"Why we read fiction" by Lisa Zunshine (Ohio State UP, 2012)

One reason that people read novels is that they enjoy a Theory-of-Mind workout, a chance to create motives and characters from observables. Novelists exploit this. Even if readers get the interpretations wrong, it's exercise, though a good piece of literature usually rewards correct interpretations. Sometimes readers are invited to work out several levels of meaning - "X thought that Y said to Z that ...". Dunbar et al in experiments found that readers had 5-10% error rates up to 4 levels, and nearly 60% on 5th level. Readers commonly short-cut some of these levels - "X doesn't like Z". Novels like "Clarissa" offer readers lot of opportunities to practice.

She then considers metarepresentation. When a piece of information ("X doesn't like Z", say) is represented, it's associated with meta-data (e.g. "K thinks that"). Sometimes this meta-data (aka tag) is implicit. It's cognitively hard work to keep in mind all the statements and related tags. She points out that in schitzophrenia, the tags are often lost - "I wish I was beautiful" and "Daddy said I was beautiful" become "I am beautiful". She writes that Our capacity for "monitoring and reestablishing the boundaries within which each representation remains useful" thus underlies crucially our practices of literary interpretation. She considers two particularly challenging scenarios -

• Unreliable narrators - "Lolita", where the tag "I think that" is elided by the narrator, and the reader forgets that when the narrator says that many other men were attracted to Lolita, we're seeing things through his eyes. In such pieces the readers' initial trust is broken.
• Detective novels - Readers initially distrust everyone. Sometimes (to increase the final surprise) the reader's not told what's happening in the detective's head. If in a novel the reader's usually told, then the exceptions are significant. She points out that "The narrative economics of the short story, which necessarily limit the number of minds that could be read and misread, makes it convenient to posit the detective's mind as one of the "mystery" minds, along with that of the main suspect" and thinks that Auden's "In his sexual life, the detective must be either celibate or happily married" observation might be because the complexities of relationships (at the same time as source-monitoring to solve the crime) will overload the reader.

I found the book useful.

Other reviews

• Goodreads
• Merja Polvinen (Broadly speaking, whereas our Theory of Mind makes it possible for us to invest literary characters with a potential for a broad array of thoughts, desires, intentions, and feelings and then to look for textual cues that allow us to figure out their states of mind and thus predict their behaviour, our metarepresentational ability allows us to discriminate among the streams of information coming at us via all this mind-reading. It allows us to assign differently weighed truth-values to representations originating from different sources)