AbstractWhat does it mean for metaphors to be “embodied”? Here we describe an influential theory of embodied cognition according to which thoughts are implemented in perceptuo-motor simulations, in the brain’s modality-specific systems. This theory is invoked in nearly every paper on “embodied metaphor,” across linguistics, philosophy, psychology, and cognitive neuroscience. There appears to be overwhelming support for the conclusion that representations of metaphorical “source domains” are embodied in perceptuo-motor simulations. Here we show, however, that when the data are evaluated appropriately there is very little evidence that metaphors are embodied in this sense. The kind of data that offer compelling support for the embodiment of concrete, literal ideas like “grasping the ball” are nearly absent for abstract, metaphorical ideas like “grasping the explanation.” There is now abundant evidence that metaphors structure our thoughts, feelings, and choices in a variety of conceptual domains. But evidence for metaphorical mental representation is not necessarily evidence for embodiment. If any metaphorical source domains are embodied in modality-specific simulations, they may be the exception rather than the rule.

People often accommodate to each other's speech by aligning their linguistic production with their partner's. According to an influential theory, the Interactive Alignment Model, alignment is the result of priming. When people perceive an utterance, the corresponding linguistic representations are primed and become easier to produce. Here we tested this theory by investigating whether pitch (F0) alignment shows two characteristic signatures of priming: dose dependence and persistence. In a virtual reality experiment, we manipulated the pitch of a virtual interlocutor's speech to find out (1) whether participants accommodated to the agent's F0, (2) whether the amount of accommodation increased with increasing exposure to the agent's speech, and (3) whether changes to participants' F0 persisted beyond the conversation. Participants accommodated to the virtual interlocutor, but accommodation did not increase in strength over the conversation and disappeared immediately after the conversation ended. Results argue against a priming-based account of F0 accommodation and indicate that an alternative mechanism is needed to explain alignment along continuous dimensions of language such as speech rate and pitch.

Processing the meaning of action language correlates with somatotopic activity in premotor cortex (PMC). A previous neurostimulation study supported a causal contribution of PMC activity to action verb understanding, but the direction of the effect was unexpected: inhibiting PMC made participants respond faster to action verbs. Here we investigated the effects of PMC excitation and inhibition on action verb understanding using tDCS. Right-handed participants received tDCS stimulation with the anodal electrode (presumed to be excitatory) and cathodal electrode (presumed to be inhibitory) placed over left and right PMC, respectively, or with the reverse configuration. After completing the stimulation protocol, participants made lexical decisions on unimanual action verbs (e.g., throw) and abstract verbs (e.g., think). tDCS configuration selectively affected how accurately participants responded to unimanual action verbs. When the anode was positioned over left PMC we observed a relative impairment in performance for right-hand responses (i.e. the hand with which these participants typically perform unimanual actions). By contrast, when the cathode was positioned over left PMC we observed a relative improvement. tDCS configuration did not differentially affect responses to abstract verbs. These complementary effects of excitatory and inhibitory tDCS clarify the functional role of premotor hand areas in understanding action language.

No abstract