Brain and Language on the Fly

The Neuroscience of Linguistic Improvisation

by Elizabeth Beam


It’s like when I’m on the mic I can squish a 

Sucka like a vice grip, my pen put ya

In the slaughterhouse cause your style’s been butchered

I’ll spin chainsaw, take off like the blades on, my brain’s on

Hyperdrive, someone put the brakes on


While maintaining a beat and a storyline, Eminem rapped the above lyrics as part of a high-energy, fast-paced, yet off-the-cuff performance that lasted over eight minutes on live radio (1). Though the lines printed here are not accompanied by Eminem’s vocals, you can hear as you read them how the sounds of the words flow through the slant rhymes “squish a,” “put ya,” and “butchered.” Meanwhile, the meaning of the words mounts through the extended metaphor, comparing Eminem’s defeat of an imagined opponent to a gruesome farmhouse slaughter. As remarkable as it is, this example is just one among the routinely astonishing feats of language that freestyle rappers can accomplish off the tops of their heads.

Aside from the exceptional talent that some individuals have for freestyle, we are all capable of playing with language, and we do not need to step up to a microphone to prompt us to do so. Sociolinguist Joan Schwann observed this firsthand one afternoon while eavesdropping on a family in the park (2). As the speakers tossed scraps of food to the birds, they commented on one pigeon that was aggressively chasing the others away:

A:  He might look scruffy but he’s seen off that one over there

B:  Obviously a thug amongst pigeons

C:  Al Capigeon

D:  The godfeather

[Laughter overlaps C & D]

This brief exchange exemplifies the sort of linguistic ingenuity that speakers display in everyday conversation. After speaker B anthropomorphizes the pigeon as a “thug,” speakers C and D riff off of the idea with references to The Godfather. The puns “Al Capigeon” and “godfeather” draw attention to the phonological similarity between “pigeon” and “Capone,” between “feather” and “father.” These jokes are not just silly, but impressively clever, eliciting laughter from the other participants before C and D have finished speaking. It is also important to note that these speakers had not been holding “Al Capigeon” and “godfeather” in arsenals of puns to be dispensed at just the right moment. Rather, the puns were invented on the spot, uttered as soon as they came to mind.

When we crack jokes and craft sonically pleasing rhythms on the fly, we engage basic cognitive systems to do something extraordinarily complex. The cognitive systems involved range from language to motivation to memory to emotion to motor control—systems that are well studied on their own but rarely altogether. The trouble with breaking down verbal creativity into simpler systems is that the superordinate behavior cannot be reconstructed easily from its parts. To do so would be like trying to solve a jigsaw puzzle without looking at the picture on the box; child’s play is turned into a formidable challenge because the design on any single piece offers very little information about the image that emerges when all the pieces are put together. Furthermore, the brain is not a picture that can be laid flat on a living room table but a three-dimensional and dynamic structure comprised of upwards of 100 billion neurons that modulate one another over time.

So, why haven’t cognitive neuroscientists pushed the pieces aside and studied verbal creativity directly? It is critical to note that normal behaviors occur in settings very different from the environment inside a neuroimaging scanner. Understandably, the above exchange between picnickers could not have transpired if the fauna of the park were exchanged for the white plastic bore of an fMRI machine. Moreover, studies conducted in the scanner are not leisurely afternoons in the park; each session is comprised of timed blocks that must be short and consistent across subjects. Because neuroscientists cannot eavesdrop on the brain like a sociolinguist in the park, they have instead investigated verbal creativity through simplified assessments like anagram puzzles (3). However, while solving anagrams and advancing an interesting conversation may both rely on the generation of insightful ideas expressed through the elements of language, one would expect the obvious differences on the behavioral level between speaking and re-arranging letters to manifest themselves in the brain.

It is here that freestyle rap meets neuroscience. Whereas the tasks designed to study verbal creativity are disconnected from real behaviors, and the real behaviors of normal individuals are compromised in the scanning environment, freestyle rappers can defy these constraints because they are specially trained to do so. For freestyle rappers, spitting a few improvised lines during a timed block in the scanner is comparable to rapping live on stage within the rules of a competition. To take advantage of this, Siyuan Liu and colleagues recently assembled a cohort of 12 freestyle rappers for a neuroimaging study (4). The neuroscientists asked the rappers in some segments of the scan to deliver improvised raps on the spot and in others to recite raps they had memorized.

By daring to study verbal creativity as it occurs in the   wild, Liu et al. gained access to the brain state in which live, inventive speech unfolds. They discovered that the network of systems engaged by freestyle rap is specialized, relying more heavily on drive and memory selection than the network for rehearsed rap. The nature of control also differs in freestyle, shifting from top-down self control to faster and more fluid control by motor regions. These findings go beyond mapping anatomical correlates of verbal creativity to the brain, lending a bird’s-eye perspective to the complex and dynamic brain state that emerges when multiple networks converge in real time. Furthermore, expanding our scope from a dozen freestyle rappers to the billions of speakers conversing constantly across the globe, this study may guide us toward a more general understanding of the mind as it engages in everyday verbal creativity.



Taking Flight: Verbal Drive and Creativity

The brains of freestyle rappers must be able to shift from cruising levels of conversational speech to the high gear of a rapid and rhythmic verbal performance. Liu et al. found that, in their cohort of rappers, the medial prefrontal cortex (mPFC) was the key to this creative ignition (5). The mPFC was especially active in improvised conditions, compared to conditions in which participants recited memorized lyrics, suggesting that it plays a role in the generation of original, on-the-spot rap. The mPFC is also more active at the beginning than at the end of a segment, consistent with the idea that it helps get the rap going.

The results of this study align well with the narrative that other neuroscientists are crafting of creative drive and the brain. Like Liu et al., the neuroscientific community has flagged the mPFC as a noteworthy region for motivation. In task-based studies, mPFC activity has been found to increase as the payoffs for good performance are raised (5). The mPFC is also preferentially activated when viewing scenes that are later described with high enthusiasm, suggesting that this region may play a role in developing the urge to communicate (6). To tie these relations between drive, language, and the mPFC to artistic creativity, neuroscientists could adapt their neuroimaging tools to studying the link between mental illness and creativity, as described in accounts of numerous eminent writers. Whereas bipolar writers like Robert Lowell and Lord Byron suffered from creative block during their depressions, they were most prolific when they were manic (7). Perhaps these mood disturbances were accompanied by the changes in mPFC connectivity that have been identified in individuals with depression and bipolar disorder (8).

While these correlations are promising, they offer more questions than they answer. If it is true that the mPFC controls verbal creative drive, by what mechanism does it do so? Furthermore, thinking beyond the brain, what would this understanding of a neurobiological mechanism mean for rappers and for everyday speakers? To step towards answers to these questions and past the standard neuroimaging paradigm of mapping a behavior to a brain region, Liu et al. sought to understand how the level of mPFC activity varies with the quality of a freestyle performance. They judged the improvised raps by factors like variation in content, use of fresh language, and coherence of the rhyme scheme. Apparently, the better the rap, the greater the mPFC activation—raising the possibility that, through the mPFC, there is a relationship between creative drive and skill.

To make sense of this, like many peculiar observations in biology and human nature, it helps to turn to the principles of Charles Darwin. According to the Darwinian model for creativity, the number of ideas that are novel and useful increases proportionally with the total number of ideas (9). The model predicts that a rapper spitting rhymes is more likely to succeed as long as he or she keeps at it. The popular journalist Malcolm Gladwell has rendered this relationship formulaic, claiming that 10,000 hours is the time it takes for a person to gain mastery in a skill (10). This runs counter to the once well-accepted belief that artists are born with special talents that elude the rest of us. While freestyle rappers do seem to have an unusual skill for rapping quickly and easily in the spur of the moment, there is hope for any one of us that, with enough time and effort, we could write the next chart-topping hit.

Thinking Backwards and Forwards:  Autocueing and Memory

In isolation, the system for driving speech would be like an automobile engine without the rest of the car. There must be another system supplying the ideas that form the content of what we say. These ideas come from our memories—from the facts and stories we read in books, from the words we learn to define and say aloud, and from our day-to-day experiences living in the world. Furthermore, while a detailed and well-organized memory is required for speech, our conversations and Eminem’s raps would be rather dull if the brain could do no more than retrieve random memories exactly as they were encoded. For performances of verbal creativity, our brains must also be able to select memories and to recombine them in new and compelling ways.

Cognitive neuroscientist Merlin Donald posits that, before there was language, there was an expansion in human memory (11). More crucial than the size of memory stores, however, was the development of the ability to tap into them voluntarily. This “self-initiated access to memory,” or briefly “autocueing,” allowed us not only to retrieve items relevant to a given set of circumstances but more amazingly to retrieve irrelevant items at will. Eventually, autocueing enabled humans to invent, recall, and actively string together symbols into words, sentences, and stories. Rather than an immutable record of history, human memory is flexible, allowing for the dynamic reorganization of items from the past into novel constructs that are useful in the present and the future.

The sequence of events that occurred during the early evolution of human memory is now recapitulated in the way that humans access memory during speech. The inferior frontal gyrus (IFG) guides the retrieval of semantic knowledge from memory in a top-down manner, allowing for target memories to be selectively recalled and articulated (12). A connectivity analysis by Liu et al. (2012) reveals that, during freestyle rap, there is a strong positive correlation between activation of the IFG and the mPFC. This could mean that, when the creative drive thought to be mediated by the mPFC is increasing, so too is the pull on semantic memory stores by the IFG. Although these correlative analyses cannot establish causation, this relationship is consistent with the possibility that Donald’s autocueing system is enhanced in freestyle rap.

Working memory, often referred to as the mental “sketchpad,” is where the mind can scribble, scratch out, and rewrite the thoughts in its consciousness. One might expect working memory to be engaged during freestyle performances, allowing rappers to play with ideas and actively organize them into rhyming, rhythmic words. Curiously, Liu et al. (2012) observe the opposite trend. The dorsolateral prefrontal cortex (dlPFC), previously shown to support working memory during creative endeavors, is actually deactivated during freestyle rap (13,14). Moreover, dlPFC activity is inversely correlated with activation of the critically involved mPFC.

A reasonable explanation for this counterintuitive phenomenon is timing. Freestyle rappers speak so quickly that they do not have time to consciously evaluate and revise the content of their utterances before articulating them. As soon as items are retrieved from memory, they are incorporated into the verbal output stream. The ability to guide the search for associations between memories at high speed may be a distinguishing quality of rappers. Future studies should test this possibility by comparing the performance of freestyle rappers and normal speakers on the same memory and verbal creativity tasks. It is also possible that, during freestyle, the memory system interacts in a special way with yet another system—a system that streamlines the motor output.

The Flow:  Cognitive Control and Motor Supervision

The paradox of freestyle rap is that, as demanding as it is on the brain to produce fast-paced utterances that are not just semantically coherent but poetically and rhythmically structured, the performance can feel just as effortless to the rapper. As previously noted, the brain region involved in cognitive control and working memory is selectively deactivated during freestyle rap. Merlin Donald would contend that the apparent lack of conscious control that rappers show during their freestyle performances is the expected result after extensive practice within the genre. This “automatization” is simply “the end result of a process of repeated sessions of rehearsal and evaluation” (15). The process of automatization is not at odds with consciousness at all but rather one of its natural consequences.

To be sure, rappers do wield some form of control during performances. While Eminem unabashedly delves into the vulgar, the violent, and the bizarre in his freestyles, he rarely misses a beat. In place of conscious cognitive control, it is motor output monitoring that keeps the verbal flow steady during freestyle rap. Merlin Donald’s insights into motor control are perhaps his most notable. “My key proposal,” Donald writes in his treatise on human nature, “is that the first breakthrough in our cognitive evolution was a radical improvement in voluntary motor control that provided a new means of representing reality” (9). This representational system conveys memories through symbolic body movements. In prehistoric times, it may have yielded a form of culture through mimesis that preceded the invention of language.

The neural system that first enabled people to control body movements has been refined since the onset of mimetic culture. Modern humans are able to monitor and adjust articulatory movements during speech, thanks in part to the cingulate motor cortex (16). In a functional connectivity analysis, Liu et al. demonstrate that there is a strong positive correlation between medial prefrontal and cingulate motor activation during improvised rap. Furthermore, the authors postulate based on anatomical studies in monkeys that there is a direct functional connection between medial prefrontal and cingulate regions, and other studies in humans have shown that portions of the mPFC are continuous with the cingulate (17, 18). As information about intentions and motor plans is transmitted along this alternative route, the drive to speak associated with the mPFC bypasses the self-conscious dlPFC.

By incorporating practiced behaviors into automatic processing in analogous ways, humans can streamline processes they know well and build on top of those processes in an hierarchical fashion, creating new processes that are ever more complex. Consider language: a child first learns to associate words with objects and other referents, then works painstakingly to combine words into grammatically correct sentences, and finally is able to produce language relatively automatically, concentrating less intensely on the forms and meanings of the words and beginning instead to focus on other functions of the speech act (19). With enough practice, the child might someday rap as easily as he or she carries on a conversation.

Let It Free, Let It Fly

Having toured the systems that intersect in the network for verbal improvisation, it is at last possible to integrate the neuroscientific findings into a complete picture of the freestyle brain at work. Regions like the mPFC, IFG, and cingulate motor area that show increased activity are thought to give rise to the distinct cognitive characteristics of verbal improvisation, including enhanced motivational drive, memory access, and motor monitoring. At the same time, working memory and self-conscious control are diminished. The attributes of freestyle rap are synthesized by the brain into what has been called a “flow state” during which the performer is so intensely engaged that the words feel as if they are pouring forth involuntarily (20). It is as if the brain takes over when the rapper steps up to the microphone, producing a stream of language that is spontaneous yet poetically crafted, guttural yet rhythmic.

Considering how well-suited the brain is for freestyle, one might be tempted to conclude that it evolved for the express purpose of enabling humans to rap. However, Donald notes that, when evolution appears to proceed in favor of humans, it bears no real concern for them. “Our major cultural achievements,” Donald remarks, “have evidently been the delayed by-products of biological adaptations for something else” (9). If freestyle rap can be regarded as one of these hallmarks of the human behavioral repertoire, then what is the “something else” from which it is derived? Taking away the beat, the pace, and the dominance of the poetic function, freestyle rap bears a striking resemblance to the linguistic improvisation that characterizes our everyday conversational play. The jokes and puns we all make are the basis for the remarkable extemporaneous speech that distinguishes freestyle rap. Every time we speak, we heed the words of musician Tunde Adebimbe when he sings, “Let it follow that we let it free, let it fly” (1).

Elizabeth Beam is a research assistant in a neuroscience lab at Harvard University.


  1. Eminem (2009, May 23). The Tim Westwood Show. BBC Radio 1Xtra. London, UK: British Broadcasting Company.
  2. Maybin, J., & Swann, J. (2007). Everyday creativity in language: Textuality, contextuality, and critique. Applied Linguistics, 28(4), 497-517.
  3. Fink, A., Benedek, M., Grabner, R. H., Staudt, B., & Neubauer, A. C. (2007). Creativity meets neuroscience: Experimental tasks for the neuroscience of creative thinking. Methods, 42, 68-76.
  4. Liu, S., Chow, H. M., Xu, Y., Erkkinen, M. G., Swett, K. E., Eagle, M. W., Rizik-Baer, D. A., & Braun, A. R. (2012). Neural correlates of lyrical improvisation: An fMRI study of freestyle rap. Scientific Reports, 2(834), 1-8.
  5. Kouneiher, F., Charron, S., & Koechlin, E. (2009). Motivation and cognitive control in the human prefrontal cortex. Nature Neuroscience, 12, 939-945.
  6. Falk, E.B., O’Donnell, M.B., & Lieberman, M.D. (2012). Getting the word out: Neural correlates of enthusiastic message propagation. Frontiers in Human Neuroscience, 6(313), 1-14.
  7. Jamison, K. R. (1993). Touched with fire: Manic-depressive illness and the artistic temperament. New York, NY: The Free Press.
  8. Price, J. L., & Drevets, W. C. (2012). Neural circuits underlying the pathophysiology of mood disorders. Trends in Cognitive Sciences, 16(1), 61-71.
  9. Simonton, D. K. (1999). Origins of genius: Darwinian perspectives on creativity. London: Oxford University Press.
  10. Gladwell, M. (2008). Outliers: The story of success. New York, NY: Little, Brown, & Company.
  11. Donald, M. W. (2004). The definition of human nature. In Rees, D., & Rose, S. (Eds.), The new brain sciences: Perils and prospects (p. 34-53). Cambridge, UK: Cambridge University Press.
  12. Badre, D., Poldrack, R. A., Pare-Blagoev, E. J., Insler, R. Z., Wagner, A. D. (2005). Dissociable controlled retrieval and generalized selection mechanisms in ventrolateral prefrontal cortex. Neuron, 47(6), 907-918.
  13. Dietrich, A. (2004) The cognitive neuroscience of creativity. Psychonomic Bulletin & Review, 11(6), 1011-1026.
  14. Vandervert, L. R., Schimpf, P. H., & Liu, H. (2007). How working memory and the cerebellum collaborate to produce creativity and innovation. Creativity Research Journal, 19(1), 1-18.
  15. Donald, M. W. (2001). A mind so rare: The evolution of human consciousness. New York, NY: Norton.
  16. Picard, N. & Strick, P. L. (1996). Motor areas of the medial wall: a review of their location and functional activation. Cerebral Cortex, 6, 342-353.
  17. Petrides, M. & Pandya, D. N. (2007). Efferent association pathways from the rostral prefrontal cortex in the macaque monkey. Journal of Neuroscience, 27, 11573-11586.
  18. Öngür, D., Ferry, A. T., & Price, J. L. (2003). Architectonic subdivision of the human orbital and medial prefrontal cortex. Journal of Comparative Neurology, 460, 425-449.
  19. Jakobson, R. (1956). Metalanguage as a linguistic problem. In Rudy, S., & Waugh, L. R. (Ed.), Selected writings VII: Contributions to comparative mythology (p. 113-121). Berlin: Mouton.
  20. Csikszentmihalyi, M. (1996). Creativity: flow and the psychology of discovery and invention. New York, NY: Harper Collins.
  21. Adebimbe, T. (2006). Province. On Return to Cookie Mountain. Santa Monica, CA: Interscope Records.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: