How does language reliably evoke emotion, as when reading a favorite novel or listening to a skilled orator? Recent evidence suggests that comprehension involves a mental simulation of sentence content that calls on the same neural systems used in literal action, perception, and emotion. Here we demonstrate a causal role of involuntary facial expression in the processing of emotional language. Subcutaneous injections of botulinum toxin-A (BTX) were used to temporarily paralyze the facial muscle used in frowning. We found that BTX selectively slowed the reading of sentences with content that normally requires the paralyzed muscle for expression of a congruent emotion. This finding demonstrates a role of peripheral feedback in language processing, supports facial feedback theories of emotional cognition, and raises questions about effects of Botox on cognition and emotional reactivity. We account for the role of facial feedback in language processing by considering neurophysiological mechanisms and reinforcement learning theory.Is language comprehension the manipulation of abstract symbols by rules of syntax (e.g., Chomsky, 1959)? Recent behavioral and neuroscientific evidence suggests that comprehension involves a mental simulation of sentence content (Barsalou, 1999) that calls on the same neural systems used in literal action, perception, and emotion (Glenberg & Kaschak, 2002;Havas, Glenberg, & Rinck, 2007;Niedenthal, 2007;Pulvermüller, 2005). Here we demonstrate a causal role of involuntary facial expression in processing of emotional language. Havas et al. (2007) provide evidence for emotion simulation in language. Participants read and judged the valence of sentences describing pleasant situations ("You execute the difficult dive flawlessly") and unpleasant situations ("The police car pulls up behind you, siren blaring") while reading times were measured. Participants were instructed to hold a pen either in the teeth (to produce a smile) or in the lips (to prevent a smile) while reading. This procedure reliably induces emotional states in participants in the absence of awareness (Strack, Martin, & Stepper, 1988). As predicted, reading times for sentences describing pleasant situations were faster while participants were smiling than while they were prevented from smiling, and the reverse was found for unpleasant sentences. Thus, facial As a prelude to the current research, previous work (e.g., Niedenthal, Winkielman, Mondillon, & Vermeulen, 2009) has demonstrated that reading words describing emotions selectively activates facial muscles: Negative emotion words activate corrugator supercilii and positive emotion words activate zygomaticus. In addition, in an unreported experiment using the sentences described below, we verified that our happy sentences activated zygomaticus more than did sad and angry sentences, whereas sad and angry sentences more strongly activated corrugator supercilii than did happy sentences (see supporting information available on-line).The question we tested here is whether th...

436Reading a passage from a favorite novel makes it clear that language evokes emotion. In addition, empirical studies have shown that emotions that are evoked by language can be powerful (Velten, 1968) and can have an impact on judgments (Johnson & Tversky, 1983). Nonetheless, the interaction between emotion and language is not well understood. Oatley (1994) suggests that since Aristotle, one explanation is based on the notion of simulation. We briefly review recent theory, data, and methods investigating simulation in language comprehension and in social cognition before introducing a new method to study the relation between emotion and language comprehension. In this article, we present initial evidence consistent with (1) a simulation account, and (2) the idea that the effect of emotion on language comprehension is likely to result from processes other than initial lexical access.By "simulation during language comprehension," we mean that a neural process ordinarily having a nonlinguistic function (e.g., action, perception, emotion) is also used during language comprehension. In addition, the simulation may be functional in that it plays a causal role in language comprehension. As we will review shortly, the case for functional simulation is strongest in the domains of action and perception. The evidence that we present for simulation using emotion is only initial; that is, the functional role has yet to be secured.The case for simulation of action during language comprehension is very strong because of converging evidence from behavioral, physiological, and neuroscience techniques. Glenberg and Kaschak (2002; see also Zwaan & Taylor, 2006) demonstrated a relation between action and language comprehension by using behavioral methods. Participants read sentences describing action away from the body (e.g., You handed Courtney the notebook) or toward the body (Courtney handed you the notebook) and judged whether the sentences were sensible. Participants responded "sensible" by reaching toward a button far from the body in one condition and toward the body in the other condition. Participants were faster to make "sensible" judgments when the action that was implied by the sentence (e.g., away from the body) was compatible with the action required to make the response. These data support the claim that language comprehension involves simulation grounded in neural systems for action. Using fMRI, Hauk, Johnsrude, and Pulvermüller (2004) demonstrated that the motor and premotor cortex are activated in a somatotopic fashion while silently reading action verbs, such as kick, pick, and lick. That is, in addition to activation in traditional language areas, understanding the word kick initiated activation in areas of the motor and premotor cortex that control the leg, whereas understanding the word pick initiated activation in areas that control the hand, and so on. Evidence consistent with the claim that the simulation is functional comes from the use of transcranial magnetic stimulation (TMS). Thus, Buccino et al. (2...

No abstract

There is no question that language and emotion affect one another. When we read a novel, the words may thrill, petrify, or titillate, and when listening to a bigot the words may invoke anger. Embodied approaches to language and cognition have a ready explanation for these effects: Language is understood through a simulation process that calls upon neural systems used in perception, action, and emotion. Thus, simulating the language produces much the same effect as being in the situation.We begin by briefly reviewing data demonstrating the embodiment of language; that is, how perception, action, and emotion systems are used in language comprehension. Next, we review work showing gender differences in emotional reactivity. Given these gender differences, we are led to the following prediction: Men and women will differ in the processing of language about emotional situations. The results from several projects document those differences. There is an important caveat to keep in mind, however. Although men and women differ in their emotional reactivity, in fact the similarities are much greater than the differences (e.g., Bradley, Codispoti, Sabatinelli, & Lang, 2001). Similarly, whereas we document theoretically significant gender differences in the processing of emotional language, the effect sizes are small. Thus it would be incorrect to assume that these differences characterize everyone to the same extent. AbstractLanguage comprehension requires a simulation that uses neural systems involved in perception, action, and emotion. A review of recent literature as well as new experiments support five predictions derived from this framework. 1. Being in an emotional state congruent with sentence content facilitates sentence comprehension. 2. Because women are more reactive to sad events and men are more reactive to angry events, women understand sentences about sad events with greater facility than men, and men understand sentences about angry events with greater facility than women. 3. Because it takes time to shift from one emotion to another, reading a sad sentence slows the reading of a happy sentence more for women than men, whereas reading an angry sentence slows the reading of a happy sentence more for men than for women. 4. Because sad states motivate affiliative actions and angry states motivate aggressive action, gender and emotional content of sentences interact with the response mode. 5. Because emotion simulation requires particular action systems, adapting those action systems will affect comprehension of sentences with emotional content congruent with the adapted action system. These results have implications for the study of language, emotion, and gender differences.

No abstract