Thursday, April 05, 2018

The Experience of Reading: Empirical Evidence

What do you experience while reading? Do you experience inner speech, as though you or the author are saying the words aloud? Do you experience visual imagery? Do you experience the black marks on the white page? All of these at once? Different ones at different times, depending on how you're engaging with the text?

Although educators, cognitive psychologists, and literary critics often make claims about readers' typical experience, few researchers have bothered to ask readers, in any systematic way, these basic questions about their experience. [Note 1] So Alan Tonnies Moore and I decided to try doing that. Alan's work on this topic became his 2016 dissertation, and we have now have a paper forthcoming in Consciousness and Cognition (final submitted manuscript available here).

In each of three experiments, we presented readers with several hundred words of text. In two of these experiments, a beep interrupted participants' reading. Immediately after the beep, readers were to report what was in their experience in the final split second before the beep. We collected both general free-response descriptions of their experience and yes/no/maybe reports about whether they were experiencing visual imagery, inner speech, and visual experience of the words on the page (all phrases defined beforehand). In all three experiments, we also collected readers' retrospective assessments of how frequently they experienced visual imagery, inner speech, and the words on the page while reading the passage we had presented.

At the end of each experiment, participants answered several questions about the text they had just finished reading. Some questions we thought might relate to visual imagery (such as memory for visual detail), other questions we thought might relate to inner speech (such as memory for rhyme), and still other questions we thought might relate to visual experience of words on the page (such as memory of the font). We were curious whether performance on those questions would correlate with reported experience. Do visual imagers, for example, remember more visual detail?

Here are the main things we found:

(1.) People differ immensely in what types of experiences they report while reading. Some people report visual imagery all the time; others report it rarely or never; and still others (the majority) report visual imagery fairly often but not all of the time. Similarly for inner speech and words on the page.

To see this, here are a couple of histograms [click to enlarge and clarify].

Readers' retrospective reports in Experiment 2 (Experiments 1 and 3 are similar):

Readers' yes/no/maybe reports immediately after the beep, also in Experiment 2:

(2.) Inner speech is less commonly reported than many researchers suppose. This has also been emphasized in Russ Hurlburt's related work on the topic. Although some researchers claim or implicitly assume that inner speech is normally present while reading, we found it in a little more than half of the samples (see the histograms above). Visual imagery was more commonly reported than inner speech.

(3.) Reported experience varies with passage type, but not by a lot. In Experiment 2, we presented readers with richly visually descriptive prose passages, rhyming poetry, and dramatic dialogue, thinking that readers might experience these types of passages differently. Differences were in the predicted directions, but weren't large. For example, visual imagery was reported in 78% of the beeped moments during richly descriptive prose passages vs 66% of the poetry passages and 69% of the dramatic dialogue (chi-square = 14.4, p = .006). Inner speech was reported in 65% of the beeped moments during dramatic dialogue passages vs 59% of the poetry passages and 53% of the descriptive prose (chi-square = 19.1, p = .001).

(4.) There was little or no relationship between reported experience and seemingly related comprehension or skill tasks. For example, people who reported seeing the words of text on the page were not detectably more likely to remember the font used. People who reported visual imagery were not detectably more likely to remember the color of objects described in passages. People who reported inner speech were not more likely to disambiguate difficult-to-pronounce words by reference to the rhyme scheme. Although all of this is possibly disappointing, it fits with some of my previous work on the poor relationship between self-reported experience and performance at behavioral tasks.

Alan and I believe that sampling studies will soon become an important tool in empirical aesthetics, and we hope that this study helps to lay some of the groundwork for that.

Full version of our paper available here.


Note 1: One important exception to this generalization is Russ Hurlburt in his 2016 book with Marco Caracciolo and in his paper with several collaborators forthcoming in Journal of Consciousness Studies.)


Related Posts:

What Do You Think About While Watching The Nutcracker? (Dec 17, 2007)

The Experience of Reading (Nov 25, 2009).

The Experience of Reading: Imagery, Inner Speech, and Seeing the Words on the Page (Aug 28, 2013).

Waves of Mind Wandering During Live Performances (Jan 15, 2014)


Anonymous said...

thanks for this, so many taken for granted (never questioned) concepts/intuitions in psych/philo good to start pulling them apart and seeing what's at work.

Adam said...

Probably worth checking out the literature on how some people can't form mental images *at all* and don't know that everyone else can - they think everyone is speaking metaphorically. I think Francis galton was the first to notice this.

Eric Schwitzgebel said...

Adam -- Yes, there's an interesting literature on that issue!

Darren Reynolds said...

This visual imagery that some people experience whilst reading: can anyone give an example? What sort of thing are we talking about? What would someone with visual imagery get, for example, with the text right here: these questions?

I ask because I'm generally a highly visual person, sometimes experiencing quite complex scenes just by closing my eyes, but whilst reading, I get nothing, not a flicker. It really surprises me that anyone would get anything at all.

On the other hand, to be able to read without hearing any imagined voices: how can you even do that? How do you process the words to extract meaning if you don't hear them?

Eric Schwitzgebel said...

Hi Darren --

If you click through to the page with the full paper, you will see links to the raw data which contain the typewritten reports of the participants, so you can see for yourself what sort of imagery they are reporting. Unfortunately, we haven't yet thought of a good way to analyze those free response data.

Anonymous said...

I can't 'see', generate, mental images but never assumed other folks were just speaking metaphorically about their own experiences/capacities, as this inability doesn't seem to be something akin to having suffered a loss/injury would be hard to draw any inferences from it for the wider population.

Arnold said...

If you have time...
Send to Brain. 2012 Mar;135(Pt 3):678-92. doi: 10.1093/brain/aws011.
Episodic memory in frontotemporal dementia: a critical review.
Hornberger M1, Piguet O.
Author information
This review offers a critical appraisal of the literature on episodic memory performance in frontotemporal dementia. Historically, description of patients diagnosed with what was then known as Pick's disease included the presence of memory deficits and an underlying amnestic syndrome was noted in some of these patients. Over the last 20 years, however, the clinical view has been that episodic memory processing is relatively intact in the frontotemporal dementia syndrome. In particular, patients with the subtypes of behavioural variant frontotemporal dementia and progressive non-fluent aphasia are reported to perform within normal limits on standard memory tests. In the third clinical presentation of frontotemporal dementia, semantic dementia, relatively intact episodic memory against a significantly impaired semantic memory was regarded as the hallmark. This position was instrumental in the development of clinical diagnostic criteria for frontotemporal dementia in which amnesia was explicitly listed as an exclusion criterion for the disease. The relative intactness of episodic memory, therefore, appeared to be a useful diagnostic marker to distinguish early frontotemporal dementia from Alzheimer's disease, in which early episodic memory disturbance remains the most common clinical feature. We argue that recent evidence questions the validity of preserved episodic memory in frontotemporal dementia, particularly in behavioural variant frontotemporal dementia. In semantic dementia, a complex picture emerges with preservation of some components of episodic memory, notably recognition-based visual memory and recall of recent autobiographical events. We propose a critical synthesis of recent neuropsychological evidence on retrograde and anterograde memory in light of neuroimaging and neuropathological findings, demonstrating involvement of medial temporal structures in frontotemporal dementia, structures known to be critical for episodic memory processing. We further argue that the multifactorial nature of most memory tests commonly used clinically fail to capture the memory deficits in frontotemporal dementia and that sensitive assessment tools of memory are needed. Together, recent clinical and experimental findings and the historical evidence represent a strong case for a re-evaluation of the importance of memory disturbance in the clinical diagnosis of frontotemporal dementia.

Then could 'multifactorial' include memories with out words, but observation continuing...