Don't Look Now

I’ve just read on LiveScience.com that “[b]ooks are just as powerful as movies” at triggering “delight, pain, or disgust” reactions in the brain. As is so often the case on LiveScience, this gripping opening represents a kind of yellow science journalism that has intensified its hold on the popular imagination in the last few years.

For one thing, in the single study writer Andrea Thompson discusses, a team of Dutch scientists (Jabba et al.) set out to test the triggering of disgust by various stimuli: print, motion picture, foul-tasting beverage. The study concerns disgust exclusively; there’s no hint of delight or pain in it, or for the rest of the article.

Moreover, contrary to what’s suggested by the article’s title—”Books Still Rival Movies For Stirring Emotions”—the scientists themselves make no quantitative comparison between the power of books and movies. That is, nobody’s saying “just as powerful” besides Thompson, not in their interviews or in the study. The parts of the study discussed on LiveScience.com only show that similar responses can be generated from various stimuli.

Finally, those portions of the study absent from the article in fact show deep differences among the responses generated by the stimuli. Let’s take a closer look at its methods to see why and how these differences emerge. At stake is not only a better understanding of the study, but the success or failure of the LiveScience article as a piece of science journalism.

First, participants in the study were shown three-second films of a disgust reaction: A person on the screen takes a drink and then appears disgusted. There were also control films showing pleasurable and neutral responses. As they watched, an fMRI machine captured changes in blood flow to various regions of their brain. During the disgust clips, these included an increase in activity in the region that injury and other studies suggest is a crucial part of the disgust reaction. (Jabba et al., 2)
One of the team members describes the next step in Thompson’s article:

“Later on, we asked them to read and imagine short emotional scenarios. […] For instance, walking along a street, bumping into a reeking, drunken man, who then starts to retch, and realizing that some of his vomit had ended up in your own mouth.”

In both cases, as well as a third in which the subjects tasted something, well, disgusting, the anterior insula and adjacent frontal operculum lit up the fMRI. (Jabba et al., 3-4) So, we conclude, books are as effective at provoking disgust as movies, right?

Not a chance. There are crucial differences between a three-second clip of a man taking a drink and then experiencing disgust and the process of reading and imagining a scenario like the one described above, and similar differences between participants’ reactions to the two stimuli.

First, the clips and the written scenarios differ in duration. The scripts given to the participants took much longer than three seconds to read and process; the study gives the reading times as 35 seconds (2). That the scripts required so much more time than the clips to induce the desired reaction belies any suggestion that the former are “just as powerful” as the latter.

Second, the clips and the scripts differ in terms of the depth of the empathetic reaction they allow. Reading a disgusting scenario and imagining oneself in that scenario seems much more likely to trigger deep emotional reaction than just seeing a face on-screen for three seconds (mirror neurons be damned). The study confirms this differing depth; the greatest average change to the fMRI signal during clip viewing was about 0.1%, compared to about .66% for the script-reading portions of the experiment (figs. 1b and 1c, page 3).

And of course, these clips don’t replicate the experience of watching a “movie.” Typically, narrative and generic context as well as cinematic style and technique allow or assist viewers in developing empathic responses to persons on-screen. All of that is absent or extremely limited in a laboratory setting, and in the films themselves. (You can find stills of the clips in another study, requiring a paid subscription or institutional access.)

Are these problems with the study? Or just with the reporting? The study’s abstract ends:

[T]his shared region however [sic] was found to be embedded in distinct functional circuits during the three modalities, suggesting why observing, imagining and experiencing an emotion feels so different. (1)

In other words, the study acknowledges the apples-to-oranges nature of the comparisons between the different stimuli-response pairs, even as it shows their similarities. So what do we make of LiveScience’s expansion of the conversation from the realm of clips and scripts to that of “movies” and “books”?

It seems to me that one of the biggest problems with science reporting in its current state—and not just on the web—is the indiscriminate use of the facts in the service of stories with artificially inflated “wow” factors. There’s a diminished ethical standard for much science journalism, in which obfuscation, equivocation, and fallacious conclusion or conjecture are all widely accepted, or at least routinely tolerated. By looking closely at one article and the study it concerns, I hope to have shown some of the problems to which this climate can lead.