Research Bulletin: Scene in the Imagination

How does a writer enable a reader to imagine a scene? A paper by Jennifer Summerfield, Demis Hassabis, and Eleanor Maguire (2010) has offered a step towards answering this. Recent studies of brain imaging have identified what researchers call a core network of brain areas that are involved in: memories of events from our lives, imaginings of future possibilities, and navigating in the world (click here). Researchers study these processes to understand how we negotiate the day-to-way world, but it seems likely that they are also involved in sustaining the simulation that a reader constructs when reading a novel or short story.

Summerfield et al. have—as they put it—slowed down the constructive process of imagination by presenting to participants in an fMRI machine between three and six spoken phrases, each of which is an element of a naturalistic scene. Here is an example of three phrases (elements) they used: "a dark blue carpet" … "a carved chest of drawers … "an orange striped pencil." As they heard each succession of phrases, participants were asked to imagine a scene containing the elements, and later to report on the vividness of the result. At the same time, the researchers were looking to see which brain areas were activated as each element was presented. They identified three brain areas of the core network that were important: the hippocampus (plus some associated areas), the intra-parietal sulcus (plus another area), and the lateral prefrontal cortex.

When the first element was presented, the hippocampus and associated areas were most activated but, with the presentation of the second element the activity of these areas decreased. With presentation of the third element activity in these areas increased again. A different pattern was seen with the inter-parietal sulcus: with the presentation of each additional element from one to three, the activity of this area and an associated area increased in a stepwise fashion. The lateral prefrontal cortex showed yet another pattern. It became activated only with the presentation of the second element, and its activity was sustained with the presentation of subsequent elements. For all three of the core brain areas, presenting phrases (elements) beyond the first three did not produce additional changes. The participants confirmed that three elements were sufficient to construct a scene to its maximum vividness.

The idea of this study bears a striking parallel to the ideas of Elaine Scarry (1999) who has proposed that, to enable a reader to imagine a scene, a writer must offer what she calls instructions. She says Thomas Hardy and Marcel Proust are particularly good at this, and she gives some helpful ideas about the kinds of instructions that work well for constructing scenes (see micro review in our Archive: Books on the Psychology of Fiction).

I find it interesting that Summerfield and her colleagues used descriptive phrases rather than just words as their elements. This reminds me of Gabriel Garcia Marquez saying in an interview that in fiction it's no good writing "There was a butterfly." One must write: "There was a yellow butterfly."

I don't know whether anyone has counted the number of phrases different writers use to prompt their readers mentally to construct a good scene. I happen to be reading and re-reading my way through Anton Chekhov's short stories. As I was writing this post, it  occurred to me to count the elements in a scene that was sufficiently vivid for me to remember from a story that I had read an hour previously. The story is "Gusev" in Richard Pevear's and Larissa Volokhonsky's translation of Chekhov's stories. It is about a discharged private solider, Gusev, who has been in the East for five years. He is on a ship supposedly on his way home. Although neither Gusev nor we the readers know it at the beginning of the story, he is dying. He imagines a scene from his homeland:

He pictures an enormous pond covered with snow … On one side of the pond, a porcelain factory, the color of brick, with a tall smokestack and clouds of black smoke: on the other side a village… Out of a yard, the fifth from the end, drives a sleigh with his brother Alexei in it (ellipsis points in the original, p. 110).

Is this three elements? It depends, I suppose, on how you count. I count the pond, the factory, the village with a sleigh. In the story, the ex-soldier starts to elaborate the scene in his memory. His brother is tipsy, and in the sleigh are his brother's children. Each element is more elaborate than any used by Summerfield and her colleagues. And—as is usual in a story—in the scene, actions begin to occur; the sleigh driving out is one such, and the next is one of the children in the sleigh laughing while the other's face can't be seen because it is wrapped up. Gusev thinks the children might be getting chilled. Perhaps the fMRI researchers are right.  Perhaps three elements are enough to get a good scene going. But perhaps the writers are right, too: the next thing for research might be that, after establishing parts of a static scene, they might consider an experiment in which actions occur. Elaine Scarry also has good things to say about giving readers instructions for imagining actions.

Anton Chekhov (1883-1903). Anton Chekhov: Stories (Richard Pevear & Larissa Volokhonsky, Trans.). New York: Bantam (current edition 2000).

Elaine Scarry (1999). Dreaming by the book. Princeton, NJ: Princeton University Press.

Jennifer Summerfield, Demis Hassabis & Eleanor Maguire (2010). Differential engagement of brain regions within a ‘core’ network during scene construction. Neuropsychologia, 48, 1501-1509.

Image: From Summerfield et al.'s paper, showing the activation of the hippocampus as greatest with the first element (1E) and third element (3E) in the presentation of a series of descriptive phrases.