Task Relevance Induces Momentary Changes in the Functional Visual Field During Reading

Abstract

In the research reported here, we examined whether task demands can induce momentary tunnel vision during reading. More specifically, we examined whether the size of the functional visual field depends on task relevance. Forty participants read an expository text with a specific task in mind while their eye movements were recorded. A display-change paradigm with random-letter strings as preview masks was used to study the size of the functional visual field within sentences that contained task-relevant and task-irrelevant information. The results showed that orthographic parafoveal-on-foveal effects and preview benefits were observed for words within task-irrelevant but not task-relevant sentences. The results indicate that the size of the functional visual field is flexible and depends on the momentary processing demands of a reading task. The higher cognitive processing requirements experienced when reading task-relevant text rather than task-irrelevant text induce momentary tunnel vision, which narrows the functional visual field.

Keywords

eye movements reading attention

Readers adjust the intake of text information to meet the demands of the reading task (Heller, 1982; Laycock, 1955; Rothkopf & Billington, 1975; Shebilske & Fisher, 1983). For example, in a previous study, we found that when participants read a text with a specific task in mind (e.g., reading a travel guide when planning a trip to Pitcairn Island), they made more and longer eye fixations on words in task-relevant sentences (containing information about Pitcairn) than on words in task-irrelevant sentences (containing information about other places; Kaakinen, Hyönä, & Keenan, 2002). In addition, we found that this relevance effect can be observed even on the very first words within a sentence (Kaakinen & Hyönä, 2007). Readers’ visual attention thus seems to zoom in on task-relevant information in text, as if readers are experiencing momentary tunnel vision. Tunnel vision is the narrowing of the functional visual field (i.e., the area from which useful information can be extracted; Rayner, 1998), which causes unawareness of the visual information beyond the central area of attention (Godnig, 2003). In the present study, we examined whether the size of the functional visual field changes during reading as a function of task relevance.

During normal reading, the size of the functional visual field is approximately 4 characters to the left and 14 to 15 characters to the right of current eye fixation (Rayner, 1998). The functional visual field thus usually extends over two or three consecutive words. Evidence for this has come from studies that have used the display-change paradigm (Rayner, 1975), in which a target word (e.g., “word”) is either initially masked with an invalid character string (“wyhj”) or shown normally (“word”). When the eyes move from the preceding word to the target word, the mask is replaced with the actual target word. As a result of saccadic suppression, readers will not become aware of the display change if it is completed during the saccade. The fixation time on the target word is shorter if the reader has seen the correct rather than the incorrect preview of the word. This effect (i.e., the difference between fixation times in masked and nonmasked conditions) is called the preview benefit (Hyönä, 2011; Rayner, 1998, 2009). Previous studies have suggested that even though readers may not always be able to extract the meaning of the upcoming word while fixating the previous word, visual and orthographic information is extracted (Hyönä, 2011; Rayner, 1998, 2009): If the preview mask comprises characters visually dissimilar to those of the target word, the preview benefit is approximately 40 ms, whereas if the mask is formed with visually similar characters, the preview benefit is only approximately 15 ms.

Previous research has suggested that the magnitude of the preview benefit depends on the processing load exerted by text characteristics (Henderson & Ferreira, 1990; Kliegl, Nuthmann, & Engbert, 2006). Henderson and Ferreira (1990) found that the preview benefit was smaller for words that appeared after low-frequency or syntactically ambiguous words, which both incur a momentary processing load. These results suggest that the functional visual field is not fixed but flexible: Less information is extracted from upcoming words when readers fixate difficult words than when they fixate words that are easy to process. Kliegl et al. (2006) reported corroborating analyses of corpus data.

Furthermore, evidence for information being picked up from the upcoming word has come from studies that have reported parafoveal-on-foveal effects (Drieghe, 2011; Kennedy, 2000), in which properties of the parafoveal word influence fixation times on the fixated word. For example, if the preview mask is an orthographically illegal character string, fixation duration on the preceding word is longer than if the preview mask is a real word (Inhoff, Starr, & Shindler, 2000). A study by Yan, Kliegl, Shu, Pan, and Zhou (2010) suggested that during reading of Chinese, the processing difficulty of the parafoveal text information (as induced by word frequency) influences whether information is extracted from words presented further in the parafovea, which indicates that the size of the functional visual field depends on the concurrent processing load. Task also seems to play a critical role in the occurrence of parafoveal-on-foveal effects (Drieghe, 2011). Wotschack and Kliegl (2013) reported that when participants were asked difficult questions about sentences they had read, they made more eye fixations on short and function words, which they typically skipped when easy questions were asked. Given that parafoveal-on-foveal effects are usually stronger for short and function words (Kliegl et al., 2006), the more careful reading strategy induced by difficult questions resulted in stronger statistical power to detect a parafoveal-on-foveal frequency effect.

Even though previous studies have suggested that the functional visual field during reading may be flexible and depend on the momentary processing load (Henderson & Ferreira, 1990) or on the performed task (Drieghe, 2011; Wotschack & Kliegl, 2013), no previous study has investigated global effects induced by task instructions on the preview benefit and parafoveal-on-foveal effects during normal reading. We believe that to better understand the interaction between perceptual and cognitive processes during reading, it is important to examine how task demands influence on-line text processing. Moreover, in previous studies, participants have read single, unrelated sentences, and this type of reading may differ from the “naturalistic” reading of longer text passages (Radach, Huestegge, & Reilly, 2008).

In the present study, we asked participants to read a lengthy expository text about different countries and instructed them to imagine that they were going to move to either Honduras or Pitcairn Island. These instructions made information related to the target country highly relevant. We expected that the size of the functional visual field would vary depending on task relevance and would result in a reduced preview benefit for words within relevant text segments in comparison with words within irrelevant text segments. Moreover, we assumed that readers would use more information from the parafovea when reading irrelevant sentences than when reading relevant sentences, which would result in stronger parafoveal-on-foveal effects for target words in irrelevant text segments than in relevant text segments.

Method

Participants

Forty university students (native speakers of Finnish) with normal or corrected-to-normal vision participated in the study.

Apparatus

Eye movements were recorded with an EyeLink 2000 eye tracker (SR Research, Mississauga, Ontario, Canada) using a 1000-Hz sampling rate. The stimuli were presented at a viewing distance of 70 cm from the participant on a 21-in. CRT monitor using a 150-Hz refresh rate and a resolution of 1,024 × 768 pixels.

Materials

An expository text describing Anguilla Island, Pitcairn Island, Honduras, and Andorra was adopted from our earlier study (Kaakinen et al., 2002). The original text contains 1,125 words (113 sentences) and has a compare-contrast rhetorical structure: The countries are discussed in varying order under different subtopics (e.g., climate, history, population). We selected 28 sentences describing Pitcairn Island and Honduras from the text and embedded the target words within these sentences. There were 14 targets (mean length = 9.21 characters, SD = 1.63; mean log frequency per million = 2.04, SD = 0.72) embedded in Pitcairn Island–related sentences and 14 targets (mean length = 9.21 characters, SD = 1.97; mean log frequency per million = 1.97, SD = 1.05) embedded in Honduras-related sentences. The targets consisted of 17 nouns (6 of which were compound words), 10 verbs, and 1 adjective. The mean length of the words preceding the target words (referred to as the target – 1 words) was 9.21 characters (SD = 2.99) in Honduras-relevant sentences and 9.43 characters (SD = 2.65) in Pitcairn Island–relevant sentences.

Procedure

After participants had signed the informed-consent form, the eye tracker was calibrated. Participants were instructed to imagine that they would be moving to Honduras (n = 20) or to Pitcairn Island (n = 20) and to read the text so that they could determine the pros and cons of their imaginary future home country. Thus, for half of the participants, sentences that described Honduras were task relevant and Pitcairn Island–related sentences were task irrelevant, whereas for the other half of the participants, Honduras-related sentences were task irrelevant and Pitcairn Island sentences were task relevant. The text was presented on a white background in 12-point Courier New black type at 2.5 line spacing with a maximum of 10 lines per page, which resulted in 15 text pages. Participants were allowed to read the text at their own pace.

We used the display-change technique (Rayner, 1975), in which a random-letter string (the first character was the same as in the target word) was used as the preview mask. There were two targets on 14 of the 15 pages of text (the 1st page did not include target words). Only 1 display change was performed on each of these pages, for a total of 14 changes (7 in relevant targets and 7 in irrelevant targets). The target words for which a display change was performed were counterbalanced across participants, and the experimental session lasted approximately 1 hr.

Results

We removed fixations shorter than 50 ms from the data. Trials in which the display change was not performed on time (< 5 ms from the start of the first fixation; McConkie & Loschky, 2002); in which the boundary change was triggered too early as a result of a blink, a return sweep, or a premature saccade to the end of the text page; and in which the target word or the target – 1 word was skipped were removed from the data. These exclusions resulted in a data set consisting of 855 observations from 40 participants and 28 items (23.7% data loss). In the final data set, there were no differences between the conditions in the mean length of the target – 1 words or in the length or frequency of the target words (Fs < 1).

The fixation-time measures were log transformed and the data were analyzed with (generalized) linear mixed-effects models (Baayen, Davidson, & Bates, 2008) using R statistical software (R Development Core Team, 2011). Participants and items were specified as crossed random effects. Preview condition (change vs. no change; deviation coded), relevance (relevant vs. irrelevant; deviation coded), and the Preview × Relevance interaction were entered in the models as fixed effects. The probability values of the t tests were estimated using posterior distributions for model parameters obtained by means of Markov chain Monte Carlo sampling. The fixed-effects estimates, standard errors, t-test and z scores, and probability values are presented in Table 1.

Table 1.

Fixed-Effects Estimates for Models of Eye Movement Measures

Measure and fixed effect	Estimate	t or z value^a	p
Single-fixation duration on target – 1 word^b
Intercept	5.35 (0.03)	164.30	.0001
Relevance	−0.04 (0.02)	−1.68	.1072
Preview	0.02 (0.02)	0.82	.4102
Relevance × Preview	−0.14 (0.05)	−3.02	.0030
Gaze duration on target – 1 word
Intercept	5.49 (0.04)	128.02	.0001
Relevance	−0.01 (0.03)	−0.55	.5902
Preview	0.02 (0.03)	0.58	.5678
Relevance × Preview	−0.10 (0.05)	−1.88	.0634
First-fixation duration on target word
Intercept	5.37 (0.03)	211.65	.0001
Relevance	0.01 (0.02)	0.60	.5488
Preview	0.04 (0.02)	1.79	.0732
Relevance × Preview	−0.10 (0.05)	−2.02	.0430
Gaze duration on target word
Intercept	5.54 (0.03)	159.87	.0001
Relevance	0.01 (0.03)	0.47	.6416
Preview	0.09 (0.03)	3.46	.0004
Relevance × Preview	−0.08 (0.05)	−1.49	.1310
Launch site of first fixation
Intercept	5.91 (0.35)	16.85	.0001
Relevance	−0.01 (0.18)	−0.05	.9872
Preview	−0.09 (0.18)	−0.47	.6594
Relevance × Preview	0.08 (0.36)	0.23	.8218
First-fixation duration on the spillover region^c
Intercept	5.31 (0.04)	150.40	.0001
Relevance	−0.0003 (0.03)	−0.01	.9856
Preview	0.01 (0.03)	0.41	.6634
Relevance × Preview	−0.02 (0.06)	−0.29	.7656
Probability of regression from the spillover region^c
Intercept	−3.00 (0.27)	−11.20	< .0001
Relevance	0.23 (0.33)	0.68	.494
Preview	0.47 (0.34)	1.41	.159
Relevance × Preview	−0.21 (0.67)	−0.31	.755

Note: Standard errors are shown in parentheses. The reported models include intercepts for participants and items as random effects. To examine individual differences in the experimental effects, we also fitted models with by-subjects random slopes for the fixed effects. The models were trimmed by removing variance components one at a time until exclusion resulted in a significantly poorer fit of the model as indicated by a likelihood-ratio test. The variance component of the random slope for the interaction term (Relevance × Preview) did not remain in any of the models. The significance of the reported fixed effects did not change when random slopes for the fixed effects were included. target – 1 = word preceding the target word.

In this column, values for the probability of regression from the spillover region are z values, and all other values are t values. ^bThis analysis was based on 657 observations of single fixations on the target – 1 word. ^cThis analysis was based on 626 observations.

Parafoveal-on-foveal effects

The analysis of the single-fixation duration on the target – 1 word revealed a significant interaction between task relevance and preview (see Table 1), which indicated that the magnitude of the parafoveal-on-foveal effect depended on task relevance. As shown in Figure 1, for words embedded in task-relevant sentences, the difference between preview conditions was not significant, t = −1.55, p = .1326. However, for words embedded in task-irrelevant sentences, a significant parafoveal-on-foveal effect (23 ms) was observed, t = 2.71,¹ p = .007.

Fig. 1.

Mean estimated (a) single-fixation duration and (b) gaze duration on the word preceding the target word (target – 1; back-transformed into milliseconds) as a function of sentence type and preview condition. Error bars represent standard errors.

The analysis of the gaze duration on the target – 1 word (summed duration of fixations during the first pass) indicated a marginally significant interaction between task relevance and preview (see Table 1). For the task-relevant sentences, there was no difference between the preview conditions, t = −0.93, p = .3462 (see Fig. 1). For the task-irrelevant sentences, the mean gaze duration was 14 ms longer in the change preview condition than in the no-change preview condition, but the difference did not quite reach significance, t = 1.73, p = .0790.

Preview benefit

The Preview × Relevance interaction was significant for the first-fixation duration on the target word, which indicated that the magnitude of the preview effect depended on the relevance of the sentence (see Table 1). As shown in Figure 2, there was a clear preview effect for the irrelevant sentences, t = 2.68, p = .0060, whereas the effect disappeared for the relevant sentences, t = −0.16, p = .8644. The magnitude of the preview effect was 21 ms for the irrelevant sentences and −6 ms for the relevant sentences.

Fig. 2.

Mean estimated (a) first-fixation duration and (b) gaze duration on the target word (back-transformed into milliseconds) as a function of sentence type and preview condition. Error bars represent standard errors.

Although the preview effect was significant for the gaze duration on the target word (see Table 1), the Preview × Relevance interaction did not reach significance. However, we computed the pairwise comparisons to examine the preview effect separately for relevant and irrelevant sentences and replicated the finding from the first-fixation duration (see Fig. 2): The preview effect was significant for the irrelevant sentences, t = 3.48, p = .0006, but not for the relevant sentences, t = 1.40, p = .1552. The preview effect was 33 ms for the irrelevant sentences and 12 ms for the relevant sentences.

Launch site

None of the effects for the launch site of the incoming fixation to the target word were significant (see Table 1).

Spillover effects

We also analyzed the first-fixation duration on the spillover region (i.e., the word following the target word) and the probability of regression back from the spillover region (see Table 1). For the first-fixation duration on the spillover region, none of the effects were significant. In addition, none of the effects proved significant for the probability of regression.

Discussion

The present results demonstrate that task relevance induces momentary tunnel vision during reading: No significant preview benefit or parafoveal-on-foveal effects were observed for words embedded in task-relevant text segments. In contrast, significant preview benefits and parafoveal-on-foveal effects were observed for irrelevant sentences; however, the effects were immediate and short-lived, given that there was no indication of spillover effects. The reported preview and parafoveal-on-foveal effects indicate that readers processed the orthography of the upcoming word in the task-irrelevant sentences but not in task-relevant sentences.

The finding of no reliable orthographic preview effect for task-relevant sentences is surprising and novel, given that previous studies using orthographic preview masks typically have shown robust preview effects (approximately 40 ms in gaze duration; Hyönä, 2011). This result suggests that the momentary processing demands of encoding relevant information to memory influence the amount of information that is extracted and processed from the parafovea. The findings are in line with results reported by Henderson and Ferreira (1990), who showed that local difficulty in lexical and syntactic processing reduces the preview benefit. The present results are further strengthened by the fact that we used relatively long target words to ensure that the words were fixated, which should have maximized the preview effect. Still, the preview effect was nonexistent in the relevant sentences (we found a nonsignificant 12-ms trend in the gaze duration).

The demonstration of parafoveal-on-foveal effects for irrelevant sentences but not for relevant sentences further shows that the engagement of attention is affected by task demands during reading. The relevance (or irrelevance) of text information was made explicit in the text: The name of the country being discussed was mentioned either in the beginning of the target sentence or in the sentence preceding it. Thus, when beginning to read a target sentence, the reader knew if it was task relevant or irrelevant. The present results suggest that when readers recognized text information to be irrelevant, their attention was attracted more toward the upcoming information in the parafovea. When the processing of text information required more attentional resources, such as when participants read task-relevant sentences, attentional span was narrowed down to cover only the fixated word. Our results are in line with previous findings that have shown that the degree of parafoveal processing may be influenced by processing difficulty (Kliegl et al., 2006; Yan et al., 2010) and that the magnitude of the observed parafoveal-on-foveal effects depends on task demands (Wotschack & Kliegl, 2013).

In sum, the present study shows that task relevance induces momentary tunnel vision and restricts the functional visual field to the currently fixated word only. The results thus indicate that the functional visual field during reading is flexible and that its size depends on the moment-to-moment processing demands induced by the reading task. In other words, when readers adjust the intake of text information to meet the demands of the reading task, they are juggling their limited attentional resources between memory encoding and integration and visual sampling of new information in text.

Footnotes

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Notes

References

Baayen

R. H.

Davidson

D. J.

Bates

D. M.

(2008). Mixed-effects modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59, 390–412.

Drieghe

(2011). Parafoveal-on-foveal effects on eye movements during reading. In Liversedge

Gilchrist

Everling

(Eds.), The Oxford handbook of eye movements (pp. 839–855). Oxford, England: Oxford University Press.

Godnig

E. C.

(2003). Tunnel vision: Its causes and treatment strategies. Journal of Behavioral Optometry, 14, 95–99.

Heller

(1982). Eye movements in reading. In Groner

Fraisse

(Eds.), Cognition and eye movements (pp. 139–154). Amsterdam, The Netherlands: North Holland.

Henderson

J. M.

Ferreira

(1990). Effects of foveal processing difficulty on the perceptual span in reading: Implications for attention and eye movement control. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 417–429.

Hyönä

(2011). Foveal and parafoveal processing during reading. In Liversedge

Gilchrist

Everling

(Eds.), The Oxford handbook of eye movements (pp. 819–838). Oxford, England: Oxford University Press.

Inhoff

A. W.

Starr

Shindler

K. L.

(2000). Is the processing of words during eye fixations in reading strictly serial? Perception & Psychophysics, 62, 1474–1484.

Kaakinen

J. K.

Hyönä

(2007). Perspective effects in repeated reading: An eye movement study. Memory & Cognition, 35, 1323–1336.

Kaakinen

J. K.

Hyönä

Keenan

(2002). Perspective effects on online text processing. Discourse Processes, 33, 159–173.

10.

Kennedy

(2000). Parafoveal processing in word recognition. Quarterly Journal of Experimental Psychology: Section A, 53, 429–455.

11.

Kliegl

Nuthmann

Engbert

(2006). Tracking the mind during reading: The influence of past, present, and future words on fixation durations. Journal of Experimental Psychology: General, 135, 12–35.

12.

Laycock

(1955). Significant characteristics of college students with varying flexibility in reading rate: I. Eye movements in reading prose. Journal of Experimental Education, 23, 311–319.

13.

McConkie

G. W.

Loschky

L. C.

(2002). Perception onset time during fixations in free viewing. Behavioral Research Methods, Instruments, & Computers, 34, 481–490.

14.

R Development Core Team. (2011). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Available from http://www.R-project.org/

15.

Radach

Huestegge

Reilly

(2008). The role of global top-down factors in local eye-movement control in reading. Psychological Research, 72, 675–688.

16.

Rayner

(1975). The perceptual span and peripheral cues in reading. Cognitive Psychology, 7, 65–81.

17.

Rayner

(1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 134, 372–422.

18.

Rayner

(2009). Eye movements and attention in reading, scene perception, and visual search. Quarterly Journal of Experimental Psychology, 62, 1457–1506.

19.

Rothkopf

E. Z.

Billington

M. J.

(1979). Goal-guided learning from text: Inferring a descriptive processing model from inspection times and eye movements. Journal of Educational Psychology, 71, 310–327.

20.

Shebilske

W. L.

Fisher

D. F.

(1983). Eye movements and context effects during reading of extended discourse. In Rayner

(Ed.), Eye movements in reading: Perceptual and language processes (pp. 157–179). New York, NY: Academic Press.

21.

Wotschack

Kliegl

(2013). Reading strategy modulates parafoveal-on-foveal effects in sentence reading. Quarterly Journal of Experimental Psychology, 66, 548–562.

22.

Yan

Kliegl

Shu

Pan

Zhou

(2010). Parafoveal load of word N+1 modulates preprocessing effectiveness of word N+2 in Chinese reading. Journal of Experimental Psychology: Human Perception and Performance, 36, 1669–1676.