The grounding of abstract concepts in the motor and visual system: An fMRI study

The grounding of concepts in the sensorimotor brain systems is controversially discussed. Grounded cognition models propose that concepts are represented in modality-specific sensorimotor, but also emotional and introspective brain areas depending on specific experiences during concept acquisition. Accumulating evidence suggests that concrete concepts are closely linked to modality-specific systems, whereas the mere existence of abstract concepts seems to contradict grounded cognition approaches. Here, using functional magnetic resonance imaging, we adopted a theory-driven approach frequently used for investigating concrete concepts to the domain of abstract concepts: We compared brain activation to abstract concepts with a known motor versus visual feature content as determined by a previous property listing study. Carefully matched motor (e.g., fitness) and visual (e.g., beauty) abstract words were presented to 24 participants along with pseudowords while performing a lexical decision task. Furthermore, participants performed two localizer tasks by actually moving their hands (motor localizer) and by looking at real pictures (visual localizer). Processing of motor abstract words specifically activated frontal and parietal motor areas, whereas processing of visual abstract words specifically elicited higher activity in temporo-occipital visual areas, albeit at a more lenient statistical threshold. According to inclusive masking analyses, this differential activity pattern to motor and visual abstract concepts overlapped with brain activations observed during hand movements (pre- and postcentral gyrus) and object perception (fusiform and lingual gyrus). Thus, consistent with the grounded cognition framework, our results suggest that, similar to concrete concepts, abstract concepts related to action and vision are grounded in modality-specific brain systems typically engaged in actual perception and action depending on their conceptual feature content.