Mapping Metaphor: This Is Your Brain on Figurative Language

“Children may not understand political alliances or intellectual argumentation, but they surely understand rubber bands and fistfights.”–Steven Pinker, from The Stuff of Thought: Language as a Window into Human Nature (Viking, 2007).

Sometimes a cigar is just a cigar. Then again, mischief is the hot smoke that curls off the end of a lit intellect. And sometimes a diamond in the rough is indeed just an ancient deposit of highly compressed carbon. But no facet of humanity’s evolved “genius,” as Aristotle put it more than 2,300 years ago, sparkles so brilliantly as our unique capacities for extra-literal description and comprehension.

Until recently, most professional sources have attributed our proficiency with language to a pair of knuckle-sized regions on the brain’s left side called Broca’s area and Wernicke’s area. The former module was responsible for grammar, the latter for word meanings. But new technologies, featuring functional Magnetic Resonance Imaging (fMRI), have allowed scientists to probe non-invasively into the brains of healthy volunteers and to discover, first, that other parts of the brain share in these responsibilities and, second, that Broca’s and Wernicke’s regions contribute to other important tasks as well.

For some, such frustrating complexity spells the end of the “modularity” hypothesis that instructs the functional specialization of certain identifiable neural systems. For others, including psychologist Gary Marcus, author of The Birth of the Mind (Basic Books, 2004), new evidence suggests “not that we should abandon modules (the Swiss Army Knife view of the brain) but that we should rethink them–in light of evolution.”

While he acknowledges that the brain’s left hemisphere appears to be devoted to both language and problem solving, R. Grant Steen, psychologist, neurophysiologist, and author of The Evolving Brain (Prometheus, 2007), agrees with Marcus, defining language (as opposed to mere communication) in very practical, adaptive terms:

[L]anguage is a system of communication that enables one to understand, predict, and

influence 

the action of others. Inherent in this definition is a concept of theory of mind: if communication is instinctual rather than having a purpose, then it should probably not be considered a language. If communication has a purpose, this assumes an awareness of other independent actors, whose actions can potentially be influenced.… [F]or communication to serve the needs of the listener as well as the needs of the speaker, the listener must be able to understand what the speaker is “really” saying. It is not enough to understand the literal meaning of speech.

Two intimately associated paradigms have suffered intense scrutiny in recent years. The standard model of figurative language processing–sometimes referred to as the “indirect” or “sequential” view–maintains that the brain initially analyzes passages for literal meaning and, if the literal interpretation makes no sense, then reprocesses the words for access to an appropriate figurative meaning. According to the related dichotomous model of “laterality,” the brain’s left hemisphere (LH) is responsible for processing literal language while its right hemisphere (RH) is enlisted only to decode figurative expressions.

Such paradigms were based on classic lesion studies beginning with those conducted in 1977 by Ellen Winner and Howard Gardner who showed that patients with RH damage had much more difficulty processing metaphors than subjects with LH damage. However, in an editorial from the February 2007 issue of Brain and Language, linguistics expert Rachel Giora argued that Winner and Gardner’s results had been widely misinterpreted. Although only the LH patients in the lesion studies were able to competently match metaphorical figures with their corresponding pictures, Giora explained, it was not true that the RH patients were unable to make such connections when asked to do so verbally. Indeed, a number of studies published at the turn of the century challenged the notion that RH damage selectively impairs people’s command over verbal figurative language.

During the last few years, researchers have begun to dissect the old paradigms more systematically. In August of 2004 Alexander Rapp’s team of German scientists published a report in Cognitive Brain Research titled “Neural Correlates of Metaphor Processing.” They used event-related fMRI technology to detect brain activity in sixteen healthy subjects as they read short, simple sentences with either a literal or a metaphorical meaning.

Consistent with the laterality model, Rapp had predicted that metaphorical versus literal sentences would induce more vigorous brain activity in participants’ right lateral temporal cortices. Instead, the strongest signal disparities occurred in the subjects’ LH, the left inferior frontal and temporal gyri, or cortical folds, in particular. In possible contradiction to the indirect or sequential view of metaphor processing, Rapp’s study noted as well that neither response times nor accuracy diverged between the two conditions. In summary, the team advised their colleagues to reassess the RH theory of figurative language comprehension and posited that, although the RH appeared to play some important role, factors other than figurativity per se might be involved.

Two years later, cognitive scientists Zohar Eviatar and Marcel Adam Just published a similar study, “Brain Correlates of Discourse Processing: An fMRI Investigation of Irony and Conventional Metaphor Comprehension” in Neuropsychologia. There, sixteen subjects digested ironic sentences (e.g., When Ann came home covered in mud, her mother said, “Thanks for staying so clean.”) in addition to literal and simple metaphorical expressions.

As one might guess, the results were considerably more complicated. First, all three types of statements stimulated the classical language areas of the LH: moving roughly from front to back, the left inferior frontal gyrus, the left inferior temporal gyrus, and the left inferior extrastriate region. Second, metaphorical sentences activated all these same areas to a significantly higher degree than did either literal or ironic statements. Third, the right superior and middle temporal gyri were significantly more sensitive to ironic statements than to any others and, finally, the right inferior temporal gyrus was differentially sensitive to metaphorical meanings.

From these varied results, Eviatar and Just concluded that because all kinds of stimuli had activated the same classical language regions of the LH, the exclusive RH theory of figurative language as such was untenable. In addition, both metaphor and irony had triggered further brain activation–metaphor most conspicuously in the LH and less forcefully in one part of the RH, and irony quite vigorously in a rather disparate region of the RH. For whatever reasons, then, the metaphors were processed in a slightly dissimilar way than the literals. Perhaps most significantly, the metaphorical and ironical expressions were processed differently in relation to one another.

The authors proposed a number of possible causes for this last distinction, but seemed inclined to attribute it to the sentences’ character rather than to their category. Recall that Eviatar and Just had chosen conventional (sometimes called “salient”) metaphors. Long hackneyed, such expressions have been “lexicalized” to the point where people really don’t have to think about them in order to understand them. In this experiment, for example, a fast worker was compared to a “hurricane” and a conscientious sister was likened to an “angel from heaven.” Simple, idiomatic metaphors like these, the authors speculated, might be processed most efficiently in the LH as a unit, not unlike long words and literal phrases.

Irony, on the other hand, is always more interpretive and complex because it implicates an association between the speaker’s thoughts and the thoughts of someone else. Citing developmental studies relating to theory of mind mechanisms, the authors alluded to the fact that, while healthy children and adults who are able to correctly attribute first order beliefs (modeling what another person knows) are also able to comprehend metaphor but not necessarily irony, subjects who can make second order attributions (modeling what another person knows about what a third person knows) are usually capable of understanding irony as well. As such, Eviatar and Just prodded, the possibility that complexity rather than figurativity per se might be responsible for RH involvement invoked “an extremely interesting set of issues for future research.”

Psychologist Gwen L. Schmidt apparently concurred before she and her team of American researchers announced the results of their study, “Right Hemisphere Metaphor Processing? Characterizing the Lateralization of Semantic Processes,” in the February 2007 edition of Brain and Language. Instead of fMRI, the authors used a divided visual field technique where the reaction times of eighty-one subjects were measured after reading the final, experimentally relevant portions of sentences either in their left visual fields (to test stimulation of the RH) or in their right visual fields (to check activation of the LH).

Three different phases were designed to investigate how the brain processes various types of figurative and literal sentences. Phases one and two compared reaction times between moderately unfamiliar (or “non-salient”) metaphors (e.g., This city is a chimney) and both familiar and unfamiliar literals (e.g., The children’s shoes were covered in dirt and Janice used fans axes). Phase three compared times between familiar and highly unfamiliar metaphors (e.g., Alcohol is a crutch and A bagpipe is a newborn baby). During the first two trials, the team recorded a RH processing time advantage for moderately unfamiliar metaphor sentence endings and a LH advantage for literal-familiar sentence endings. Literal-unfamiliar sentences, like novel metaphors, produced an advantage for the RH. During the final trial, the authors found a LH advantage for familiar metaphors and a RH advantage for their highly unfamiliar complements.

In other words, Schmidt and company got exactly what they had expected consistent with the “coarse coding model” of semantic processing. Displacing the old indirect/sequential processing and dichotomous laterality paradigms, the coarse coding model predicts that any sentence depending on a close semantic relationship (e.g., The camel is a desert animal) will activate the LH, and that any sentence relying on a distant semantic relationship (e.g., either The camel is a desert taxi, or The camel is a good friend) will activate the RH, regardless of whether the expression is intended metaphorically or literally. More hackneyed stimuli can be efficiently processed in a fine semantic field in the LH. Novel ones with multiple possible meanings, however, must be dealt with more methodically in a much coarser field in the RH.

All of which makes good, practical sense from an evolutionary point of view. Adaptations are cumulative, of course, and nature builds ever so slowly and imperfectly, if at all, upon existing structures. While theoretically possible, we should never assume a priori that an isolated region of the brain would take sole responsibility for any behavior or the accomplishment of any task. Recent investigations make it clear that the human brain has evolved into a highly integrated (not to mention surprisingly plastic) organ.

But why should cognitive scientists of all people agonize over literary minutia normally regarded only in university humanities departments? Generally, because the days are long past when science could be neatly segregated from “other subjects.” More specifically, because significant clinical interests are at stake as well. Several patient populations reliably suffer from diminished or otherwise altered comprehension of irony, humor, metonymy, and non-salient metaphors in particular. Certain diseases, therefore, might well find their causes in brain anomalies also responsible for linguistic deficiencies. Regardless, such deficiencies surely exacerbate the existing social impairments experienced among patients overwhelmed by serious psycho- and neuropathologies.

One of the more unfortunate features of schizophrenic thought disturbance, for example, is the still mysterious problem of “concretism,” the inability to grasp non-literal language. In the January 2007 issue of NeuroImage, Tilo Kircher’s team published “Neural Correlates of Metaphor Processing in Schizophrenia,” an fMRI study involving twelve subacute in- and outpatients and twelve control subjects who inspected brief sentences with either a literal or novel metaphorical connotation. Kircher’s goal, of course, was to begin the process of exposing the disease’s neural bases.

As predicted, all participants’ brains activated in the left inferior frontal gyrus more forcefully for metaphors than for literals. With respect to metaphors only, controls clearly reacted more strongly than patients in the RH (more specifically, the right precuneus and right middle/superior temporal gyrus). LH results were more complicated. Healthy subjects activated most vigorously in the anterior portion of the left inferior frontal gyrus, a locus equivalent to what researchers call Brodmann’s areas 45 and 47 (which, incidentally, is just anterior-inferior to the classical Broca’s area). Remarkably, this region has been closely associated with sentence-level semantic language comprehension. By contrast, patients activated most impressively in Brodmann’s area 45, three centimeters dorsal to peak stimulation among controls.

While first acknowledging prior evidence demonstrating the RH’s valuable role in complex syntactic and semantic processing, Kircher’s team stressed their findings that the inferior frontal and superior temporal gyri “are key regions in the neuropathology of schizophrenia,” and that “[t]heir dysfunction seems to underlie the clinical symptom of concretism, reflected in the impaired understanding of non-literal, semantically complex language structures.” In other words, the patients’ shared failure to recruit now specifically identified areas in the LH appears to be at least pertinent if not vital to our struggle against this horribly debilitating illness.

In an even more recent edition of Brain and Language, a group of Italian psychologists and neurologists led by Martina Amanzio published “Metaphor Comprehension in Alzheimer’s Disease: Novelty Matters,” a study comparing both conventional and novel metaphor comprehension among twenty probable Alzheimer’s sufferers and twenty matched controls. Based in part on some of the above-referenced experiments, Amanzio successfully predicted that patients would perform relatively well with salient metaphors but significantly less so with non-salient ones.

While maintaining a healthy skepticism, the team hypothesized that the distinction might this time involve the prefrontal cortex, the brain’s executive center, because prefrontal dysfunction is a common symptom of Alzheimer’s disease and because the comprehension of non-salient metaphors requires the executive ability to compare and combine vehicles and topics in order to appreciate figurative meanings. “These findings,” the Italians concluded, “may have some clinical implications for the real life communication with [Alzheimer’s] patients. Salience matters.”

And, thus, so does metaphor. Figurative language is surely more than an intellectual extravagance. It is as much a fiber of our very being as each of the countless neurons contained in our big, beautiful brains. Most fortunately, however, comprehension of novel expression serves as a useful barometer of our personal and communal health as well. So one might permit a writer the guilty pleasure of mixing his metaphors on occasion, despite academic decorum.