- Email: [email protected]
Activation in left primary visual cortex representing parafoveal visual field during reading Japanese texts
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
available at www.sciencedirect.com
www.elsevier.com/locate/brainres
Research Report
Activation in left primary visual cortex representing parafoveal visual field during reading Japanese texts Yoichi Shimadaa, b,⁎, Kazumi Hirayamaa, c, 1, 2 , Satoshi Nakadomarid, e, 3, 4 , Ayumu Furuta f, g, 5, 6 , Masaya Misaki h, i, 7, 8 , Shigeyuki Kanh, i, 7, 8 , Takahiko Koikeh, i, 7, 8 , Satoru Miyauchih, i, 7, 8 , Etsuro Moria, 1 a
Department of Behavioral Neurology and Cognitive Neuroscience, Tohoku University Graduate School of Medicine, 2-1, Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan b Department of Rehabilitation, Faculty of medical science and welfare, Tohoku Bunka Gakuen University, 6-45-1, Kunimi, Aoba-ku, Sendai 981-8551, Japan c Department of Occupational Therapy, Yamagata Prefectural University of Health Sciences, 260, Kamiyanagi, Yamagata City, Yamagata 990-2212, Japan d Department of Medical Treatment (2), Hospital, National Rehabilitation Center for Persons with Disabilities, 4-1, Namiki, Tokorozawa, Saitama 359-8555, Japan e Department of Ophthalmology, Jikei University, School of Medicine, 3-25-8, Nishishinbashi, Minato-ku, Tokyo 105-8461, Japan f Maeda Ophthalmic Clinic, 3-30, Naka-machi, Aizu Wakamatu, Fukushima 965-0878, Japan g Department of Ophthalmology, Fukushima Medical University, 1, Mitsugaoka, Fukushima City, Fukushima 960-1295, Japan h Kobe Advanced ICT Research Center, National Institute of Information and Communications Technology (NICT), 588-2, Iwaoka-machi, Nishi-ku, Kobe 651-2492, Japan i Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency ( JST), 5, Sanban-machi, Thiyoda-ku, Tokyo 102-0075, Japan
A R T I C LE I N FO
AB S T R A C T
Article history:
Activation in the left primary visual cortex (V1) representing the parafoveal field during text
Accepted 16 June 2011
reading has been interpreted as attentional modulation in the process of deciding saccadic
Available online 24 June 2011
target for reading ahead. Kanji words serve the main cue to decide the goal of saccades in Japanese. We aimed to determine the exact location of this modulation in the V1 and to
Keywords:
determine whether the area of the modulation changes according to the location where the
Parafoveal activation
next Kanji word appears or it is fixed on a certain region in V1. Using functional magnetic
Reading saccade
resonance imaging, we determined the area in V1 representing each eccentricity on the
⁎ Corresponding author at: 6-45-1, Kunimi, Aoba-ku, Sendai 981-8551, Japan. Fax: + 81 22 233 4054, + 81 22 717 7360. E-mail addresses: [email protected] (Y. Shimada), [email protected] (K. Hirayama), [email protected] (S. Nakadomari), [email protected] (A. Furuta), [email protected] (M. Misaki), [email protected] (S. Kan), [email protected] (T. Koike), [email protected] (S. Miyauchi), [email protected] (E. Mori). 1 Fax: +81 22 717 7360. 2 Fax: +81 23 686 6674. 3 Fax: +81 4 2995 3102. 4 Fax: +81 3 3433 1111. 5 Fax: +81 242 29 3556. 6 Fax: +81 24 547 1989. 7 Fax: +81 78 969 2215. 8 Fax: +81 3 3222 2066. 0006-8993/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.brainres.2011.06.043
73
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
Attentional modulation
horizontal meridian of the visual field for each participant. Then we investigated brain
Primary visual cortex
activation while they were reading two sets of Japanese texts that scrolled leftward as the
Japanese text
participants. In set 1, the distance between the heads of adjacent Kanji words was about 3°.
Functional magnetic resonance
In set 2, the distance was about 5°. From the results of these experiments, we obtained
imaging
activation amplitude of the area corresponding to each eccentricity. We recorded eye movements simultaneously with the acquisition of fMRI data. The maximum peak of the activation was found in the region representing about 4.5° of eccentricity on the horizontal meridian in the left V1 for each participant. The activation pattern did not essentially differ between the two text conditions, although the location of the saccades made for reading next section of the text corresponds to the head of the next Kanji word. The activation modulation during reading Japanese texts occurs in the parafoveal V1 of the left hemisphere. The attentional modulation did not change with the distance to the next goal of saccade but was fixed on the area representing about 4.5° of eccentricity. © 2011 Elsevier B.V. All rights reserved.
1.
Introduction
In order to achieve smooth, efficient reading, one has not only to recognize each word embedded in a text but also to plan the eye movements to the next viewing point. As visual acuity decreases sharply with increasing eccentricity (Anstis, 1974), only the visual information from the foveal and parafoveal visual fields contributes to these process (Fig. 1). Word form is perceived in the foveal field extending 1° of visual angle either side of fixation (Rayner and Bertera, 1979), where the high spatial acuity enables one to discriminate constituent letters clearly. In the parafoveal visual field, one can extract only more gross features. However, the information from the parafoveal region can be enhanced by attentional modulations and thus contribute to reading process (Miellet et al., 2009). Perception of the ensuing words, to plan the next saccade, occurs in the right parafoveal field extending 5° for those who have left to right reading habit. Zihl (2000) named foveal and right parafoveal region on the horizontal meridians together (Fig. 1) as “perceptual window” for reading. When this parafoveal preview of upcoming text is not a valid extension of the current text, reading time is slowed. In contrast, invalid
Fig. 1 – Critical visual field for efficient reading and an example of a Japanese written sentence. Foveal vision is depicted in white and parafoveal vision in pale gray. The black boundary represents the critical area for efficient reading which is called perceptual window. †Kannji. *Hiragana.
information from text in the left parafovea does not affect reading time (McConkie and Rayner, 1975). This asymmetry reflects attentional demands linked to reading direction: In Hebrew (which is read from right to left) the interference by invalid information occurs in the left parafovea (Pollatsek et al., 1981). Patients with right homonymous hemianopia with macular sparing of less than 5° may show an impairment of reading. This is caused by a difficulty in acquiring information necessary for deciding the next saccadic target for reading (hemianopic dyslexia) (Zihl, 1995, 2000). Abnormal eye movement patterns during reading are also found in developmental dyslexics (Biscaldi et al., 1998). Thus, it is important to know the neural basis of saccades guided by the information acquired from peripheral vision in healthy subjects in order to understand the mechanisms of these impairments. Leff et al. (2000) investigated reading process of English word array with positron emission tomography (PET). They found a peak of activation in an area in the left calcarine cortex representing the right parafoveal visual field, and interpreted it as attentional modulation in the primary visual cortex (V1) in the process of deciding saccadic targets necessary for reading ahead. However, there remain some questions. The result might not reflect natural process of reading a text. The activation reported in their study was obtained by contrasting the viewing condition of reading meaningless strings of words and of viewing a single word. It is not clear what part of the parafoveal region of V1 was activated. They did not ascertain the retinotopic organization of V1 for each participant. It remains uncertain if exact locations of the activation found for each participant were comparable or not, as they just showed the activated locations averaged across the 5 participants. In the languages that use Roman script, selection of saccade targets is based on information about word boundaries, which is provided visually salient spaces inserted between words (McConkie et al., 1988). However, since there is no such spacing between words or between individual letters in the Japanese writings, space does not provide the information necessary to locate word boundaries. Typical modern Japanese texts are written primarily in a mixture of two basic scripts: Kanji, a script, and Hiragana, a syllabarie (Fig. 1). Usually, Kanji encodes grammatical categories such as nouns, verb stems and adjective stems. Hiragana serves to
74
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
represent grammatical element of sentences such as particles, auxiliary verbs, and inflectional affixes of noun, adjectives or verbs. However, words usually written in Kanji can also be expressed in Hiragana, as each Kanji word has its specific Japanese syllables. In general, Kanji characters are visually more complex. They also consist of more straight lines than Hiragana characters, most of which are composed of curved lines. Kajii et al. (2001) investigated eye movements in reading Japanese texts, and found that the eyes tend to land at the head of Kanji words. Thus, Kanji words that occur in the right parafoveal visual field seem to serve the main cue to decide the goal of saccades in reading a text of this language. Thus, an important research question is to localize the exact region of the “attentional modulation” in the parafoveal V1 while ones are reading Japanese texts. There are two possibilities. One possibility is that an area corresponding to the goal of next saccade where the next Kanji word is laid would be activated (goal-dependent modulation). Another possibility is that the activation would be found at a fixed location corresponding to somewhere in the perceptual window, for example the right edge of it (fixed-location modulation). This study aimed to test which of these two areas is activated more prominently in the healthy subjects while they were reading texts in Japanese. Using functional magnetic resonance imaging (fMRI), we investigated brain activation when healthy participants read meaningful Japanese sentences that scrolled leftward as the participants read them. We attempted to determine the exact location of activation in the parafoveal V1, for each participant, rather than to obtain the averaged location for the participants. Furthermore, we tried to find out whether the attentional modulation was dynamically controlled in accordance with the location of next saccade which corresponded to the head of next Kanji word or was fixed at some location in the perceptual window, by comparing fMRI data and eye movement record.
2.1° of eccentricity (Fig. 4-C). The activation during reading 5° texts showed a maximum peak on 5.5° and another peak on 2.8°, and showed a rise on 2.1° of eccentricity (Fig. 4-D). For amplitude of activation, we performed a three-way repeated measures ANOVA with text type (3 texts and 5 texts), hemisphere (left and right), and eccentricity (20 locations: 2.2, 2.3, 2.5, 2.7, 2.8, 3.0, 3.2, 3.4, 3.6, 3.9, 4.1, 4.4, 4.7, 5.0, 5.3, 5.7, 6.0, 6.4, 6.8, and 7.3°) as factors. The three-way ANOVA showed a significant main effect of hemisphere (F (1, 5) = 7.17, p < .05). The average of the all amplitude in left hemisphere was 1.55 and that of left hemisphere was 0.97. Thus, the activation of left parafoveal V1 was greater than that in right parafoveal V1. An interaction between text type and eccentricity was also significant (F (19, 95) = 2.48, p < .01). To follow up this interaction, a two-way repeated measures ANOVA was conducted separately for each hemisphere. In the left hemisphere, twoway ANOVA showed a significant interaction between text type and eccentricity (F (1, 19) = 2.41, p < .01). In the right hemisphere, there was no significant effect. Fig. 5 shows the activation amplitude of each participant along the horizontal meridian. As the rising activations toward the area representing 2.1° of eccentricity in the left and right hemispheres thought to reflect the margin of foveal V1, we sought the eccentricity of each participant which had the amplitude of largest local maximum other than these rising toward 2.1°. In left hemisphere the average of the eccentricities was 4.6° (range 4.1° to 4.7°) for 3° texts and 4.6° (range 3.9° to 5.3°) for 5° texts. In the right hemisphere the average of the eccentricities was 3.8° (range 2.2° to 7.3°) for 3° texts and 5.1° (range 2.2° to 7.3°) for 5° texts. Thus all participants had the largest peak somewhere in the restricted range close the eccentricity of 4.6°, and it did not change according to the distance to the head of the next Kanji word. In contrast, the largest peak of each participant was distributed among broad range.
2.
Results
2.2.
2.1.
fMRI data of text reading
Saccadic eye movement data obtained from two participants (subjects 1 and 2 of Fig. 5) showed that reading related saccades occurred actually from the head of a Kanji word to the next head of subsequent Kanji word (Fig. 6). In subject 1, the mean amplitude of the saccades during reading 3° texts was 3.22° ± 1.07 of visual angle while that of 5° texts was 5.14° ± 1.57. In subject 2, mean amplitude of the saccades during reading 3° texts was 2.96° ± 0.78 of visual angle while that of 5° texts was 4.27° ± 1.01. Thus, the location of the saccades made for reading next section of the text appeared to correspond to the head of the next Kanji word.
All of the participants answered sufficiently well to the questions concerning the texts. Thus, they seemed to understand the texts. We presented to the participants scrolling texts whose distance between the heads of adjacent Kanji words was about 3° (3° texts) and whose distance between the heads of adjacent Kanji words was about 5° (5° texts) (Fig. 3-A). If the attentional modulation was dynamically controlled in accordance with the location of next saccade which corresponded to the head of next Kanji word, the peaks of activation in the left V1 would shift according the distances. If it was fixed at some location in the perceptual window, the location would not shift according the distances. The mean Z score of activation amplitude on the horizontal meridians represented in the left V1 during reading 3° texts and reading 5° texts showed essentially similar patterns. Both of them had a maximum peak on 4.5° and other peaks on 2.7° and 6.7° of eccentricities, and showed a rise on 2.1° of eccentricity (Figs. 4-A, B). In the right V1, the activation during reading 3° texts showed no obvious peak, and showed a rise on
3.
Saccadic eye movement data of text reading
Discussion
Using fMRI, we studied activation in V1 during reading meaningful Japanese sentences. In the left V1 area representing the parafoveal horizontal meridian, the maximum peak of the activation was found in the region representing about 4.5° of eccentricity not only in the average data of all the participants but also in each participant. In contrast to the left V1, no common pattern of activation was found in the
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
75
Fig. 2 – Determination of horizontal meridian and location of the eccentricity of 5° in V1. (A) The dartboard like stimuli to determine the area in V1 representing horizontal meridian and location of the eccentricity of 5° along the meridian. These two sections were presented alternatively with a cycle of 25.6 s. The center of the dartboard, changed from red to white every 10 s for 0.5 s. (B) Response of a participant (participant 4 of Fig. 5) elicited during the viewing of these stimuli is shown on an inflated cortical surface of his left hemisphere. (C) The response was converted into a flat map. Black boundary represents the area V1 obtained beforehand, using the paradigm reported by Furuta et al., 2009. HM: Red lines represent horizontal meridian. Ec5°: the eccentricity of 5°. Fo: the area which represents fovea. (D) The color of the activated area reflects phase, i.e. timing (second) in the cycle of visual stimuli, as indicated by this color bar.
right V1. Thus, the findings of this study demonstrated that with the naturalistic material the activation occurred in the parafoveal V1 of the left hemisphere as was shown by Leff et al. (2000) with a somewhat artificial text of random word sequence. The greatest activated region was the right edge of the perceptual window throughout the participants. That focused attention can modulate activity in the visual cortex has been reported in many human fMRI studies (Martinez et al., 1999). In those studies visual stimuli were displayed in the neighborhood of the parafoveal area and the participants were instructed not to make eye movements. Leff et al. (2000) interpreted that the activation in the parafoveal V1 was caused by attention that was directed to the region to determine next saccadic target. The results of our experiments strongly support their interpretation. The activation pattern of each participant showed no essential difference between the two text conditions with different distance between the heads of adjacent Kanji words. In our experiment, the distances of Kanji words within a set of texts were constant. Thus, it might be expected that the peak activation is different with the longer distance condition having a peak at a further peripheral area in comparison with the one found for the shorter distance condition, if attention was directed to the next goal of reading saccade. However, the result showed that the maximum peak of activation was in the area representing about 4.5° of eccentricity in the left V1 even for the stimuli of texts of 3° distance. This was also true for the
two participants whose eye movements during reading were monitored. Their eye movements were found to be sensitive to the distance condition. The mean amplitude of the saccades during reading 3° texts was about 3°, while that of 5° texts was about 5° for each of them. Thus, the attentional modulation in V1 did not change with the distance to the next goal of saccade (goal-dependent modulation) but was fixed on the area representing the right edge of perceptual window (fixedlocation modulation). However, as the amplitudes were obtained as a total effect of 12.8-second text reading. The average location of activation during text reading is not adequate to test the effect of flexible ocular control on the neural activity in V1. It should be kept in mind that our result does not deny the presence of such a transient and dynamic modulations related to the goal of saccade. The reason why the modulation occurred at the edge of the right parafoveal region is uncertain. In parafoveal region, the acuity was lowest at its peripheral edge. Thus, elevating the sensitivity of this region would be helpful in getting information necessary for smooth and efficient control of reading. In the paper, we regarded the eye movement as voluntary saccade. However, the movement may be interpreted as a sort of optokinetic nystagmus (OKN). OKN is generally defined as involuntary rhythmic movements of the eyes caused by unidirectional motion of the visual field. However, Honrubia et al. (1968) classified OKN into two types; “look OKN” induced by active following of the targets without turning of the head,
76
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
Fig. 3 – Stimuli presentation and visualizing activation. (A) An example of the 3 texts. The distance between the heads of adjacent Kanji words was about 3° (top). An example of the 5 texts. The distance between the heads of adjacent Kanji words was about 5° (bottom). (B) In a trial, a drifting text was presented for 12.8 s (top), which was replaced by a blank screen for 12.8 s (bottom). (C) Response of a participant (participant 4 of Fig. 5) elicited during the reading task is shown on an inflated cortical surface of his left hemisphere. (D) The response was mapped onto flat map. Black boundary represents the area V1 obtained beforehand, using the paradigm reported by Furuta et al., 2009. Red lines represent horizontal meridian (HM) and the eccentricity of 5° (Ec5°), obtained beforehand, using the alternating dartboard like stimuli. A linear ROI along the horizontal meridian was determined. The voxels that corresponded to the eccentricity of 5° along the horizontal meridian as a node of these two lines. The amplitude of the activations at each voxel in the ROI was measured. Fo: the area which represents fovea. (E) The color of the activated area reflects phase, i.e. timing (second) in the cycle of visual stimuli, as indicated by this color bar.
Fig. 4 – Mean Z scores of activation amplitude of the 6 participants. (A) Response in the left V1 during reading the 3° texts. (B) Response in the left V1 during reading the 5° texts. (C) Response in the right V1 during reading the 3° texts. (D) Response in the right V1 during reading the 5° texts.
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
Fig. 5 – Activation amplitude of each participant. (A) Response in the left V1 during reading the 3° texts. (B) Response in the left V1 during reading the 5° texts. (C) Response in the right V1 during reading the 3° texts. (D) Response in the right V1 during reading the 5° texts. Italic numbers indicate each participant.
and “stare OKN” involuntary appearing when the subject inattentively directs gaze towards the moving patterns of targets. In our study, as the instruction to the subjects was to read and understand the stimulus texts, the induced eye movements should be corresponded to “look OKN”. Kashou et al. (2010) demonstrated the difference of activation on fMRI between two types of OKN by changing the instruction, and concluded that “look OKN” is somewhat of a misnomer and is indeed a voluntary series of pursuit and saccadic eye movements. Therefore, we think that the eye movements in our study are regarded as “look OKN” or saccade, although both are two aspects of the same phenomenon. There are two limitations in our experiment. First, the text was scrolled leftward as the participant. We chose this presentation in order to keep off leftward shifts of attention during reading on another line and keep the amount of the visual stimuli similar on either side of the fixation during the task. Thus, the stimuli used in our experiment were artificial in presentation method, although they were more natural in material comparing with the stimuli used by Leff et al. (2000). Reading the texts scrolling leftward may induce larger
77
Fig. 6 – Examples of eye movement during reading task. Arrows indicate horizontal locations of viewing point on a text. The change from 1 to 2 represents a pursuit for a Kanji meaning ‘one’ which is the head of Kanji word meaning ‘generally’. The change from 2 to 3 represents a saccade from a Kanji meaning ‘one’ to another Kanji meaning ‘same’. The change from 3 to 4 represents a pursuit for a Kanji meaning ‘same’. The change from 4 to 5 represents a saccade from a Kanji meaning ‘same’ to another Kanji meaning ‘occur’. Reading-related saccades occurred usually from the head of a Kanji word to the next head of Kanji word.
attention toward the right edge of the perceptual window comparing to the static texts. Thus, there was a possibility that the movement of the text enhanced the activation of the V1 area representing this portion of visual field. Second, the number of participants was small. To generalize our conclusion about the hemispheric difference of activation pattern it should be supported by statistical analyses over a larger group of participants. Finally, clinical implication of our finding is discussed. Leff et al. (2000) found that the activation patterns in the posterior parietal cortices and in the frontal eye fields were different between patients with hemianopic dyslexia and those of the healthy subjects, and interpreted that this difference was caused by a disorder in the control of visual attention. If in hemianopic dyslexic patients with a lesion in the optic radiation without damage in V1, the method employed in this study might show disordered modulation with inadequate activation peaks in V1. Abnormal eye movement patterns during reading have been reported also in developmental dyslexics (Adler-Grinberg and Stark, 1978). Biscaldi et al. (1998) reported that saccadic eye movements toward the visual stimuli displayed outside of the parafoveal vision were impaired in developmental dyslexics, also
78
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
impairment was correlated with the degree of reading disability, and maintained that reading process and saccade system are both controlled by visual attention systems, which may well be impaired in developmental dyslexics. Our method is readily applied to developmental dyslexics to examine how the peak locations are changed.
4.
Experimental procedure
4.1.
Participants
Six healthy males, aged 33 to 54 years took part. They were Japanese speaking, with normal vision and were right handed. All gave informed written consent. All procedures adhered to protocols based upon the world medical association declaration of Helsinki ethical principles for medical research involving human subjects, approved by the ethical committees of the National Institute of Information and Communications Technology (NICT), Japan.
4.2.
Visual tasks
4.2.1.
Localizing task
Before the reading task, we determined the area V1 and its direction of con-eccentric areas in each participant using the paradigm reported by Furuta et al. (2009) in which they used a stimulus of expanding ring and clockwise rotating wedges made of periodically contrast reversing checkerboard pattern. Then, we determined the horizontal meridian that extends to both the left and right directions encompassing the fixation point and the exact locations of the eccentricity of 5° along this meridian in each visual field for bilateral V1 of each participant. For these purposes, we prepared a circular figure with a radius of 16°, which was divided into 126 sectors by 18 radial lines and 7 concentric circles. Each sector was colored in black or white in alternating fashion to make a dartboard-like figure. Each sector of the dartboard reversed its color at 4 Hz. The dartboard was separated into two larger sections by its horizontal meridian and a circle of 5° of eccentricity as shown in Fig. 2-A. These two sections were presented alternatively with a cycle of 25.6 s and the cycle was repeated 18 times. To secure the fixation, participants were asked to push a button when the color of the fixation point, at the center of the dartboard, changed from red to white every 10 s for 0.5 s.
4.2.2.
Reading task
Nine plain passages taken from scientific essay of popular magazines were used as the reading materials. Fig. 7 shows an example of the texts. The passages consisted of 2 to 5 sentences with 111 to 142 characters. Each text was modified as follows. Kanji words and Hiragana words in the passage were rearranged to make an array of Kanji and Hiragana characters, in which the interval of the heads of Kanji words was about 4 to 6 characters ahead without deranging both grammatical appropriateness and the meaning of the original text. We prepared two sets of stimuli from these modified passages. The characters of both sets had identical font and height (1.56°). Each set was different only in the widths of the characters and space between the characters. In set 1, the
Fig. 7 – An example of the Japanese texts used in the reading task. It means, ‘Why do we sleep. We have not found even the answer of this question. It is not true that sleep has only a static role for rest. During sleeping, various phenomena occur in our brain. I will tell you above, about dreams which everyone call to his mind when he hears the word, “phenomenon in the brain during sleeping”.’ To distinguish between Kanji characters and Kana characters, the former are underlined with solid lines and the latter are underlined with broken lines.
width of the characters was adjusted to be 0.7–1.0° and the width of the space between the characters was made to be 0.25–0.4° of visual angle. Thus, the distance between the heads of adjacent Kanji words was about 3° (3° texts). In set 2, the width of the characters was 1.2–1.6° and that of the space between the characters was 0.4–0.6°. Thus, the distance was made to be about 5° (5° texts) (Fig. 3-A). The texts were displayed on the screen in white characters against a black background (Fig. 3-B). They drifted leftward at the velocity of 7.15°/s for the text from set 1 and 11.9°/s for the text from set 2. The drifting speed was adjusted to compensate the horizontal extent of the texts with higher speed given to set 2 texts to compensate their longer horizontal extent. Thus, the information conveyed by the two sets was the same. The texts were shown on the screen horizontally and occupied 38° of visual angle from left to right. In the trials the text was presented over 12.8 s, which was replaced by a blank screen for 12.8 s. Participants were requested to read the texts silently in the first half of the trail and to watch the black screen in the latter half. This cycle was repeated 9 times. All texts were presented with random order in these trials. In this way, first 3° texts or 5° texts were presented then the other set of texts were presented. The presentation order of the sets was counterbalanced for the participants. It took more than 12.8 s to show the whole text. Thus, only one text could appear in a trial. Before the reading task, we instructed the participants to read the texts so that they comprehended them. During debriefing after the reading tasks, we checked their comprehension by asking them questions concerning the texts.
4.3.
fMRI scanning
4.3.1.
Data acquisition
The fMRI measurements were performed using a 3T scanner with an 8 channel phased-array coil (Siemens Trio, Erlangen, Germany). Blood oxygenation level-dependent (BOLD) responses were acquired using 1-shot gradient-echo echo-planar imaging with the following settings: 20 planes; TR/TE 1600/36 ms; flip
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
angle, 90 deg; voxel size, 2 × 2 × 3 mm; fov, 192 mm. We chose axial slices parallel to the AC–PC line including the almost entire occipital lobe, recorded 148 imaging volumes.
4.3.2.
Anatomical data acquisition
To classify the gray and white matter, we acquired wholebrain T1-weighted anatomical data (magnetization prepared rapid acquisition gradient-echo imaging sequence; 190 planes; TR/TE 9.7/4 ms; flip angle, 12°; voxel size, 1 × 1 × 1 mm; fov, 256 mm; sagittal sections). Each fMRI session included acquisition of both functional and anatomical data in the same measurement planes (20 planes; TR/TE 580/17 ms; flip angle, 55°; voxel size, 1 × 1 × 3 mm; acquisition order, interleave; averaging, 2; phase-partial Fourier, 7/8; fov, 192 mm).
4.4.
Data processing
4.4.1.
Anatomical processing
Gray and white matter regions were segmented from the anatomical MRI using custom software and hand-edited to minimize segmentation errors (Teo et al., 1997). The cortical surface was reconstructed at the white/gray matter border and rendered as a smoothed 3D surface (Wandell et al., 2000) (http://white.stanford.edu/software/).
4.4.2.
Functional processing
The first 4 time-frames of each functional run were discarded due to start up magnetization transients. Motion compensation for head movement within and between scans (Friston et al., 1994) was applied with Statistical Parametric Mapping (SPM) 2 software program (http://www.fil.ion.ucl.ac.uk/spm/ software/spm2/). The drift in the time series mean level was removed by high-pass temporal filtering (fN0.025 Hz). Data, which were not spatially smoothed, were analyzed using the mrVista software (Wandell et al., 2000) (Stanford, http://white. stanford.edu/software). Activations were visualized on the unfolded representation of the white–gray matter boundary. We visualized the activated regions on the flat map with the two localization tasks (Figs. 2-B, C). In this way we could identify the voxels representing areas of V1, their directions of eccentricity, horizontal meridians and the locations of 5° of eccentricity for both the left and the right occipital cortices, in each subject. Then, we visualized the activation areas during the reading tasks on flat map (Figs. 3-C, D). We determined a linear ROI area along the horizontal meridians, and calculated the mean activation amplitudes of the 9 trials for the ROI along the horizontal meridian. We determined the voxels that corresponded each eccentricity along the horizontal meridian from their distance from the 5° point using a formula developed by Engel et al. (1997). Thus, we could measure the amplitude of the activations at each eccentricity in each subject. As the representations of the fovea in the early visual cortices (V1, V2, V3, and hV4) converge on the neighborhood of the occipital pole, the boundaries of early cortices are intricate with each other and thus it is hard to separate them in this area (Dougherty et al., 2003). Therefore, we restrict this analysis to the parafoveal areas of 2.1–9.0°. To standardize the intersubject difference, the amplitude of each eccentricity was converted to Z score based on the averages and standard deviations of the amplitude of activated voxels of this region. Then, we calculated the mean Z
79
scores of 6 participants for each eccentricity, smoothed them with 1 + 2 + 1 filter, and graphically illustrated them. In addition to the average graph, we made graphs of the activation of each individual after smoothing the mean activation amplitudes of each eccentricity with 1 + 2 + 1 filter.
4.5.
Eye movement recording
In 2 of the participants, we recorded eye movements using an infrared-video eye-monitoring system (ST-661; NAC Image Technology, Tokyo, Japan) to verify whether the saccades landed on the expected boundary region during the reading of the text. The records were performed simultaneously with the acquisition of fMRI data. To decide whether a particular eye movement was really associated with reading, we set up the criterion of the eye movement magnitude of 1.5° of visual angle and the latency criterion of 100 ms. Eye movements that were either smaller than this magnitude or shorter than this latency were not included in the analysis, regarding them to be made for some other purposes, such as corrective eye movements.
Acknowledgments We are grateful to Ryusaku Hashimoto and Syoichi Iwasaki for their assistance and helpful comments. This study was supported by a Grant-in-Aid for Scientific Research (C) (19500439) of The Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan.
REFERENCES
Adler-Grinberg, D., Stark, L., 1978. Eye movements, scanpaths, and dyslexia. Am. J. Optom. Physiol. Opt. 55, 557–570. Anstis, S.M., 1974. A chart demonstrating variations in acuity with retinal position. Vision Res. 14, 589–592. Biscaldi, M., Gezeck, S., Stuhr, V., 1998. Poor saccadic control correlates with dyslexia. Neuropsychologia 36, 1189–1202. Dougherty, R.F., Koch, V.M., Brewer, A.A., Fischer, B., Modersitzki, J., Wandell, B.A., 2003. Visual field representations and locations of visual areas V1/2/3 in human visual cortex. J. Vis. 3, 586–598. Engel, S.A., Glover, G.H., Wandell, B.A., 1997. Retinotopic organization in human visual cortex and the spatial precision of functional MRI. Cereb. Cortex 7, 181–192. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.P., Heather, J.D., Frith, R.S.J., 1994. Statistical parametric maps in functional imaging: a general linear approach. Hum. Brain Mapp. 2, 189–210. Furuta, A., Nakadomari, S., Misaki, M., Miyauchi, S., Iida, T., 2009. Objective perimetry using functional magnetic resonance imaging in patients with visual field loss. Exp. Neurol. 217, 401–406. Honrubia, V., Downey, W.L., Mitchell, D.P., Ward, P.H., 1968. Experimental studies on optokinetic nystagmus. II. Normal humans. Acta Otolaryngol. 65, 441–448. Kajii, N., Nazir, T.A., Osaka, N., 2001. Eye movement control in reading unspaced text: the case of the Japanese script. Vision Res. 41, 2503–2510. Kashou, N.H., Leguire, L.E., Robert, C.J., Fogt, N., Smith, M.A., Rogers, G.L., 2010. Instruction dependent activation during optokinetic nystagmus (OKN) stimulation: An FMRI study at 3T. Brain Res. 1336, 10–21.
80
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
Leff, A.P., Scott, S.K., Crewes, H., Hodgson, T.L., Cowey, A., Howard, D., Wise, R.J., 2000. Impaired reading in patients with right hemianopia. Ann. Neurol. 47, 171–178. Martinez, A., Anllo-Vento, L., Sereno, M.I., Frank, L.R., Buxton, R.B., Dubowitz, D.J., Wong, E.C., Hinrichs, H., Heinze, H.J., Hillyard, S.A., 1999. Involvement of striate and extrastriate visual cortical areas in spatial attention. Nat. Neurosci. 2, 364–369. Miellet, S., O'Donnell, P.J., Sereno, S.C., 2009. Parafoveal magnification. Visual acuity does not modulate the perceptual span in reading. Psychol. Sci. 20, 721–728. McConkie, G.W., Kerr, P.W., Reddix, M.D., Zola, D., 1988. Eye movement control during reading. The location of initial eye fixations on words. Vis. Res. 28, 1107–1118. McConkie, G.W., Rayner, K., 1975. The span of effective stimuli during a fixation in reading. Prcept. Psychophys. 17, 578–586.
Pollatsek, A., Bolozky, S., Well, A.D., Rayner, K., 1981. Asymmetries in the perceptual span for Israeli readers. Brain Lang. 14, 174–180. Rayner, K., Bertera, J.H., 1979. Reading without a fovea. Science 206, 468–469. Teo, P.C., Sapiro, G., Wandell, B.A., 1997. Creating connected representations of cortical gray matter for functional MRI visualization. IEEE Trans. Med. Imaging 16, 852–863. Wandell, B.A., Chial, S., Backus, B.T., 2000. Visualization and measurement of the cortical surface. J. Cogn. Neurosci. 12, 739–752. Zihl, J., 1995. Eye movement patterns in hemianopic dyslexia. Brain 118, 891–912. Zihl, J., 2000. Rehabilitation of Visual Disorders After Brain Injury. Psychology Press, Hove.
available at www.sciencedirect.com
www.elsevier.com/locate/brainres
Research Report
Activation in left primary visual cortex representing parafoveal visual field during reading Japanese texts Yoichi Shimadaa, b,⁎, Kazumi Hirayamaa, c, 1, 2 , Satoshi Nakadomarid, e, 3, 4 , Ayumu Furuta f, g, 5, 6 , Masaya Misaki h, i, 7, 8 , Shigeyuki Kanh, i, 7, 8 , Takahiko Koikeh, i, 7, 8 , Satoru Miyauchih, i, 7, 8 , Etsuro Moria, 1 a
Department of Behavioral Neurology and Cognitive Neuroscience, Tohoku University Graduate School of Medicine, 2-1, Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan b Department of Rehabilitation, Faculty of medical science and welfare, Tohoku Bunka Gakuen University, 6-45-1, Kunimi, Aoba-ku, Sendai 981-8551, Japan c Department of Occupational Therapy, Yamagata Prefectural University of Health Sciences, 260, Kamiyanagi, Yamagata City, Yamagata 990-2212, Japan d Department of Medical Treatment (2), Hospital, National Rehabilitation Center for Persons with Disabilities, 4-1, Namiki, Tokorozawa, Saitama 359-8555, Japan e Department of Ophthalmology, Jikei University, School of Medicine, 3-25-8, Nishishinbashi, Minato-ku, Tokyo 105-8461, Japan f Maeda Ophthalmic Clinic, 3-30, Naka-machi, Aizu Wakamatu, Fukushima 965-0878, Japan g Department of Ophthalmology, Fukushima Medical University, 1, Mitsugaoka, Fukushima City, Fukushima 960-1295, Japan h Kobe Advanced ICT Research Center, National Institute of Information and Communications Technology (NICT), 588-2, Iwaoka-machi, Nishi-ku, Kobe 651-2492, Japan i Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency ( JST), 5, Sanban-machi, Thiyoda-ku, Tokyo 102-0075, Japan
A R T I C LE I N FO
AB S T R A C T
Article history:
Activation in the left primary visual cortex (V1) representing the parafoveal field during text
Accepted 16 June 2011
reading has been interpreted as attentional modulation in the process of deciding saccadic
Available online 24 June 2011
target for reading ahead. Kanji words serve the main cue to decide the goal of saccades in Japanese. We aimed to determine the exact location of this modulation in the V1 and to
Keywords:
determine whether the area of the modulation changes according to the location where the
Parafoveal activation
next Kanji word appears or it is fixed on a certain region in V1. Using functional magnetic
Reading saccade
resonance imaging, we determined the area in V1 representing each eccentricity on the
⁎ Corresponding author at: 6-45-1, Kunimi, Aoba-ku, Sendai 981-8551, Japan. Fax: + 81 22 233 4054, + 81 22 717 7360. E-mail addresses: [email protected] (Y. Shimada), [email protected] (K. Hirayama), [email protected] (S. Nakadomari), [email protected] (A. Furuta), [email protected] (M. Misaki), [email protected] (S. Kan), [email protected] (T. Koike), [email protected] (S. Miyauchi), [email protected] (E. Mori). 1 Fax: +81 22 717 7360. 2 Fax: +81 23 686 6674. 3 Fax: +81 4 2995 3102. 4 Fax: +81 3 3433 1111. 5 Fax: +81 242 29 3556. 6 Fax: +81 24 547 1989. 7 Fax: +81 78 969 2215. 8 Fax: +81 3 3222 2066. 0006-8993/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.brainres.2011.06.043
73
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
Attentional modulation
horizontal meridian of the visual field for each participant. Then we investigated brain
Primary visual cortex
activation while they were reading two sets of Japanese texts that scrolled leftward as the
Japanese text
participants. In set 1, the distance between the heads of adjacent Kanji words was about 3°.
Functional magnetic resonance
In set 2, the distance was about 5°. From the results of these experiments, we obtained
imaging
activation amplitude of the area corresponding to each eccentricity. We recorded eye movements simultaneously with the acquisition of fMRI data. The maximum peak of the activation was found in the region representing about 4.5° of eccentricity on the horizontal meridian in the left V1 for each participant. The activation pattern did not essentially differ between the two text conditions, although the location of the saccades made for reading next section of the text corresponds to the head of the next Kanji word. The activation modulation during reading Japanese texts occurs in the parafoveal V1 of the left hemisphere. The attentional modulation did not change with the distance to the next goal of saccade but was fixed on the area representing about 4.5° of eccentricity. © 2011 Elsevier B.V. All rights reserved.
1.
Introduction
In order to achieve smooth, efficient reading, one has not only to recognize each word embedded in a text but also to plan the eye movements to the next viewing point. As visual acuity decreases sharply with increasing eccentricity (Anstis, 1974), only the visual information from the foveal and parafoveal visual fields contributes to these process (Fig. 1). Word form is perceived in the foveal field extending 1° of visual angle either side of fixation (Rayner and Bertera, 1979), where the high spatial acuity enables one to discriminate constituent letters clearly. In the parafoveal visual field, one can extract only more gross features. However, the information from the parafoveal region can be enhanced by attentional modulations and thus contribute to reading process (Miellet et al., 2009). Perception of the ensuing words, to plan the next saccade, occurs in the right parafoveal field extending 5° for those who have left to right reading habit. Zihl (2000) named foveal and right parafoveal region on the horizontal meridians together (Fig. 1) as “perceptual window” for reading. When this parafoveal preview of upcoming text is not a valid extension of the current text, reading time is slowed. In contrast, invalid
Fig. 1 – Critical visual field for efficient reading and an example of a Japanese written sentence. Foveal vision is depicted in white and parafoveal vision in pale gray. The black boundary represents the critical area for efficient reading which is called perceptual window. †Kannji. *Hiragana.
information from text in the left parafovea does not affect reading time (McConkie and Rayner, 1975). This asymmetry reflects attentional demands linked to reading direction: In Hebrew (which is read from right to left) the interference by invalid information occurs in the left parafovea (Pollatsek et al., 1981). Patients with right homonymous hemianopia with macular sparing of less than 5° may show an impairment of reading. This is caused by a difficulty in acquiring information necessary for deciding the next saccadic target for reading (hemianopic dyslexia) (Zihl, 1995, 2000). Abnormal eye movement patterns during reading are also found in developmental dyslexics (Biscaldi et al., 1998). Thus, it is important to know the neural basis of saccades guided by the information acquired from peripheral vision in healthy subjects in order to understand the mechanisms of these impairments. Leff et al. (2000) investigated reading process of English word array with positron emission tomography (PET). They found a peak of activation in an area in the left calcarine cortex representing the right parafoveal visual field, and interpreted it as attentional modulation in the primary visual cortex (V1) in the process of deciding saccadic targets necessary for reading ahead. However, there remain some questions. The result might not reflect natural process of reading a text. The activation reported in their study was obtained by contrasting the viewing condition of reading meaningless strings of words and of viewing a single word. It is not clear what part of the parafoveal region of V1 was activated. They did not ascertain the retinotopic organization of V1 for each participant. It remains uncertain if exact locations of the activation found for each participant were comparable or not, as they just showed the activated locations averaged across the 5 participants. In the languages that use Roman script, selection of saccade targets is based on information about word boundaries, which is provided visually salient spaces inserted between words (McConkie et al., 1988). However, since there is no such spacing between words or between individual letters in the Japanese writings, space does not provide the information necessary to locate word boundaries. Typical modern Japanese texts are written primarily in a mixture of two basic scripts: Kanji, a script, and Hiragana, a syllabarie (Fig. 1). Usually, Kanji encodes grammatical categories such as nouns, verb stems and adjective stems. Hiragana serves to
74
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
represent grammatical element of sentences such as particles, auxiliary verbs, and inflectional affixes of noun, adjectives or verbs. However, words usually written in Kanji can also be expressed in Hiragana, as each Kanji word has its specific Japanese syllables. In general, Kanji characters are visually more complex. They also consist of more straight lines than Hiragana characters, most of which are composed of curved lines. Kajii et al. (2001) investigated eye movements in reading Japanese texts, and found that the eyes tend to land at the head of Kanji words. Thus, Kanji words that occur in the right parafoveal visual field seem to serve the main cue to decide the goal of saccades in reading a text of this language. Thus, an important research question is to localize the exact region of the “attentional modulation” in the parafoveal V1 while ones are reading Japanese texts. There are two possibilities. One possibility is that an area corresponding to the goal of next saccade where the next Kanji word is laid would be activated (goal-dependent modulation). Another possibility is that the activation would be found at a fixed location corresponding to somewhere in the perceptual window, for example the right edge of it (fixed-location modulation). This study aimed to test which of these two areas is activated more prominently in the healthy subjects while they were reading texts in Japanese. Using functional magnetic resonance imaging (fMRI), we investigated brain activation when healthy participants read meaningful Japanese sentences that scrolled leftward as the participants read them. We attempted to determine the exact location of activation in the parafoveal V1, for each participant, rather than to obtain the averaged location for the participants. Furthermore, we tried to find out whether the attentional modulation was dynamically controlled in accordance with the location of next saccade which corresponded to the head of next Kanji word or was fixed at some location in the perceptual window, by comparing fMRI data and eye movement record.
2.1° of eccentricity (Fig. 4-C). The activation during reading 5° texts showed a maximum peak on 5.5° and another peak on 2.8°, and showed a rise on 2.1° of eccentricity (Fig. 4-D). For amplitude of activation, we performed a three-way repeated measures ANOVA with text type (3 texts and 5 texts), hemisphere (left and right), and eccentricity (20 locations: 2.2, 2.3, 2.5, 2.7, 2.8, 3.0, 3.2, 3.4, 3.6, 3.9, 4.1, 4.4, 4.7, 5.0, 5.3, 5.7, 6.0, 6.4, 6.8, and 7.3°) as factors. The three-way ANOVA showed a significant main effect of hemisphere (F (1, 5) = 7.17, p < .05). The average of the all amplitude in left hemisphere was 1.55 and that of left hemisphere was 0.97. Thus, the activation of left parafoveal V1 was greater than that in right parafoveal V1. An interaction between text type and eccentricity was also significant (F (19, 95) = 2.48, p < .01). To follow up this interaction, a two-way repeated measures ANOVA was conducted separately for each hemisphere. In the left hemisphere, twoway ANOVA showed a significant interaction between text type and eccentricity (F (1, 19) = 2.41, p < .01). In the right hemisphere, there was no significant effect. Fig. 5 shows the activation amplitude of each participant along the horizontal meridian. As the rising activations toward the area representing 2.1° of eccentricity in the left and right hemispheres thought to reflect the margin of foveal V1, we sought the eccentricity of each participant which had the amplitude of largest local maximum other than these rising toward 2.1°. In left hemisphere the average of the eccentricities was 4.6° (range 4.1° to 4.7°) for 3° texts and 4.6° (range 3.9° to 5.3°) for 5° texts. In the right hemisphere the average of the eccentricities was 3.8° (range 2.2° to 7.3°) for 3° texts and 5.1° (range 2.2° to 7.3°) for 5° texts. Thus all participants had the largest peak somewhere in the restricted range close the eccentricity of 4.6°, and it did not change according to the distance to the head of the next Kanji word. In contrast, the largest peak of each participant was distributed among broad range.
2.
Results
2.2.
2.1.
fMRI data of text reading
Saccadic eye movement data obtained from two participants (subjects 1 and 2 of Fig. 5) showed that reading related saccades occurred actually from the head of a Kanji word to the next head of subsequent Kanji word (Fig. 6). In subject 1, the mean amplitude of the saccades during reading 3° texts was 3.22° ± 1.07 of visual angle while that of 5° texts was 5.14° ± 1.57. In subject 2, mean amplitude of the saccades during reading 3° texts was 2.96° ± 0.78 of visual angle while that of 5° texts was 4.27° ± 1.01. Thus, the location of the saccades made for reading next section of the text appeared to correspond to the head of the next Kanji word.
All of the participants answered sufficiently well to the questions concerning the texts. Thus, they seemed to understand the texts. We presented to the participants scrolling texts whose distance between the heads of adjacent Kanji words was about 3° (3° texts) and whose distance between the heads of adjacent Kanji words was about 5° (5° texts) (Fig. 3-A). If the attentional modulation was dynamically controlled in accordance with the location of next saccade which corresponded to the head of next Kanji word, the peaks of activation in the left V1 would shift according the distances. If it was fixed at some location in the perceptual window, the location would not shift according the distances. The mean Z score of activation amplitude on the horizontal meridians represented in the left V1 during reading 3° texts and reading 5° texts showed essentially similar patterns. Both of them had a maximum peak on 4.5° and other peaks on 2.7° and 6.7° of eccentricities, and showed a rise on 2.1° of eccentricity (Figs. 4-A, B). In the right V1, the activation during reading 3° texts showed no obvious peak, and showed a rise on
3.
Saccadic eye movement data of text reading
Discussion
Using fMRI, we studied activation in V1 during reading meaningful Japanese sentences. In the left V1 area representing the parafoveal horizontal meridian, the maximum peak of the activation was found in the region representing about 4.5° of eccentricity not only in the average data of all the participants but also in each participant. In contrast to the left V1, no common pattern of activation was found in the
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
75
Fig. 2 – Determination of horizontal meridian and location of the eccentricity of 5° in V1. (A) The dartboard like stimuli to determine the area in V1 representing horizontal meridian and location of the eccentricity of 5° along the meridian. These two sections were presented alternatively with a cycle of 25.6 s. The center of the dartboard, changed from red to white every 10 s for 0.5 s. (B) Response of a participant (participant 4 of Fig. 5) elicited during the viewing of these stimuli is shown on an inflated cortical surface of his left hemisphere. (C) The response was converted into a flat map. Black boundary represents the area V1 obtained beforehand, using the paradigm reported by Furuta et al., 2009. HM: Red lines represent horizontal meridian. Ec5°: the eccentricity of 5°. Fo: the area which represents fovea. (D) The color of the activated area reflects phase, i.e. timing (second) in the cycle of visual stimuli, as indicated by this color bar.
right V1. Thus, the findings of this study demonstrated that with the naturalistic material the activation occurred in the parafoveal V1 of the left hemisphere as was shown by Leff et al. (2000) with a somewhat artificial text of random word sequence. The greatest activated region was the right edge of the perceptual window throughout the participants. That focused attention can modulate activity in the visual cortex has been reported in many human fMRI studies (Martinez et al., 1999). In those studies visual stimuli were displayed in the neighborhood of the parafoveal area and the participants were instructed not to make eye movements. Leff et al. (2000) interpreted that the activation in the parafoveal V1 was caused by attention that was directed to the region to determine next saccadic target. The results of our experiments strongly support their interpretation. The activation pattern of each participant showed no essential difference between the two text conditions with different distance between the heads of adjacent Kanji words. In our experiment, the distances of Kanji words within a set of texts were constant. Thus, it might be expected that the peak activation is different with the longer distance condition having a peak at a further peripheral area in comparison with the one found for the shorter distance condition, if attention was directed to the next goal of reading saccade. However, the result showed that the maximum peak of activation was in the area representing about 4.5° of eccentricity in the left V1 even for the stimuli of texts of 3° distance. This was also true for the
two participants whose eye movements during reading were monitored. Their eye movements were found to be sensitive to the distance condition. The mean amplitude of the saccades during reading 3° texts was about 3°, while that of 5° texts was about 5° for each of them. Thus, the attentional modulation in V1 did not change with the distance to the next goal of saccade (goal-dependent modulation) but was fixed on the area representing the right edge of perceptual window (fixedlocation modulation). However, as the amplitudes were obtained as a total effect of 12.8-second text reading. The average location of activation during text reading is not adequate to test the effect of flexible ocular control on the neural activity in V1. It should be kept in mind that our result does not deny the presence of such a transient and dynamic modulations related to the goal of saccade. The reason why the modulation occurred at the edge of the right parafoveal region is uncertain. In parafoveal region, the acuity was lowest at its peripheral edge. Thus, elevating the sensitivity of this region would be helpful in getting information necessary for smooth and efficient control of reading. In the paper, we regarded the eye movement as voluntary saccade. However, the movement may be interpreted as a sort of optokinetic nystagmus (OKN). OKN is generally defined as involuntary rhythmic movements of the eyes caused by unidirectional motion of the visual field. However, Honrubia et al. (1968) classified OKN into two types; “look OKN” induced by active following of the targets without turning of the head,
76
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
Fig. 3 – Stimuli presentation and visualizing activation. (A) An example of the 3 texts. The distance between the heads of adjacent Kanji words was about 3° (top). An example of the 5 texts. The distance between the heads of adjacent Kanji words was about 5° (bottom). (B) In a trial, a drifting text was presented for 12.8 s (top), which was replaced by a blank screen for 12.8 s (bottom). (C) Response of a participant (participant 4 of Fig. 5) elicited during the reading task is shown on an inflated cortical surface of his left hemisphere. (D) The response was mapped onto flat map. Black boundary represents the area V1 obtained beforehand, using the paradigm reported by Furuta et al., 2009. Red lines represent horizontal meridian (HM) and the eccentricity of 5° (Ec5°), obtained beforehand, using the alternating dartboard like stimuli. A linear ROI along the horizontal meridian was determined. The voxels that corresponded to the eccentricity of 5° along the horizontal meridian as a node of these two lines. The amplitude of the activations at each voxel in the ROI was measured. Fo: the area which represents fovea. (E) The color of the activated area reflects phase, i.e. timing (second) in the cycle of visual stimuli, as indicated by this color bar.
Fig. 4 – Mean Z scores of activation amplitude of the 6 participants. (A) Response in the left V1 during reading the 3° texts. (B) Response in the left V1 during reading the 5° texts. (C) Response in the right V1 during reading the 3° texts. (D) Response in the right V1 during reading the 5° texts.
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
Fig. 5 – Activation amplitude of each participant. (A) Response in the left V1 during reading the 3° texts. (B) Response in the left V1 during reading the 5° texts. (C) Response in the right V1 during reading the 3° texts. (D) Response in the right V1 during reading the 5° texts. Italic numbers indicate each participant.
and “stare OKN” involuntary appearing when the subject inattentively directs gaze towards the moving patterns of targets. In our study, as the instruction to the subjects was to read and understand the stimulus texts, the induced eye movements should be corresponded to “look OKN”. Kashou et al. (2010) demonstrated the difference of activation on fMRI between two types of OKN by changing the instruction, and concluded that “look OKN” is somewhat of a misnomer and is indeed a voluntary series of pursuit and saccadic eye movements. Therefore, we think that the eye movements in our study are regarded as “look OKN” or saccade, although both are two aspects of the same phenomenon. There are two limitations in our experiment. First, the text was scrolled leftward as the participant. We chose this presentation in order to keep off leftward shifts of attention during reading on another line and keep the amount of the visual stimuli similar on either side of the fixation during the task. Thus, the stimuli used in our experiment were artificial in presentation method, although they were more natural in material comparing with the stimuli used by Leff et al. (2000). Reading the texts scrolling leftward may induce larger
77
Fig. 6 – Examples of eye movement during reading task. Arrows indicate horizontal locations of viewing point on a text. The change from 1 to 2 represents a pursuit for a Kanji meaning ‘one’ which is the head of Kanji word meaning ‘generally’. The change from 2 to 3 represents a saccade from a Kanji meaning ‘one’ to another Kanji meaning ‘same’. The change from 3 to 4 represents a pursuit for a Kanji meaning ‘same’. The change from 4 to 5 represents a saccade from a Kanji meaning ‘same’ to another Kanji meaning ‘occur’. Reading-related saccades occurred usually from the head of a Kanji word to the next head of Kanji word.
attention toward the right edge of the perceptual window comparing to the static texts. Thus, there was a possibility that the movement of the text enhanced the activation of the V1 area representing this portion of visual field. Second, the number of participants was small. To generalize our conclusion about the hemispheric difference of activation pattern it should be supported by statistical analyses over a larger group of participants. Finally, clinical implication of our finding is discussed. Leff et al. (2000) found that the activation patterns in the posterior parietal cortices and in the frontal eye fields were different between patients with hemianopic dyslexia and those of the healthy subjects, and interpreted that this difference was caused by a disorder in the control of visual attention. If in hemianopic dyslexic patients with a lesion in the optic radiation without damage in V1, the method employed in this study might show disordered modulation with inadequate activation peaks in V1. Abnormal eye movement patterns during reading have been reported also in developmental dyslexics (Adler-Grinberg and Stark, 1978). Biscaldi et al. (1998) reported that saccadic eye movements toward the visual stimuli displayed outside of the parafoveal vision were impaired in developmental dyslexics, also
78
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
impairment was correlated with the degree of reading disability, and maintained that reading process and saccade system are both controlled by visual attention systems, which may well be impaired in developmental dyslexics. Our method is readily applied to developmental dyslexics to examine how the peak locations are changed.
4.
Experimental procedure
4.1.
Participants
Six healthy males, aged 33 to 54 years took part. They were Japanese speaking, with normal vision and were right handed. All gave informed written consent. All procedures adhered to protocols based upon the world medical association declaration of Helsinki ethical principles for medical research involving human subjects, approved by the ethical committees of the National Institute of Information and Communications Technology (NICT), Japan.
4.2.
Visual tasks
4.2.1.
Localizing task
Before the reading task, we determined the area V1 and its direction of con-eccentric areas in each participant using the paradigm reported by Furuta et al. (2009) in which they used a stimulus of expanding ring and clockwise rotating wedges made of periodically contrast reversing checkerboard pattern. Then, we determined the horizontal meridian that extends to both the left and right directions encompassing the fixation point and the exact locations of the eccentricity of 5° along this meridian in each visual field for bilateral V1 of each participant. For these purposes, we prepared a circular figure with a radius of 16°, which was divided into 126 sectors by 18 radial lines and 7 concentric circles. Each sector was colored in black or white in alternating fashion to make a dartboard-like figure. Each sector of the dartboard reversed its color at 4 Hz. The dartboard was separated into two larger sections by its horizontal meridian and a circle of 5° of eccentricity as shown in Fig. 2-A. These two sections were presented alternatively with a cycle of 25.6 s and the cycle was repeated 18 times. To secure the fixation, participants were asked to push a button when the color of the fixation point, at the center of the dartboard, changed from red to white every 10 s for 0.5 s.
4.2.2.
Reading task
Nine plain passages taken from scientific essay of popular magazines were used as the reading materials. Fig. 7 shows an example of the texts. The passages consisted of 2 to 5 sentences with 111 to 142 characters. Each text was modified as follows. Kanji words and Hiragana words in the passage were rearranged to make an array of Kanji and Hiragana characters, in which the interval of the heads of Kanji words was about 4 to 6 characters ahead without deranging both grammatical appropriateness and the meaning of the original text. We prepared two sets of stimuli from these modified passages. The characters of both sets had identical font and height (1.56°). Each set was different only in the widths of the characters and space between the characters. In set 1, the
Fig. 7 – An example of the Japanese texts used in the reading task. It means, ‘Why do we sleep. We have not found even the answer of this question. It is not true that sleep has only a static role for rest. During sleeping, various phenomena occur in our brain. I will tell you above, about dreams which everyone call to his mind when he hears the word, “phenomenon in the brain during sleeping”.’ To distinguish between Kanji characters and Kana characters, the former are underlined with solid lines and the latter are underlined with broken lines.
width of the characters was adjusted to be 0.7–1.0° and the width of the space between the characters was made to be 0.25–0.4° of visual angle. Thus, the distance between the heads of adjacent Kanji words was about 3° (3° texts). In set 2, the width of the characters was 1.2–1.6° and that of the space between the characters was 0.4–0.6°. Thus, the distance was made to be about 5° (5° texts) (Fig. 3-A). The texts were displayed on the screen in white characters against a black background (Fig. 3-B). They drifted leftward at the velocity of 7.15°/s for the text from set 1 and 11.9°/s for the text from set 2. The drifting speed was adjusted to compensate the horizontal extent of the texts with higher speed given to set 2 texts to compensate their longer horizontal extent. Thus, the information conveyed by the two sets was the same. The texts were shown on the screen horizontally and occupied 38° of visual angle from left to right. In the trials the text was presented over 12.8 s, which was replaced by a blank screen for 12.8 s. Participants were requested to read the texts silently in the first half of the trail and to watch the black screen in the latter half. This cycle was repeated 9 times. All texts were presented with random order in these trials. In this way, first 3° texts or 5° texts were presented then the other set of texts were presented. The presentation order of the sets was counterbalanced for the participants. It took more than 12.8 s to show the whole text. Thus, only one text could appear in a trial. Before the reading task, we instructed the participants to read the texts so that they comprehended them. During debriefing after the reading tasks, we checked their comprehension by asking them questions concerning the texts.
4.3.
fMRI scanning
4.3.1.
Data acquisition
The fMRI measurements were performed using a 3T scanner with an 8 channel phased-array coil (Siemens Trio, Erlangen, Germany). Blood oxygenation level-dependent (BOLD) responses were acquired using 1-shot gradient-echo echo-planar imaging with the following settings: 20 planes; TR/TE 1600/36 ms; flip
BR A IN RE S E A RCH 1 4 08 ( 20 1 1 ) 7 2 – 8 0
angle, 90 deg; voxel size, 2 × 2 × 3 mm; fov, 192 mm. We chose axial slices parallel to the AC–PC line including the almost entire occipital lobe, recorded 148 imaging volumes.
4.3.2.
Anatomical data acquisition
To classify the gray and white matter, we acquired wholebrain T1-weighted anatomical data (magnetization prepared rapid acquisition gradient-echo imaging sequence; 190 planes; TR/TE 9.7/4 ms; flip angle, 12°; voxel size, 1 × 1 × 1 mm; fov, 256 mm; sagittal sections). Each fMRI session included acquisition of both functional and anatomical data in the same measurement planes (20 planes; TR/TE 580/17 ms; flip angle, 55°; voxel size, 1 × 1 × 3 mm; acquisition order, interleave; averaging, 2; phase-partial Fourier, 7/8; fov, 192 mm).
4.4.
Data processing
4.4.1.
Anatomical processing
Gray and white matter regions were segmented from the anatomical MRI using custom software and hand-edited to minimize segmentation errors (Teo et al., 1997). The cortical surface was reconstructed at the white/gray matter border and rendered as a smoothed 3D surface (Wandell et al., 2000) (http://white.stanford.edu/software/).
4.4.2.
Functional processing
The first 4 time-frames of each functional run were discarded due to start up magnetization transients. Motion compensation for head movement within and between scans (Friston et al., 1994) was applied with Statistical Parametric Mapping (SPM) 2 software program (http://www.fil.ion.ucl.ac.uk/spm/ software/spm2/). The drift in the time series mean level was removed by high-pass temporal filtering (fN0.025 Hz). Data, which were not spatially smoothed, were analyzed using the mrVista software (Wandell et al., 2000) (Stanford, http://white. stanford.edu/software). Activations were visualized on the unfolded representation of the white–gray matter boundary. We visualized the activated regions on the flat map with the two localization tasks (Figs. 2-B, C). In this way we could identify the voxels representing areas of V1, their directions of eccentricity, horizontal meridians and the locations of 5° of eccentricity for both the left and the right occipital cortices, in each subject. Then, we visualized the activation areas during the reading tasks on flat map (Figs. 3-C, D). We determined a linear ROI area along the horizontal meridians, and calculated the mean activation amplitudes of the 9 trials for the ROI along the horizontal meridian. We determined the voxels that corresponded each eccentricity along the horizontal meridian from their distance from the 5° point using a formula developed by Engel et al. (1997). Thus, we could measure the amplitude of the activations at each eccentricity in each subject. As the representations of the fovea in the early visual cortices (V1, V2, V3, and hV4) converge on the neighborhood of the occipital pole, the boundaries of early cortices are intricate with each other and thus it is hard to separate them in this area (Dougherty et al., 2003). Therefore, we restrict this analysis to the parafoveal areas of 2.1–9.0°. To standardize the intersubject difference, the amplitude of each eccentricity was converted to Z score based on the averages and standard deviations of the amplitude of activated voxels of this region. Then, we calculated the mean Z
79
scores of 6 participants for each eccentricity, smoothed them with 1 + 2 + 1 filter, and graphically illustrated them. In addition to the average graph, we made graphs of the activation of each individual after smoothing the mean activation amplitudes of each eccentricity with 1 + 2 + 1 filter.
4.5.
Eye movement recording
In 2 of the participants, we recorded eye movements using an infrared-video eye-monitoring system (ST-661; NAC Image Technology, Tokyo, Japan) to verify whether the saccades landed on the expected boundary region during the reading of the text. The records were performed simultaneously with the acquisition of fMRI data. To decide whether a particular eye movement was really associated with reading, we set up the criterion of the eye movement magnitude of 1.5° of visual angle and the latency criterion of 100 ms. Eye movements that were either smaller than this magnitude or shorter than this latency were not included in the analysis, regarding them to be made for some other purposes, such as corrective eye movements.
Acknowledgments We are grateful to Ryusaku Hashimoto and Syoichi Iwasaki for their assistance and helpful comments. This study was supported by a Grant-in-Aid for Scientific Research (C) (19500439) of The Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan.
REFERENCES
Adler-Grinberg, D., Stark, L., 1978. Eye movements, scanpaths, and dyslexia. Am. J. Optom. Physiol. Opt. 55, 557–570. Anstis, S.M., 1974. A chart demonstrating variations in acuity with retinal position. Vision Res. 14, 589–592. Biscaldi, M., Gezeck, S., Stuhr, V., 1998. Poor saccadic control correlates with dyslexia. Neuropsychologia 36, 1189–1202. Dougherty, R.F., Koch, V.M., Brewer, A.A., Fischer, B., Modersitzki, J., Wandell, B.A., 2003. Visual field representations and locations of visual areas V1/2/3 in human visual cortex. J. Vis. 3, 586–598. Engel, S.A., Glover, G.H., Wandell, B.A., 1997. Retinotopic organization in human visual cortex and the spatial precision of functional MRI. Cereb. Cortex 7, 181–192. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.P., Heather, J.D., Frith, R.S.J., 1994. Statistical parametric maps in functional imaging: a general linear approach. Hum. Brain Mapp. 2, 189–210. Furuta, A., Nakadomari, S., Misaki, M., Miyauchi, S., Iida, T., 2009. Objective perimetry using functional magnetic resonance imaging in patients with visual field loss. Exp. Neurol. 217, 401–406. Honrubia, V., Downey, W.L., Mitchell, D.P., Ward, P.H., 1968. Experimental studies on optokinetic nystagmus. II. Normal humans. Acta Otolaryngol. 65, 441–448. Kajii, N., Nazir, T.A., Osaka, N., 2001. Eye movement control in reading unspaced text: the case of the Japanese script. Vision Res. 41, 2503–2510. Kashou, N.H., Leguire, L.E., Robert, C.J., Fogt, N., Smith, M.A., Rogers, G.L., 2010. Instruction dependent activation during optokinetic nystagmus (OKN) stimulation: An FMRI study at 3T. Brain Res. 1336, 10–21.
80
BR A IN RE S EA RCH 1 4 08 ( 20 1 1 ) 7 2 – 80
Leff, A.P., Scott, S.K., Crewes, H., Hodgson, T.L., Cowey, A., Howard, D., Wise, R.J., 2000. Impaired reading in patients with right hemianopia. Ann. Neurol. 47, 171–178. Martinez, A., Anllo-Vento, L., Sereno, M.I., Frank, L.R., Buxton, R.B., Dubowitz, D.J., Wong, E.C., Hinrichs, H., Heinze, H.J., Hillyard, S.A., 1999. Involvement of striate and extrastriate visual cortical areas in spatial attention. Nat. Neurosci. 2, 364–369. Miellet, S., O'Donnell, P.J., Sereno, S.C., 2009. Parafoveal magnification. Visual acuity does not modulate the perceptual span in reading. Psychol. Sci. 20, 721–728. McConkie, G.W., Kerr, P.W., Reddix, M.D., Zola, D., 1988. Eye movement control during reading. The location of initial eye fixations on words. Vis. Res. 28, 1107–1118. McConkie, G.W., Rayner, K., 1975. The span of effective stimuli during a fixation in reading. Prcept. Psychophys. 17, 578–586.
Pollatsek, A., Bolozky, S., Well, A.D., Rayner, K., 1981. Asymmetries in the perceptual span for Israeli readers. Brain Lang. 14, 174–180. Rayner, K., Bertera, J.H., 1979. Reading without a fovea. Science 206, 468–469. Teo, P.C., Sapiro, G., Wandell, B.A., 1997. Creating connected representations of cortical gray matter for functional MRI visualization. IEEE Trans. Med. Imaging 16, 852–863. Wandell, B.A., Chial, S., Backus, B.T., 2000. Visualization and measurement of the cortical surface. J. Cogn. Neurosci. 12, 739–752. Zihl, J., 1995. Eye movement patterns in hemianopic dyslexia. Brain 118, 891–912. Zihl, J., 2000. Rehabilitation of Visual Disorders After Brain Injury. Psychology Press, Hove.