University of Wisconsin: Bookworms can “read” people, too

More than any other genre, fiction is the realm of emotion. “Getting lost in a story” means entering a world we don’t want to leave, where we are fully absorbed not only in the actions of the characters, but also in their thoughts, feelings and motivations. Throughout the experience, readers pass through their own emotion states, triggered by the words and phrases.

Now it appears that this rich, often fraught, journey of the imagination—so often considered a solitary pleasure—is good training for reading the emotions of people in real life.

A new study by a team of psychology researchers at UW-Madison provides important new insight into a likely causal link between reading fiction and emotion recognition, combining behavioral experiments with methods from the digital humanities to show that exploring the mental states of fictional characters helps people recognize emotion expression in other human beings.

“For years, researchers had this sense that fiction readers were getting practice simulating other people’s minds,” explains Steven Schwering, a PhD student in psychology and the lead author on the study. “Our experiments help explain why people’s experience with fiction affects their emotion recognition capabilities.”

The project started several years ago, when Natalie M. Ghaffari-Nikou, then an undergraduate student in psychology exploring a senior thesis, began following the thread of an idea she’d encountered through conversations with psychology professors Maryellen MacDonald and Paula Niedenthal: Did literacy rates affect emotion recognition capabilities? MacDonald and Niedenthal helped her develop an experiment to test whether UW students who scored higher on a fiction-reading assessment would be better able to recognize not only simple emotions (anger, joy, surprise, sadness, fear) but also complex emotions (despair, relief, anxiety, irritation, pride).

Using the Author Recognition Task (ART), a common assessment of fiction reading that MacDonald helped develop in 2008, along with the Geneva Emotion Recognition Test (which features videos of actors expressing emotions), Ghaffari-Nikou’s experiment revealed that participants with higher ART scores were significantly better able to recognize complex emotions.


The findings were exciting – but what role, exactly, did literature play?

MacDonald, who leads UW’s Language & Cognitive Neuroscience Lab and studies how we comprehend and produce language, says fiction is rich with descriptions of emotion.

“Descriptions of characters’ behaviors and emotions may provide “a virtual reality of sorts” that readers can bring to their real-life experiences,” MacDonald says. “It seems almost counterintuitive that an introverted activity like reading can actually help people learn to better recognize emotion in others.”

MacDonald said previous studies had showed that fiction reading was correlated with the ability to recognize emotion – but “there was never a story as to why that might be.”

To determine what it is about fiction that affects people’s ability to recognize emotion, the team turned to the methods of the digital humanities, in which software searches an enormous body of work—all of Shakespeare’s plays, say, or the entire corpus of Victorian literature—to uncover patterns or connections. In this case, the team decided to trawl the 600 million words found in the Corpus of Contemporary English (COCA). They searched for “emotion words,” used in an emotive sense, across three genres: “Fiction,” “Spoken” and “Other.” “Fiction” included short stories and plays from magazines for both children and adults (such as Scholastic Scope and The New Yorker) and first chapters of fiction books. The “Spoken” category included transcriptions from popular television and radio programs (from The Jerry Springer Show to All Things Considered). “Other” included newspapers, academic journals and popular magazines.

A total of 20,172 emotion category labels and their contexts were analyzed. Schwering, who worked with Niedenthal as he wrote the software program, says the analysis revealed that simple emotions were used equally often across all of the corpora. But it was in the “Fiction” category where complex emotions, such as disgust or joy, were found much more frequently.

“With these results, we felt like we could argue that it is people’s experience with fiction texts, rather than with other kinds of texts, that affects their emotion recognition capabilities,” says Schwering.

In addition to testing UW students—who likely come into the university having read many of the same texts— the team wanted to test one more group, to make sure their results would replicate.

In the second study group, drawn from people across the country, from a wide variety of backgrounds and age groups, those who scored higher on the ART (meaning they read more fiction) were not necessarily better at recognizing complex emotions. However, those fiction readers were better at recognizing all kind of emotions.

“So, we didn’t have the neat little correlation between simple and complex, but we still got the overall effect,” explains Schwering.

Combining a “big data” approach to analyzing fiction and other kinds of language with psychological experiments is a new approach, says MacDonald. The analyses helped point out an answer to the seemingly contradictory question of why reading—a completely non-social activity—results in better social cognition.

Schwering says the results are exciting, tying our language comprehension and production systems with our social experiences, which “are two quite different domains.”

“We really want to know how language affects the ways we interact with the world. How does language shape our social lives? How do we recognize our emotions in others?” says Schwering.