prominent deficits in activating and selecting task-appropriate tool actions. Little is known about the cognitive mechanisms and brain regions enabling such selection in the neurologically intact brain, or how these processes go awry in apraxia. In several other cognitive domains, it has been suggested that appropriate response selection occurs via biased competition—that is, the prioritization of competing incoming information to enable appropriate response selection. Capitalizing on the promise of such frameworks, we have developed a new functional-neuroanatomic model of biased competition in a specific left hemisphere Tool Use network. Called “Two Action Systems Plus” (2AS+), the model generates testable hypotheses about the major principles determining tool action selection, and their deficiencies in apraxia. Specifically, we hypothesize that 1) Competition between tool actions is influenced by the graded similarity of tool action representations, as implemented primarily by the left posterior temporal cortex (pTC), 2) The outcome of the competitive process is affected by the strength and timing of activation of tool action representations, and depends on the dynamic interplay of left pTC and the parietal lobes, 3) Outcome is further influenced by a mechanism that biases competition towards the tool action that is appropriate to goals and context, as implemented by the left inferior frontal gyrus (IFG) and its connections with the supramarginal gyrus (SMG), and 4) There are two subtypes of apraxia characterized by distinct failures in the competitive selection process: an anterior subtype characterized by inability to appropriately resolve tool action competition, and a posterior subtype reflecting weakened competition. These hypotheses will be tested using a number of complementary methods with healthy and brain-lesioned participants, including voxel-based lesion symptom mapping, resting functional connectivity, fMRI with multi-voxel pattern analyses, and eyetracking. By specifying when and how visuomotor information plays a role in tool representations, the proposed experiments promise to critically constrain “embodied” cognition theories claiming that tools automatically evoke their actions. The proposed research will also advance the theoretical understanding of tool action by anchoring relevant constructs in a cognitive-neuroanatomic model, clarify how action representations are organized and activated, and improve our understanding of the mechanisms affecting errors and re-learning in apraxia, with implications for rehabilitation.