Embodied language: a review of the role of the motor system in language comprehension
Fischer, M. H., & Zwaan, R. A. (2008). Embodied language: a review of the role of the motor system in language comprehension. Quarterly Journal Of Experimental Psychology, 61(6), 825–850. doi:10.1080/17470210701623605
Summary
This is a review article by Fischer & Zwaan about whether and how the motor system is engaged in language comprehension. They are primarily interested in understanding how internal “simulations” by the motor system of an action affects, or is affected by, language describing the same action—a phenomena they term motor resonance. They don’t give a precise definition of it, beyond that it is the “possibility that language comprehension may incorporate, and possibly even require as an essential component, some activity of the motor system that could be characterized as ‘motor resonance’” (pg. 826).
Fischer & Zwaan first review a set of theories regarding the relationship between perception and action. These include the two-visual-pathways theory (which holds that there are two separate pathways in the brain, one for perception and one for action), the theory of event coding or TEC (which holds that action and perception use the same underlying representations and thus compete for cognitive resources of those representations), mirror neurons (which are neurons that are active when we engage in an action, and when we see others engaging in the same action), and motor cognition (which is based on the idea that “we simulate our own as well as other people’s behavior as part of understanding it”, pg. 831).
Next, Fischer & Zwaan discuss what mechanisms are involved in action simulation. The first mechanism they discuss is computation of affordances, i.e., that “the motor system spontaneously uses object information to compute possible actions in the light of one’s current posture and to select favourable responses” (pg. 831). Similarly, they discuss evidence that planning an action to a particular location can facilitate perception in that area. The second mechanism they discuss is motor resonance during action observation, which seems to be the idea that people use their motor system to make predictions about the actions of others during observation of those actions. The third mechanism is the time course of motor simulation, which they seem to suggest is involved specifically with forward prediction (and therefore facilitates judgements that involve a prediction, as opposed to a reverse inference or something unrelated). However, Fischer & Zwaan also state that “viewing the result of an action activates the processes that would bring about that result”, which is really an inference, not a forward simulation.
Fischer & Zwaan now turn to discussing how the motor system is involved in actual language comprehension. They make the distinction between communicative motor resonance (i.e., motor simulation of speech production) versus referential motor resonance (i.e., motor simulation of the action described by the language). They discuss evidence for communicative motor resonance in lower level phonological processing, and evidence for referential motor resonance in lexical access (i.e. the semantic meaning of individual words) as well as full sentence comprehension (in which information from multiple words must be integrated) and also discourse comprehension (which is, as they note, more ecologically valid as it is how we interpret language in real-world contexts).
They close with three questions. First, the association question is whether activation of the motor system co-occurs when performing a cognitive task (they state that they think the evidence reviewed in this article supports an answer of “yes” to this question). Second, the necessity question is whether the activation of the motor system is required for language comprehension. Third, the sufficiency question is whether activation of the motor system is sufficient for language comprehension.
Takeaways
Fischer & Zwaan seem to use the term “simulation” here to mean neural activation (in the mirror neuron sense) of the same motor regions that are used for actually executing actions. This view, to me, isn’t that useful, though—talking about simulation as neural simulation is either tautological or unrealistic, depending on how you view it. On the one hand, if we recognize actions in other people as the same types of actions we ourselves make, it would be incredibly surprising if there were no shared neural activity. On the other hand, if it is more than just a shared representation and is actually involved in producing the action, then it doesn’t make much sense why motor simulation wouldn’t also produce the action.
I don’t doubt that some sort of simulation that unifies perception, action, and cognition is involved in language comprehension, but talking about it at the level of neural simulation seems like the wrong level of abstraction. For me, it’s more helpful to think of it in terms of something like a mental model, a la Matlock, Bergen et al., or Grush. I don’t think these accounts are at odds with one another: you can still discuss motor simulation at the level of a mental model or emulator, but you don’t need to get into the tricky business of trying to interpret what it actually means to observe correlations of neural activity.
I do like the distinction that Fischer & Zwaan make between communicative and referential motor resonance. I would say that communicative motor resonance has more to do with perception—inferring the underlying motor process that is producing the auditory sensations. On the other hand, the referential motor resonance is more about cognition—constructing a mental simulation of percepts and actions for the purposes of comprehension and interpretation. (I am purposefully not using the terms “embodied” or “grounded” here: while I think it is incredibly useful for the mind to construct models of how perception and action work, I don’t think it is a requirement for all of cognition to ground out in the real world).
Overall, I’m a bit dissatisfied with the account in this article: the overarching hypothesis is that “motor simulation is required for language comprehension” but Fischer & Zwaan don’t really make an attempt to explain why that’s the case. Simply imitating someone else’s action if you don’t know what the action is for doesn’t help you understand it, and I think the same is true for the motor simulation account as well. I agree motor simulation is real, and believe it is involved in comprehension, but I want an account for why it helps and what it brings to the table. The argument that I’d make is that we construct mental models/simulations from which we can make predictions, inferences, etc., and taking actions within the model (i.e. imagining those actions) is the way in which we manipulate the model or test out our predictions. It’s not simply an imitation of the motor action, but a way of identifying and choosing hypotheses about what information a sentence is conveying.