Sensory and Perceptual Processing in Reading Disability

Applications of Parallel Processing in Vision I. B r m a n (Editor) 0 1992 Elsevier Science Publishers B.V. All rights reserved

263

Sensory and Perceptual Processing in Reading Disability MARY C. WILLIAMS and WILLIAM LOVEGROVE

Introduction Specific reading disability is a broad term which encompasses A reading disabilities arising from a number of sources. specific-reading-disabled child (SRD)is defined here as one of normal or better intelligence with no known behavioral or organic disorders who, despite normal schooling and average progress in other subjects, has a reading disability of at least 2.5 years (Badcock and Lovegrove, 1981; Critchely, 1964; Lovegrove et al., 1978. 1986; Slaghuis and Lovegrove, 1984; Stanley, 1975). Since reading involves a dynamic visual processing task that requires the analysis and Integration of visual pattern information across fixation-saccade sequences, studies in the area of reading disability have explored the possibility that visual processing abnormalities contribute to reading difficulties. A number of studies have provided evidence for basic visual processing differences between normal and disabled readers, especially a t early states of visual processing. Differences have been reported in visual information store duration (Lovegrove and Brown, 1978; Stanley, 1975; Stanley and Hall, 1973a). in the rate of transfer of information from visual information store to short term memory (Lovegrove and Brown, 1978; Stanley and Hall, 1973a). and in the characteristics of visual short term memory itself (Stanley and Hall, 1973b). These results indicate that some disabled readers process information more slowly and have a more limited processing capacity than normal readers. Studies that used tasks relying less on dynamic visual processing and temporal resolution, and more on pattern-formation processes and long-term visual memory, however, have failed to show visual processing differences between normal and disabled readers (Benton, 1962, 1975; Vellutino, 1977, 1979a. 1979b. 1987; Vellutino et al.. 1975a. 1975b). although the validity of these studies has been called to question (Fletcher and Satz, 1979a, 1979b). Thus the long-standing debate as to whether visual factors play a significant role in reading disabilities has been complicated by the differences in methodological factors and the failure to distinguish between the measurement of temporal versus pattern-formation processes.

264

CHAPTER 8

Two processing systems It has been suggested that the processing of temporal and pattern information is accomplished by two separate but interactive subsystems in vision with different spatiotemporal response characteristics (Kulikowski and Tolhurst, 1973: Tolhurst. 1973). Psychophysical work on visual spatiotemporal processing channels has indicated that low spatial frequency channels may elicit faster visual responses than high spatial frequency channels and are more sensitive to temporally modulated or moving stimuli, while high spatial frequency channels seem to be best designed for the detection of stationary patterns and resolution of fine pattern detail (Breitmeyer, 1975: Breitmeyer and Ganz, 1977; Breitmeyer et al., 1981: Tolhurst. 1975: Vassilev and Mitov, 1976; Watson and Nachmias. 1977). Additionally, for a given spatial frequency stimulus, response is faster and more transient when motion is being detected than when pattern is being detected (Kulikowski and Tolhurst. 1973; Watson and Nachmias, 1977). This conception of separable motion and pattern processing systems has been extensively studied and elaborated over the past two decades (Breitmeyer and Ganz, 1976: Maunsell, 1987; Stone et al., 1979), and has been incorporated in what has come to be known as the transient/sustained theory of visual perception. Breitmeyer and Ganz (1976) and Weisstein. Ozog and Szoc (1975). among others, have proposed two separate but overlapping subsystems in the visual system that respond selectively to different spatial and temporal frequencies. The transient system is most sensitive to low spatial frequencies, has a high temporal resolution, and responds transiently to quickly moving targets and to stimulus on- and offsets. The sustained system is most sensitive to high spatial frequencies, has a long response persistence and low temporal resolution, and responds in a sustained fashion to stationary or slowly moving targets. This dual processing system has recently been reconceptualized in terms of the magnocellular and parvocellular systems of the primate visual system, differing in color, acuity, speed, and contrast sensitivity (Livingstone and Hubel, 1987. 1988). The magnocellular and parvocellular systems are closely analogous to the previously proposed transient and sustained systems, respectively. [ I t should be noted t h a t the validity of the transient/sustained distinction in human vision has been questioned (Burbeck, 1981; Lennie. 1980). While this distinction is still quite schematic and requires additional work for confirmation, it has provided a useful organizing hypothesis in the study of reading disability, and has been found to be a remarkable predictor of the visual processing characteristics of the reading disabled. The usefulness of the transient/sustained analysis as an organizing tool thus seems to warrant its continued use at the present time.] It has also been suggested that these subsystems contribute to different aspects of vision and have different perceptual functions. The transient system is thought to be involved in the perception of motion and depth, brightness discrimination, the control of eye movements, and the localization of targets in space, and seems to function to accomplish a quick global analysis of a visual scene. The sustained system seems to be best designed for the identification of patterns,

READING DISABILITY

265

resolution of fine detail, and the perception of color (Breitmeyer, 1984; Breitmeyer and Ganz, 1976;Weisstein et al.. 1975: Livingstone and Hubel. 1988). Although these two subsystems operate in parallel. it is believed that the transient system has temporal precedence: it operates preattentively and functions as an early warning system. I t performs a global analysis of the incoming stimulus, parsing the field into units and regions and coding the position and movement of objects in space. The transient system may function to direct the sustained system to particularly salient areas where it might be most efficacious to perform a more detailed analysis of the shape and color of objects. The functioning of the sustained system, then, would depend to a degree on the prior output of the transient system.

Transient and sustained channels and reading It has been demonstrated physiologically (Singer and Bedworth. 1973) and psychophysically that transient and sustained systems may mutually inhibit each other (Breitmeyer and Ganz, 1976). In particular. if the sustained system is responding when the transient system is stimulated, the transient activity can terminate the sustained activity. These two subsystems and the interactions between them may serve a number of functions essential to the reading process. When reading, the eyes move through a series of rapid eye movements called saccades. Saccades are separated by fixation intervals lasting 200-250msec. It is during these stationary periods that information from the printed page is seen. The average saccade length is 6-8characters, or about 2 degrees of visual angle (Rayner and McConkie, 1976). Saccadic eye movements function to bring unidentified regions of text into foveal vision for detailed analysis during fixations. Foveal vision is the area of high acuity in the center of vision extending approximately 2 degrees (6-8letters) around the fixation point on a line of text. Beyond this, foveal acuity drops off rather dramatically. The role of transient and sustained subsystems in reading has recently been considered by Breitmeyer (Breitmeyer. 1980, 1983; Breitmeyer and Ganz, 1976). Figure 1 represents the hypothetical activity in the transient and sustained channels over a sequence of 3 fixations of 250 msec duration separated by 2 saccades of 25 msec duration. The sustained channel response occurs during fixations and may last for several hundred milliseconds. This response provides the details of what is being seen. The transient channel response is initiated by eye movements and lasts for much shorter durations. Consequently both systems are involved in reading. The duration of the sustained response may outlast the physical duration of the stimulus. This is one form of visible persistence produced by the activation of sustained channels. The duration of visible persistence can reach several hundred milliseconds and increases with increasing spatial frequency (Bowling et al.. 1979;Meyer and Maguire. 1977). If sustained activity (as shown in Figure 1, panel 2)generated in a preceding fixation persists into the succeeding one, it would interfere with processing in the second fixation. Consequently, it is evident that

266

CHAPTER 8

for tasks such as reading, persistence across saccades presents a problem a s it may lead to superimposition of successive inputs. Breitmeyer proposes that the problem posed by visible persistence is solved by rapid saccades, as shown in the bottom two panels of Figure 1.

-I

-I -I

Figure 1. A hypothetical response sequence of sustained and transient

channels during 3 250 msec fixation intervals separated by 25 msec saccades (panel 1). Panel 2 illustrates persistence of sustained channels acting as a forward mask from preceding to succeeding fixation intervals. Panel 3 shows the activation of transient channels shortly after each saccade which exerts inhibition (arrows with minus signs) on the trailing, persisting sustained activity generated in prior fixation intervals. Panel 4 shows the resultant sustained channel response after the effects of the transient-on-sustained inhibition have been taken into account. Saccades not only change visual fixations, they also activate short latency transient channels (panel 3). which are very sensitive to stimulus movement. This transient activity, in turn, inhibits the sustained activity persisting from a previous fixation and prevents it from interfering with the succeeding one (Breitmeyer and Ganz. 1976; Matin, 1974). The result is a series of clear, unmasked, and temporally

READING DISABILITY

267

segregated frames of sustained activity, each one of which represents the pattern information contained in a single fixation (Figure 1, panel 4).

In these terms, clear vision on each fixation results from the interactions between sustained and transient channels. The two subsystems and the interactions between them, therefore, seem to be important in facilitating normal reading. A deficit in either the transient or the sustained system, or in their interaction, may have harmful consequences for reading.

Transient and sustained channels and reading disability There is evidence that this transient-sustained relationship is different in normal and disabled readers. Lovegrove and coworkers have shown that visual processing differences between normal and disabled readers are evident when transient system processing is involved, but fail to surface under sustained processing conditions. For example, several studies have compared SRDs and controls on measures of visible persistence. Visible persistence is one measure of temporal processing in spatial frequency channels and refers to the continued perception of a stimulus after it has been physically removed. Visible persistence is assumed to reflect on going neural activity initiated by the stimulus presentation. In adults, the duration of visible persistence increases with increasing spatial frequency (Bowling et al.. 1979: Meyer and Maguire. 1977). In a series of experiments, Babcock and Lovegrove (1981) and Slaghuis and Lovegrove [ 1985) have compared visible persistence in SRD and normal readers aged 8 to 15 years. The duration of visible persistence was determined by measuring the temporal separation required for the detection of a blank interval between two successively presented gratings. In normal readers, the duration of visible persistence varied as a function of the spatial frequency of the test grating. Lower spatial frequency stimuli produced shorter visible persistence durations. Visible persistence increased monotonically with increasing spatial frequency (Figure 2). The SRDs had a significantly smaller increase in visible persistence duration with increasing spatial frequency than did controls. In the SRD group, visible persistence was longer for low spatial frequencies and shorter for the higher spatial frequencies as compared with controls (Figure 2). The slope of the function relating visible persistence to spatial frequency was significantly flatter in SRDs than in normal readers (Lovegrove et al.. 1980). The slope was 14.9 for normals but only 4.8 for SRDs. A normal sloping function can be explained in terms of transient-on-sustained inhibition. At low spatial frequencies, transient inhibition is relatively strong. At the offset of the stimulus, a transient system response inhibits the prolonged activity of the sustained channels. Because of this action. visible persistence is relatively short. As spatial frequency increases, transient system activity decreases, producing less sustained system inhibition. At high spatial frequencies, sustained channel activity persists longer after stimulus offset, resulting in an increase in visible persistence duration. This difference in visible persistence as a function of spatial frequency between controls and SRDs can be explained by suggesting a

268

CHAPTER 8

disparate type of transient-sustained interaction present in those with specific reading disability. In this group, the existence of a transient system deficit would elevate visible persistence at low spatial frequencies by creating deficient transient-on-sustained inhibition. At higher spatial frequencies, if SRDs do have a weak transient system, their sustained systems are "disinhibited" from the normal tonic transient-on-sustained inhibition compared to controls (Lovegrove. Martin, and Slaghuis, 1986). This would increase the activity in their sustained systems and produce shorter persistence durations. This is argued to be the case because a manipulation known to reduce transient system activity - uniform field flicker masking (Breitmeyer et al.. 1981)has a much greater effect on visible persistence in controls than in SRDs. Furthermore, uniform field flicker masking reduces the persistence differences between the two groups.

h

0

%

E

v

W

0

z W

trw K W

n

SPATIAL FREQUENCY (cldeg)

Duration of visible persistence as a function of spatial frequency for controls and reading disabled.

Figure 2.

Other evidence for a transient system deficit in specific reading disability has been advanced by the study of contrast sensitivity. Contrast sensitivity is the minimum amount of contrast needed to perceive a grating pattern. Sensitivity is greatest for patterns of intermediate frequency and decreases for patterns that are of lower or

READING DISABILITY

269

higher spatial frequency. Contrast sensitivity plotted as a function of stimulus spatial frequency is referred to as the contrast sensitivity function (CSF).

4.24.0 \

3.8 - + 3.6 3.4 3.2 3.0 2.8 2.6

-

\ \

2.4 -

\ \ \ \

2.2 -

\ \ \

\

2.0 -

1

c

1.8

+

I

0

f 2

b

Reading Disabled Controls I

I

4

8

I

12

SPATIAL FREQUENCY (ddeg) Figure 3. Contrast sensitivity function for controls and reading disabled. CSFs have been measured in five separate samples of disabled and normal readers aged 8-14 years of age (Lovegrove et al., 1980, 1982; Martin and Lovegrove. 1984). SRDs showed a consistent pattern of lower sensitivity to low spatial frequencies (1-4 c/deg) than did

270

CHAPTER

8

controls (Figure 3). The pattern of differences to the high spatial frequencies (12-16 c/deg) is less exact. In some studies, the two groups did not differ in the CSF and in others the SRDs were slightly more sensitive than controls in that range. The differences between the groups were greatest with stimulus durations ranging from 150-500 msec. I t should be noted that the magnitude of the differences between the groups on measures of pattern CSF are not as great as those found on measures of visible persistence (see Lovegrove. Martin, and Slaghuis. 1986). The finding of a small but consistent sensitivity loss a t low spatial frequencies in SRDs is consistent with the proposal of a transient system deficit a s argued by Lovegrove et al. (1982). The pattern CSF data are consistent with the visible persistence data reported above. SRDs are less sensitive than controls at low spatial frequencies but not a t high spatial frequencies, where they are sometimes more sensitive. As the pattern CSF experiments used a two-alternative temporal forced choice procedure, the differences between the groups are unlikely to result from criterion differences. The finding that SRDs are at least as sensitive as controls to high spatial frequencies is also consistent with the general finding that SRDs have normal or correctable-to-normal acuity. This finding also further explains why some studies have, and some have not, found differences in visual processing between the two groups. The presence or absence of visual processing differences should reflect the channel whose activity has been measured. Transient system functioning can be investigated more directly by the measurement of flicker contrast sensitivity where, instead of a static display of a stimulus, a test grating is counterphased. It has been argued that flicker thresholds are mediated by the transient system (Kulikowski and Tolhurst, 1973). A transient system deficit, then, should result in decreased sensitivity to flicker in SRDs. and this decrement should increase as temporal frequency increases. In these experiments, 13-year-old subjects detected a 2 c/deg sine wave grating that counterphased a t 5, 10, 15. 20, and 25 Hz. On the average, controls were found to be more sensitive than SRDs across the range of temporal frequencies tested (Figure 4; Martin and Lovegrove. 1987). The sensitivity difference between the two groups increased with increasing temporal frequency. Similar results were obtained by Brannan and Williams (1988a) using uniform field flicker. These results add further to the argument that a difference in transient system function exists between SRDs and normal readers. In a second experiment, the flicker contrast sensitivity function was determined for the same two groups using spatial frequencies from 1-12 c/deg counterphasing a t 20 Hz (Martin and Lovegrove, 1987). These results also indicated that the controls were more sensitive than SRDs across all spatial frequencies, the differences being larger at the higher spatial frequencies. A further series of experiments has been conducted comparing sustained system processing in controls and SRDs (Lovegrove et al.. 1986). Using similar procedures, equipment, and subjects as the experiments outlined above, this series h a s failed to show any significant differences between the two groups in orientation or spatial frequency tuning. This implies that either there are no differences

READING DISABILITY

27 1

between the groups in the functioning of their sustained systems, or that such differences are small compared to the transient system differences demonstrated,

4.1 4.0

3.9

>.

t 2

3.8 3.7

v)

z

v)

3.6

I-

2K

3.5

z 0

3.4

I-

\ \

0

f

\ \ \

3.3

\ \

\

\

3.2

+ 0

3.1

\

Reading Disabled Controls

\ \ \

6

T I

I

I

I

10

15

I

5

20

25

TEMPORAL FREQUENCY (Hz) Figure 4. Flicker contrast sensitivity for a counterphasing 2 c/deg grating as a function of temporal frequency for controls and reading disabled. In summary, four converging lines of evidence suggest a transient system deficit in SRDs. The results are internally consistent and consistent with the proposal of a transient system deficit. The differences between the groups are quite large and discriminate well between individuals In the different groups, with approximately 75% of SRDs showing reduced transient system sensitivity (Slaghuis and Lovegrove, 1985). At the same time, evidence to date suggests that the two groups do not differ in sustained system functioning. The two

272

CHAPTER 8

findings taken together may help to explain some of the confusion reported in the literature over many years. In these terms whether or not differences are found will depend on which system is investigated.

Perceptual consequences of a transient deficit The studies reported above demonstrate that a large subgroup of disabled readers do have visual deficits. The visual deficits are specific spatiotemporal processing abnormalities, are systematic, and occur early in the visual processing hierarchy: that is, in transient system operations. Given what is known of the perceptual functions that the transient system performs, it is reasonable to expect that children with a reading disability stemming from transient malfunctions would show deficits in global, preattentive processing operations and on tasks requiring fine temporal resolution. A number of studies have expanded on Lovegrove's work by studying the perceptual consequences of a transient deficit in reading disabled children. These studies employed subject populations consisting of children reading a t least one year below grade level ("disabled readers"), and children reading a t or above grade level ("normal readers"). Since normal performance on standardized reading tests includes a range of scores +/- one standard deviation from the mean. this classification criterion would theoretically designate 84% of a normally distributed population as normal readers, and 16% as disabled readers, which is consistent with most estimates of the prevalence of reading disability in the general population (e.g.. Critchely. 1964). All children in these studies were aged 8-12 years, were of normal or above normal intelligence, had normal color vision and normal or corrected-to-normal visual acuity, and scored within the normal range on tests of auditory discrimination. The normal and disabled reader groups were matched for age and IQ. These studies have demonstrated the existence of a number of perceptual deficits in disabled readers that would be predicted by a transient deficit hypothesis, t e . . the perceptual skills affected are those that are most likely to be mediated by the transient system. Williams and Bologna (1985) found that disabled readers show stronger perceptual grouping effects than normal readers, and are less proficient a t selective attention operations. Given that perceptual grouping and selective attention operations have been linked to the activity of transient visual channels (Williams and Weisstein. 1980). these findings suggest that the suspected transient deficit in disabled readers may be producing perceptual deficits in perceptual grouping and selective attention operations. There is evidence that this inclination towards global processing in disabled readers is a consequence of a sluggish response from the transient system. Brannan and Williams (1987) found that disabled readers were not able to utilize information provided by a cue to target location in a target detection task if the cue preceded the target by less than 50 msec.. whereas normal readers could utilize such information a t shorter temporal separations. Likewise, it has been shown that disabled readers require more time than either normal readers or adults to make accurate judgements when asked to specify which of two simple words were presented first (Brannan and Williams, 1988b; May,

READING DISABILITY

-

-

2 73

CLEAR IMAGE ARRAYS 20,

0 18-

v, 16-14--

w

I I-

I 0

5

E

121086--

42 -0

0

3

10

15

20

POSITION OF TARGET IN ARRAY

BLURRED IMAGE ARRAYS

I-

I

10-

0

8-

4

6--

[1L

0

5

10

15

20

POSITION OF TARGET IN ARRAY Figure 5 . Search time in seconds is plotted as a function of the target's

position in the array for adults (triangles). normal readers (circles). and disabled readers (diamonds). (a) (top panel) Clear image arrays. (b) (bottom panel) Blurred image arrays.

274

CHAPTER 8

Williams, and Dunlap, 1988). Given that the transient system is a likely candidate for the mediation of temporal order, these results suggest that disabled readers may have a slow or sluggish transient system. Transient operations may then consume too much of available processing capacity in disabled readers, leading to a n over-restriction to global processing operations, and difficulty in proceeding to detail processing operations.

Beyond a unitary deficit explanation Although there is considerable evidence that the visual processing of disabled readers is characterized by poor temporal processing, which surfaces in what have been conceptualized a s transient processing operations, it is misleading to consider the visual processing abnormalities under a unitary transient deficit hypothesis. For example, a number of studies have found that the functioning of sustained pattern formation processes are affected by the poor temporal processing of disabled readers. Williams, Brannan and Lartigue (1987) investigated detail processing operations in normal and disabled readers using a visual search task, where subjects were required to search for a target letter embedded in a list of distractor letters. This task requires a scrutiny of local differences between target and distractor items. They found that search times were much longer in disabled readers than in either normal readers or adults (Figure 5a). This difference was diminished, however, when high spatial frequencies (above 15 cycles/degree) were removed from the display by image blurring (Figure 5b). This method of spatial frequency filtering produces 100% contrast reduction a t spatial frequencies above 15 cycles/degree. and reduces the contrast of the remaining high spatial frequencies as well. Such contrast reduction would decrease the amplitude of the high spatial frequency component of visual response, and as a result, it is quite plausible that the latency and/or rise time would be increased as well. If high spatial frequency channels mediate local processing operations, then this manipulation may weaken and/or slow the temporal development of local information. This would reestablish the temporal precedence of global information, and simulate a normal relationship between global and local operations. Thus the over-restriction to global processing and consequential difficulty in the progression to detail processing produced by a sluggish transient system in disabled readers may be overridden by manipulating the relative timing of low and high spatial frequency information. Although the primary visual deficit of disabled readers seems to involve temporal, or transient, processing operations, the sustained pattern formation processes are necessarily affected through a different pattern of interactions over time.

Visual masking studies More direct measures of the time course of visual processing in normal and disabled readers have been obtained in recent visual masking studies (Williams et al.. 1989. 1990a). In visual masking, two temporally separated visual stimuli are presented, and one stimulus, called the mask, interferes with the processing of the other stimulus,

275

READING DISABILITY

ADULTS

0-0

A-A

FOVEAL PERIPHERAL

5t

-240 -180 -120 -60

0

60

120

180

2

DELAY (MSEC) 25

l5

t

I NORMAL READERS

0-0

A-A

FOVEAL PERIPHERAL

54

T

I 1

I

-240 -180 -120 -60

I

I

0

60

DELAY (MSEC)

I

120

I I

180

240

276 n

R

w

W

CHAPTER 8 25

I DISABLED READERS

-25-1 -240 -180 I I

I

-120

I

I

-60

a-0 A-A

I

I

0

60

I

FOVEAL PERIPHERAL

I

120

I I

180

2 LO

D E LAY (M S EC) Figure 6. Masking functions obtained from adult subjects (a), (top panel, preceding page): normal readers (b), (bottom panel, preceding page); and disabled readers (c) under foveal (circles) and peripheral (triangles) viewing. Accuracy (measured a s percent correct) for detecting the target lines when preceded or followed by the masking stimulus at various delays is plotted relative to target lines-alone accuracy level. Negative delays indicate that the mask preceded the target (forward masking), and positive delays indicate that the mask followed the target (backward masking). called the target. Williams et al. (1990a)employed a masking of pattern by light paradigm to measure visual integration and persistence characteristics of normal and disabled readers. Masking of pattern by light is a special case of visual masking where the visibility of the target pattern is reduced by a spatially uniform luminance mask flash that overlaps the target. I t is assumed that target visibility is degraded to the extent that the sensory activity of the target and mask persist and overlap in time or are integrated during a brief temporal interval. Thus masking by light constitutes a measure of the temporal resolution limits of the visual system imposed by either response persistence or response integration. Disabled readers showed more prolonged masking as compared with normal subjects (Figure 6a, b. c), suggesting that visual processing is characterized by a longer integration time and/or longer visual persistence. Disabled readers also showed enhancement effects rather than masking effects when stimuli were presented in the peripheral retina, suggesting that peripheral visual processing is characterized by a disinhibition or enhancement of sustained pattern information due to a diminished inhibitory effect

READING DISABILITY

277

imposed by peripheral transient channels. This disinhibition effect provides additional evidence for the proposal that sustained pattern formation processes are affected by the temporal processing characteristics of disabled readers through an abnormal pattern of interactions over time. a

b.

20

40

60

80

Delay lmrecl

Figure 7 . A schematic U-shaped metacontrast function together with hypothetical visual responses to a target and to a mask. (a) Schematic U-shaped metacontrast function: accuracy is plotted against delay. (b) Hypothetical visual responses. (1) simultaneous onset of target and mask. Transient responses do not overlap with sustained responses, and there is no masking, as shown by the arrow labelled on the left. (2) Target leads the mask by 20 msec. The transient response to the mask slightly overlaps the beginning of the sustained response to the target and some interference occurs. There is a small amount of masking, as shown by arrow 2 on the left. (3) difference in onsets of target and mask produce maximum overlap of transient and sustained components, and thus. the greatest amount of interference. As shown by arrow 3 on the left, this is the point of maximum masking. (4) Target leads mask by 60 msec. The transient response to the mask again only slightly overlaps the sustained response to the target, and the amount of interference is again small. Masking begins to decrease, as shown by arrow 4 on the left. (5)Target leads mask by a long delay. No interference occurs, and from this point on, no masking occurs either.

Additional measures of the time course of visual processing in normal and disabled readers have been obtained by Williams et al. (1989). In this study a metacontrast masking paradigm was used to index processing rate in both foveal and peripheral vision. In metacontrast. a target is briefly presented, and is followed at various delays by a spatially adjacent masking stimulus. Accuracy for the target

278

CHAPTER 8

is measured as a function of the delay between the target and the mask. The time course of the accuracy function is thought to reveal the time course of the processing of the target and mask. The accuracy functions typically obtained in metacontrast experiments are U-shaped. much like the schematic one shown in Figure 7a. Accuracy first decreases, reaches a low point at a n intermediate delay, and then increases again to baseline level. Two-component metacontrast theories (Breitmeyer and Ganz. 1976; Matin, 1975; Weisstein, 1968, 1972; Weisstein et al., 1975) attribute U-shaped metacontrast functions to the interaction of transient and sustained components of visual response. These models posit metacontrast masking as the result of the transient response to the later occurring mask catching up with, and inhibiting, the slower sustained response to the target. For this to occur, the mask must be delayed in time relative to the target. Figure 7b illustrates these timing assumptions. The dip, or lowest accuracy point in the function, is the point of maximum inhibition. As the dip shifts rightward toward longer delays, it could be assumed that some aspect of the transient (inhibitory) response to the mask is traveling faster. This is simply because something that occurs later has to travel faster to catch up. Thus, according to these models, dips at long delays

n W

25-

0-ONORMAL A-A DISABLED ADU LTS

>

a 1 W

w

+-+

-25! 0 10 30

II

I

1

I

I

I

60

90

120

150

180

210

240

DELAY (MSEC) Figure 8. Metacontrast functions obtained from adults, normal readers, and disabled readers. Accuracy (measured aa percent correct) for detecting the target lines when followed by the masking stimulus a t various delays is plotted relative to target lines-alone accuracy level (horizontal line). which was set at a level between 70 and 80% before each session.

READING DISABILITY

279

between the target and mask imply fast processing, and dips a t short delays imply slower processing. Williams et al. (19891, using diagonal line segments as targets and a surrounding outlined square a s a masking stimulus, obtained the metacontrast functions shown in Figure 8. The differences in dip location in the functions obtained from adults, normal readers, and disabled readers indicated that the rate of foveal visual processing was fastest in normal adults, slowest in reading disabled children, and intermediate in normal reading children. These findings are consistent with previous reports of increased temporal resolution with age (Brannan and Williams, 1988a,b), and sluggish temporal processing in disabled readers, as described above. The magnitude of metacontrast masking increased in the peripheral retina in adults and normal readers (Figure 9a,b), which is consistent with previous reports of increased masking effects in the periphery (Kolers and Rosner. 1960; Stewart and Purcell. 1970; Williams and Weisstein, 1981). There was, however, an absence of metacontrast masking in disabled readers with peripheral presentations (Figure 9c), a finding which is compatible with Gelger and Lettvin's (1987) finding that dyslexic subjects show a smaller magnitude of simultaneous lateral masking in the periphery. Geiger and Lettvin attribute the reduced masking effect to a n attentional strategy of dyslexic subjects to allocate more processing capacity to peripheral as compared to foveal areas of the visual field. An alternate explanation can be derived from the two-component masking theories described above, which attribute metacontrast masking to the inhibition of relatively slow pattern formation processes by short-latency temporal processing channels. These theories would predict that a temporal processing deficit would lead to a n attenuation or elimination of metacontrast.

The role of visual masking in the reading process The implication of these different patterns of response persistence and integration and of transient/sustained interactions in the visual processing of normal and disabled readers can best be understood within the context of the role of masking in the reading process (Breitmeyer and Ganz, 1976; Breitmeyer, 1980, 1983. 1984; Matin et al., 1972). As described above, sustained channel activity is generated during each fixation interval of a fixation-saccade sequence in reading (Figure 1). Due to the long response persistence of sustained channels, this sustained channel activity could interfere, via forward masking by integration, with the sustained activity generated during the following fixation. The finding that disabled readers show longer visual integration times (Figure 6) suggests that this masking effect may be more severe in the disabled readers than in normal readers. The models additionally propose that transient channel stimulation produced by a saccade normally serves to inhibit the trailing persistence of the sustained channels, thus producing clear fixation intervals. Metacontrast masking, then, serves as an afferent neural mechanism for saccadic suppression. Breitmeyer (1980, 1983) has suggested that metacontrast, as a short-range mechanism for saccadic suppression, is too weak in the fovea to produce the amount of

CHAPTER 8

280

25

-

ADULTS

1

0-0

5

FOVEAL

v

-

-254

!

8

0 10 30

I

II

60

90

120

1

150

I

180

I

210

2 0

DELAY (MSEC)

NORMAL READERS

-25

0 10 30

60

90

120

.-.FOVEAL A-A

150

DELAY (MSEC)

PERIPHERAL

180

210

240

28 1

READING DISABILITY

25

- DISABLED READERS

.-@FOVEAL A-A PERIPHERAL

15t

-15

t

0 1'0 30

60

90

120

150

180

210

240

DELAY (M SEC) Figure 9. Metacontrast functions obtained from adults (a). (top panel, preceding page); normal readers (b). (bottom panel, preceding page); and disabled readers (c) with foveal (circles) and peripheral (triangles) viewing. Accuracy (measured as percent correct) for detecting the target linea when followed by a masking stimulus a t various delays is plotted relative to target lines-alone accuracy level (horizontal line), which was set a t a level between 70 and 80% before each session. inhibition necessary to suppress forward masking by integration effects. He suggests that a n additional long-range suppression mechanism, where peripheral transient mechanisms generated by eye movements inhibit the foveal sustained response, would be required to produce the necessary saccadic suppression. Breitmeyer and Valberg (1979) have, in fact, provided evidence for the existence of such a long-range suppression mechanism. The fact that disabled readers show U-shaped metacontrast functions in the fovea (Figure 8) suggests that transient-on-sustained inhibitory interactions are occurring, and thus the short-range mechanism for saccadic suppression is operational in these readers. The fact that the dip in the metacontrast function of the disabled readers occurs a t a shorter delay than that of the other subject groups suggests that their foveal transient response is sluggish, which may render the metacontrast mechanism less effective in producing saccadic suppression. Increased fixation durations or intersaccade intervals would be required to compensate for a sluggish metacontrast mechanism in order to produce clear fixation intervals, a n effect commonly observed in the reading behavior of disabled readers (Plrozollo. 1979).

282

CHAPTER 8

One major performance difference between disabled and normal readers involved the peripheral transient inhibitory response (Figure 9). with disabled readers falling to show local transient inhibitory operations in the periphery. Since the long-range masking mechanism proposed by Breitmeyer (1980. 1983) depends on the spatial pooling of local transient activity in the periphery, it may be that the long-range mechanism for saccadic suppression is dysfunctional in disabled readers.

The time course of the processing of words Williams, Weisstein, and LeCluyse (1990) investigated how the temporal processing differences observed between normal and disabled readers affect the processing of words. A metacontrast masking procedure was employed to obtain estimates of the time course of the processing of words in these subject groups. Williams et al. used the single letters "S" and "N'as targets. The target letters were presented either alone, with a three letter mask that together with the target formed a word, or followed a t various delays by the three letter mask (Figure 10). Figure 11 shows the metacontrast functions collected on normal and disabled readers.

S

N

RUST

RUNT

RU T

RU T

Figure 10. Target letters alone (top panel), target letters with the

three-letter masks that together with the target form words (middle panel), and three-letter mask-Q alone (bottom panel) The points located above the baseline represent enhancement of accuracy for detecting the targets when followed by a mask over accuracy for the target letters when presented alone. Normal readers showed significant enhancement of accuracy for the target letters when followed by a word mask a t simultaneous presentation of the target letter and word mask (0 delay) and a t the shortest delay between the target and mask (30 msec). This enhancement is a manifestation of the enhancement phenomenon known a s the word-letter effect (Johnston and McClelland. 1973: Matthews, Weisstein, and Williams, 1974: Reicher, 1969; Wheeler, 1970). where letters are detected better

READING DISABILITY

283

within the context of words than when presented alone. Disabled readers failed to show a word-letter effect. According to two-component metacontrast theories (Breitmeyer. 1984; Breitmeyer and Ganz. 1976; Matin. 1975;Weisstein, Ozog, and Szoc, 1975).the early portion of the metacontrast function is where the sustained, pattern formation response to the target and mask can interact, suggesting that the enhancement effect found in normal readers is related to the sustained, high spatial frequency component of visual response. Disabled readers did not show the enhancement effect that was found with normal readers, suggesting that the sustained pattern information of the target with a word mask is different in normal and disabled readers.

20 --

, t 02 nc

-30

0

WORD

30

0-ONORMAL A-A DISABLED

60

90

120

150

180

210

240

DELAY (MSEC) Figure 11. Metacontrast functions collected on normal and disabled readers with clear stimulus images. Accuracy for detecting the target letters when followed by a word mask is plotted relative to accuracy for the target letters-alone (horizontal line). Positive accuracy indicates that the mask enhanced the visibility of the target, and negative accuracy indicates that the mask impaired the visibility of the target. Another important aspect of these data is that the metacontrast functions are U-shaped. with maximum masking, or a dip in accuracy, occurring at intermediate delays. The temporal location of the dip occurs at longer delays in the function obtained from normal readers as compared with that of disabled readers, suggesting that the processing of words is slower in disabled readers. Specifically. metacontrast theories would interpret this difference as indicating that the transient

284

CHAPTER 8

response to the word mask has a slower rise time or longer latency in disabled readers. Disabled readers also showed a smaller magnitude of masking at the dip, which is consistent with either a weaker transient response to the word or a more resilient sustained response to the target . Thus the time course of the processing of words appears to be different in normal and disabled readers. Williams, Weisstein. and LeCluyse (1990) also measured the time course of the processing of words using blurred stimulus images to evaluate the contribution of high spatial frequency information. Image blurring was accomplished by covering the monitor screen with a sheet of frosted acetate. By measuring contrast reduction of sine wave gratings on a n oscilloscope screen, it was determined that this method of image blurring produced contrast reduction mainly in the high spatial frequency range. The blurred stimulus images were found to produce a distinctly different set of functions than were obtained with the clear stimulus images (Figure 12). In the function obtained from normal readers, the enhancement effect was eliminated and the dip shifted leftward to a shorter delay. The finding that diminishing high spatial frequency response diminishes the enhancement effect implies that high spatial frequency channels in the visual system are involved in the processing of words. The leftward shift of the dip can be accounted for in one of two ways. According to metacontrast theories. temporal shifts of the dip from one delay to another would result if either the timing of the transient response to the mask or the sustained response to the target changed. Since image blurring presumably leaves the transient component of response unaltered, it seems more likely that the sustained response to the target becomes relatively faster. Accordingly, the word-mask would have to be delayed less in order for it to coincide in time with the target, producing a shift in the maximal masking effect to a shorter delay. The function obtained from disabled readers showed enhancement and a late dip - characteristics evident in the data of normal readers in the clear image condition. The clear image data suggested that the transient response to the word stimulus may be faster in normal than in disabled readers. In disabled readers, there may be a lack of temporal separation and therefore less temporal distinction between transient and sustained responses (Williams et al., 1987) due to a sluggish response from the transient system. Image blurring, which is thought to diminish the contrast of mainly high spatial frequencies (Ikeda and Wright. 1972). may function to decrease the amplitude of the sustained component of visual response, and a s a result, may increase the latency and/or rise time as well. Decreasing the amplitude of the longer-persisting high spatial frequency response may function to eliminate this temporal smearing and to create a temporal separation between transient and sustained components of visual response. The result may be a disinhibition of each component of response, thus allowing for a normal pattern of interactions over time. The fact that there was an increase in the magnitude of masking with image blurring suggests that when sustained activity is diminished, the relationship between the two systems may be restored, and transient-on-sustained inhibition improved. Comparison of the normal reader data obtained with clear stimulus images and the disabled reader

READING DISABILITY

285

data obtained with blurred stimulus images shows quite clearly that image blurring renders the performance of disabled readers comparable to that of normal readers in the processing of words.

+

-20-

W

-30-

a 1

1 I

I

I I

1 1

I

LO

DELAY (MSEC) Figure 12. Metacontrast functions collected on normal and disabled

readers with blurred stimulus images. Accuracy for the target letters when followed by a word mask is plotted relative to accuracy for the target letters-alone (horizontal line). Positive accuracy indicates that the mask enhanced the visibility of the target, and negative accuracy indicates that the mask impaired the visibility of the target. Williams, Weisstein. and LeCluyse (1990) investigated the nature of the contribution of transient, low spatial frequency information to the processing of words by observing changes produced by ramping the stimuli. Since the transient component of visual response is more sensitive to high temporal frequencies than to low temporal frequencies, and ramping a stimulus diminishes high temporal frequencies, then ramping a stimulus should diminish the activation of the transient component of visual response. Metacontrast theories attribute U-shaped metacontrast functions to transient versus sustained latency differences, and would predict that an increase in the contribution of sustained relative to transient channel processing should produce a shift in maximal masking effects to shorter delays. Moreover, since the masking effect depends on the inhibitory effect exerted by the transient response to the mask, diminishing the transient component of response should attenuate the overall masking effect. Both predictions were borne out in the functions produced by

286

CHAPTER 8

the word-mask in normal subjects (Figure 13). The temporal functions produced by the word-mask showed an attenuated masking effect and a shift in dip location to shorter delays. Moreover, the masking effect, though diminished, was broader or more prolonged, as would be predicted if the contribution of higher spatial frequency channels, having a longer response persistence, was increased relative to that of the shorter persistence low spatial frequency response. The fact that the data of normal subjects in the ramped condition resembles that of disabled readers in the clear image condition supports the contention that the visual processing of disabled readers is characterized by a slower transient response and increased visual persistence. This finding is consistent with previous studies showing that temporal processing differences between normal and disabled readers disappear when transient system activity is reduced (Slaghuis and Lovegrove, 1984). The data of disabled readers in the ramped condition (Figure 13) is not distinctly different from the clear image condition, indicating that with clear images, the visual processing of disabled readers is already characterized by a slower transient response and increased visual persistence. The masking studies described above provide additional evidence that the visual processing differences observed between normal and disabled readers are related to the relative timing of low and high spatial frequency channels in disabled readers. The results of these studies suggest that a sluggish transient system in disabled readers may result in a lack of temporal separation between transient and sustained processes. The .functioning of the sustained system is affected through a different pattern of interactions over time. image blurring, which was assumed to reestablish normal temporal relationships, also rendered the performance of disabled readers comparable to that of normal readers.

The effect of wavelength on the time course of visual processing Recent psychophysical and physiological data indicate that color or wavelength differentially affect the response characteristics of transient and sustained processing channels. and that wavelength can affect the relative contributions of transient and sustained channels to the processing of a stimulus. Physiological observations of the primate visual system indicate that there are differences in the color selectivity of these systems (Livingstone and Hubel, 19881, and that a steady red background light attenuates the response of transient channels (Dreher. Fukuda, and Rodieck. 1976: Kruger. 1977; Schiller and Magpeli, 1978). A recent investigation by Breitmeyer and Williams (1980) provides evidence that variations in wavelength produce similar effects in the human visual system. They found that the magnitude of both metacontrast and stroboscopic motion was decreased when red as compared with equiluminant green or white backgrounds were used. According to transient-sustained theories of metacontrast and stroboscopic motion, these results indicate that the activity of transient channels is attenuated by red backgrounds. Williams, Breitmeyer, and Lovegrove (1990). using a metacontrast paradigm, additionally found that the rate of processing in transient channels increases a s

287

READING DISABILITY

wavelength decreases, and that red light enhances the activity of sustained channels. Other human psychophysical studies have shown that transient channels are not sensitive to changes in hue when luminance transients are not also present (Bowen et al.. 19771, and that the time course of visual processing is wavelength-specific (Walters, 1970; Foster, 1979).

W

% 0

QL

20 --

WORD

0-0

A-A

-304 0

30

I 1

60

90

I

120

I

150

NORMAL DISABLED

I

180

210

240

DELAY (MSEC) Figure 13. Metacontrast functions collected on normal and disabled readers with ramped stimulus images. Accuracy for the target letters when followed by a word mask is plotted relative to accuracy for the target letters-alone (horizontal line). Positive accuracy indicates that the mask enhanced the visibility of the target, and negative accuracy indicates that the mask impaired the visibility of the target. Solman, Dain, and Keech (1990) recently investigated the effect of color on the visual processing characteristics of reading disabled as compared with normal subjects. Measuring color mediated contrast sensitivity in normal and disabled readers, they found that there is a decrement in contrast sensitivity with increasing spatial frequency in disabled but not normal readers when colored versus noncolored filters were used. It was postulated that restricting the wavelength of the incoming light reduces the level of sustained activity. This manipulation thus may have compensated for a transient system deficit by limiting the activity in the sustained system, thus improving transient-on-sustained inhibition. Williams, Faucheux and LeCluyse (1990) utilized a metacontrast paradigm to obtain direct measures of the effects of color on temporal

CHAPTER 8

288

visual processing in normal and disabled readers. Using white diagonal lines as targets, and a white, red, or blue 12 c/deg flanking grating as a mask, Williams et al. obtained the metacontrast functions shown in Figures 14 and 15. Normal readers showed differences in both enhancement and dip location with the different colored masks (Figure 14). The fact that the delay of maximum masking occurred a t a shorter

-20

--

-30 - _i

1

I

1

I

1

DELAY (MSEC) Figure 14. Metacontrast functions collected on normal readers with masks varying in wavelength. Accuracy for the target lines when followed by a flanking grating mask is plotted relative to accuracy for the target lines-done (horizontal line). Positive accuracy indicates that the mask enhanced the visibility of the target, and negative accuracy indicates that the mask impaired the visibility of the target. delay for the red as compared with the other masks suggests that the processing rate in transient channels is slowest for the red masks. This finding may be related to previous findings that red light inhibits the activity of transient channels (Dreher. Fukuda. and Rodieck, 1976; Breitmeyer and Williams, 1990). Along the same lines, the fact that the delay of maximum masking occurred a t a longer delay for the blue as compared with the other masks suggests that blue light may enhance the processing rate in transient channels. Next, consider the differences found in the magnitude of masking a t the dips in the functions. Again, this is the point in the metacontrast function where the transient response to the target maximally overlaps with, and inhibits, the sustained response to the target. Since the target was always the same, differences in the

READING DISABILITY

289

magnitude of masking at the dip can be attributed to differences in the response magnitude of transient channel activity generated by the mask. The fact that there was a smaller magnitude of masking with the red as compared with the longer wavelength masks suggests that transient channels respond less vigorously to short wavelength stimuli.

0-

+

- -0BLUE

> Q

W + J 0 2-30 -

QL

0

30

60

90

120

150

180

210

2

DELAY (MSEC) Figure 15. Metacontrast functions collected on disabled readers with masks varying in wavelength. Accuracy for the target lines when followed by a flanking grating mask is plotted relative to accuracy for the target lines-alone (horizontal line). Positive accuracy indicates that the mask enhanced the visibility of the target, and negative accuracy indicates that the mask impaired the visibility of the target. At simultaneous presentation of the target and mask, target identification accuracy was enhanced over the accuracy level for the targets when they were presented alone. This finding is consistent with previous reports of contextual information enhancing the detectability of briefly presented targets (Weisstein and Harris. 1974: Williams and Weisstein. 198 1, 1984). According to masking models based on transient/sustained theory, this is the part of the function where the sustained components of response to the target and mask can interact. Since the enhancement effect varied with the wavelength of the mask, it appears that the sustained component of visual response is sensitive to variations in wavelength. The results indicate that the sustained channels respond with greater sensitivity to red light as compared with blue and white light.

290

CHAPTER 8

Disabled readers also showed differences in dip location and magnitude of masking with the wavelength of the mask (Figure 15). Overall, dip locations occurred a t shorter delays for disabled as compared with normal readers, suggesting that the processing rate in transient channels is slower in disabled readers. As with the normal readers, however, the processing rate in transient channels appears to be slowest with the red mask and fastest with the blue mask. Disabled readers generally showed a smaller magnitude of masking than normal readers, again suggesting that, overall, transient channels respond less vigorously in disabled readers. As with the normal readers, there was a smaller magnitude of masking with the red as compared with the longer wavelength masks, suggesting that transient channels respond less vigorously to short wavelength stimuli. Finally, it is interesting to note that the function produced by the blue mask in disabled readers is similar in time course to the function produced by the white mask in normal readers. This finding suggests that blue light produces a normal time course of processing in disabled readers, and is consistent with the contention that blue light may enhance the processing rate in transient channels.

The effect of image blurring and wavelength on reading performance Given the systematic effects of image blurring and color on the perceptual performance of the reading disabled, and the fact that these manipulations can render their performance comparable to that of normal readers, Williams, LeCluyse and Faucheux (1990) investigated the effects of image blurring and color on actual reading performance. To assess reading performance. reading comprehension was measured for standardized reading passages under three temporal presentation conditions. In the first condition, the passages were presented one word a t a time, each word being centered on a computer monitor. In this reading condition, eye movements were not required for successful reading. In the second condition, the passages were presented one line a t a time. with the words painted from left to right in a moving window fashion. In this condition, eye movements were required, but were guided by the presentation of the text. In the third condition, the passages were presented one line at a time, with all of the words in each line being painted simultaneously. This was a free eye movement condition: eye movements were required and were under each subject's control. The grade level and presentation rate of the passages were determined by each subject's performance on a standardized reading test. The passages were presented with white, red, blue, and white-blurred text on a black background in separate blocks. Figure 16 shows the reading comprehension scores obtained with clear and blurred white text. When the passages were presented with clear images, the three presentation conditions did not have differential effects on the reading comprehension of normal readers. Disabled readers, on the other hand, performed more poorly in the free eye movement condition than in the other conditions. When the passages were presented with blurred images, the performance of normal readers deteriorated as eye movement control became more difficult Disabled readers showed a n opposite trend: their performanc improved in the free eye movement condition, again demonstrating t l

READING DISABILITY

29 1

n

K

NORMAL READERS

u

Z 0 -

.-.CLEAR A-A

90 --

BLURRED

v,

Z

W

I W

Qi

Q

I 0

0

50

!

I I

NONE

I 1

GUIDED

I

FREE

EYE MOVEMENTS n

K

DISABLED READERS

U

Z

0 v,

90 --

T

Z

80 --

W

70 --

w I Qi

n

I 0 0

0-@CLEAR A-A BLURRED

60 --

50

!

I

NONE

I I

GUIDED

I

FREE

EYE MOVEMENTS Figure 16. Reading comprehension (measured as percent correct on literal recall questions) for clear white text (circles) and blurred white text (triangles) under three temporal presentation conditions. (a) (top panel) Normal readers. (b) (bottom panel) Disabled readers.

292

CHAPTER 8

beneficial effects of image blurring on the performance of this subject group. Figure 17 shows the reading comprehension scores obtained with the white as compared with colored text. The pattern of results with the blue text is very similar to that found with the blurred text. The red text, on the other hand, had a detrimental effect on the reading performance of both normal and disabled readers. These results can best be interpreted within the model of fixation-saccade processes in reading described above (Breitmeyer and Ganz. 1976: Breitmeyer, 1980. 1983: Matin et al., 1972). First, the blur manipulation improved the reading comprehension performance of normal readers in the no eye movement condition, and decreased their performance in the free eye movement condition. Under normal circumstances, when no eye movements are involved in reading, there would be a lack of transient inhibition generated by eye movements, and thus a greater degree of masking by integration from one fixation to the next. Image blurring, which may function to diminish the trailing persistence of high spatial frequency channels, may serve to diminish this masking effect and thus render the reading process more efficient. Thus the reading performance of normal readers actually improves with blurred images in this condition. Similarly, the reading performance of normal readers may improve in this condition when blue text is used if the use of a short wavelength stimulus results in a faster, more vigorous response from the transient channels, resulting in improved transient-on-sustained inhibition (Williams. Breitmeyer, and Lovegrove, 1990: Williams, Faucheux and LeCluyse. 1990). Again, improving transient-on-sustained inhibition would function to diminish the trailing persistence of high spatial frequency channels. Finally, the deterioration of the reading performance of normal readers in this condition when red text is used may be due to an increased masking by integration effect. This increased masking by integration effect would be the result of an increase in sustained channel sensitivity and weaker transient-on-sustained inhibition produced by a long wavelength stimulus (Williams, Breitmeyer, and Lovegrove, 1990; Williams, Faucheux and LeCluyse. 1990). When eye movements are involved, image blurring and color may disrupt the normal pattern of transient-sustained interactions due to the fact that the timing and sensitivity of transient and sustained channel response are altered. Thus a decrement in performance is observed in the eye movement conditions with both color and the blur manipulation. Disabled readers performed poorly in the free eye movement condition when clear, white text was used, but performed better in the conditions requiring no eye movements or guided eye movements. In the latter conditions it could be assumed that transient activity was minimized. Disabled readers showed a n improvement in reading performance with blurred images when eye movements were required, again suggesting that the image blurring manipulation may be reestablishing a normal pattern of transient-sustained interactions in visual processing. When red text was used, disabled readers showed a smaller decrement in performance in the no eye movement condition than normal readers showed. This result may be related to the finding that disabled readers do not show the increase in sustained channel sensitivity that normal readers show with red stimuli (as indexed by an

READING DISABILITY n

R W

100

Z

0 -

90

Z

80 --

v)

W

I

W

0-0

A-A

-- B-B

CY

70 --

0

60 --

e I 0

50

n

R W Z

-

0 v)

Z

W

I

100-

WHITE BLUE 90-'B-RRED

0

60 --

0

0-0

II

I

DISABLED READERS

80 --

70 --

e I

I

NORMAL READERS

A-A

CY

W

WHITE BLUE RED

293

50

I

I 1

1

Figure 17. Reading comprehension (measured as percent correct on literal recall questions) for clear white text (circles), blue text (triangles), and red text (squares) under three temporal presentation conditions. (a) (top panel) Normal readers. (b) (bottom panel) Disabled readers.

294

CHAPTER 8

enhancement effect in the metacontrast function, Figure 15). although they do show a less vigorous response from transient channel activity (as indexed by a smaller magnitude of masking, Figure 15). The increased masking by integration effect, therefore, in disabled readers may be smaller in this condition, resulting in a smaller decrement in performance. When blue text was used, the effects on reading performance were similar to those produced by the blur manipulation, suggesting that the use of the short wavelength stimulus simulates a normal pattern of visual processing during reading. In a related study, Lovegrove and Macfarlane (1990) measured the number of errors, comprehension, and reading rate under the same three temporal presentation conditions. Here it was found that disabled readers made fewer errors in the no eye movement condition than controls, and more errors in the free eye movement condition. Furthermore, reading rate increased in the no eye movement condition as compared with the free eye movement condition in disabled readers, and there was a trend, although insignificant, towards higher comprehension scores in the no eye movement condition. Lovegrove and Macfarlane interpreted their results within the framework of a central-peripheral processing dichotomy. The no eye movement condition required the processing of only central (foveal) information, whereas the free eye movement condition required the integration of central and peripheral information from successive fxations. When disabled readers only have to process central information, which requires little transient system involvement, they make fewer errors and read at a faster rate than in the condition that approximates normal reading. This improvement increased over a five session training period, but no increase in reading age on a post-test of reading ability was noted. There were several differences between Williams et a l . 3 and Lovegrove and Macfarlane's studies that may account for the fact that normal a n d disabled readers did not show differences in comprehension scores in the latter study. Most importantly, in Lovegrove and Macfarlane's study reading rate was increased on each trial that subjects correctly answered 80% of the comprehension questions posed, while in Williams et a1.k study reading rate remained constant. Thus, improvement in comprehension would have been obscured by a speed-accuracy tradeoff in the former study. Williams, LeCluyse and Faucheux (1990) also measured several characteristics of eye movement patterns for reading passages presented one page a t a time. The image blurring manipulation was found to increase reading rate and decrease span of apprehension (number of characters processed in each fixation) in disabled readers. This manipulation decreased the span of apprehension in normal readers, but had no effect on reading rate. This result may be related to the finding that disabled readers show a lack of metacontrast masking effects in peripheral vision (Williams et al., 1989). I t may be that, with clear stimulus images, this lack of peripheral inhibition results in some degree of interference of peripheral visual information on the reading process. Decreasing the amount of peripheral information available in each fixation may diminish the degree of interference experienced, resulting in an increase in the efficiency of the reading process as indexed by increased reading rate.

READING DISABILITY

295

Remaining questions A number of important questions concerning reading disability remain. First, does a transient system deficit exist before children begin to read or is it produced because disabled readers are delayed in reading acquisition? A corollary question is whether subsequent reading performance can be predicted on the basis of visual measures. To answer this, Lovegrove et al. (1986b) measured contrast sensitivity in 123 kindergarten pre-readers with a mean age of 5 years, 11 months. The sample included all children in a particular classroom, thus ensuring a broad sample of school achievement. Contrast thresholds were determined using 2 and 4 c/deg gratings with Vocabulary and digit span exposure durations of 350 msec. performance were also measured. Two years later the reading performance of these children was assessed. The extent to which the vocabulary, digit span, and contrast sensitivity scores predicted reading ability 2 years later was determined using a stepwise multiple regression procedure. On the first step, only contrast sensitivity was entered, with the multiple R = 0.27 (p<.Ol). Contrast sensitivity, therefore, was a significant predictor of reading ability. Vocabulary was entered on the second step, making the multiple R = 0.35. On the third step, digit span was entered, increasing the multiple R to 0.40. The finding that the visual measures predict later reading ability better than the vocabulary test suggests a very close link between visual processing and reading ability. On the basis of these studies, a low level visual processing difference between future good readers and poor readers is present before formal reading instruction begins. The visual deficits are not the result of a disability in reading. Other studies have considered whether the visual deficits associated with reading disability are the result of a developmental lag which corrects with age. Brannan and Williams (1988a) measured flicker thresholds for homogeneous fields of light in 8. 10.and 12 year old normal and disabled readers. A multiple regression analysis of their data was performed using reading level, age, and flicker threshold, and showed that flicker threshold accounted for 56% of the unique variance in reading level, with age contributing another 10%. Post hoc comparisons of flicker detection performance were made between normal and disabled readers matched for reading level (e.g., 10 year old normal readers and 12 year old disabled readers reading a t the same grade level). Flicker sensitivity differences remained when the children were matched for reading ability, indicating that the differences between the groups in flicker sensitivity were not due primarily to differences in reading skill: there was some fundamental difference in visual processing between normal and disabled readers. Brannan and Williams (1988b) obtained similar results with measures of perceptual grouping and temporal order judgments; performance on these perceptual tasks was not found to be due to differences in reading skill, but to a fundamental temporal processing difference between the groups. Questions may also be raised about the possibility that reading disabled children also have attentional deficits. Attentional deficits

296

CHAPTER 8

could result in poor test sensitivity and therefore could account for an apparent transient system deficit. There are two major considerations that argue against this explanation for the experimental results. In addition to the fact that in many of the studies testing was conducted with simple, nonverbal stimuli using a two-alternative forced choice paradigm, there were clear instances when the reading disabled subjects performed better or were more sensitive than controls. For example, disabled readers were shown to have slightly higher contrast sensitivity than controls to high spatial frequencies (Figure 3). Furthermore, Williams and Bologna (1985) found that providing the optimal processing strategy in a selective attention task does not eliminate performance differences between normal and disabled readers, suggesting that the performance differences are attributable to automatic rather than directed attention factors. Additionally, although normal and disabled readers differ in their ability to allocate attention across visual space, their performance in a target detection task is equivalent when prior allocation of attention is not a factor (Brannan and Williams, 1987). In the metacontrast studies reported here, the location of the target was known and did not vary. As such, selective attention mechanisms would be concentrated on a single spatial location, and transient activity in shifting or allocating attention would not be involved. Thus differences between subject groups in the integrity of the transient system a s related to the ability to flexibly allocate and direct attention would not enter into task requirements. It is possible, however, that automatic attention processes, being mediated by the transient system, are involved in the performance differences reported here. Finally, the question remains as to how a transient system deficit in disabled readers relates to the observation that disabled readers frequently have various forms of language deficiency. Some of these language based difficulties are phonological coding deficits, phonemic segmentation deficits, poor vocabulary development, and difficulty in discriminating grammatical and syntactic differences among words and sentences, One possible involvement of a transient system deficit in language deficiencies can be seen by considering the problems that disabled readers would have in integrating the inputs from successive fixations. As described above, in normal readers this problem could be solved by the transient system inhibiting the trailing persistence of sustained responses generated during each fixation. In disabled readers, there is a probability that inputs from successive inputs would be superimposed, making the task of reading confusing. This could manifest itself in a number of ways, Disabled readers may see only parts of words, and if they did not know which fixation the information came from, they could know very little about the spatial arrangement of the letters. This may lead to reading errors and word or letter reversals. In addition, depending on exactly where the reader fixated each time, the amount and type of masking or interference which occurred may vary from one reading task to another. As a result, it would be diffkult to learn any systematic grapheme-to-phoneme rule if the appearance of the graphemes was in some way unstable. It has been found that many visually disabled readers also demonstrate a phonological coding deficit in the form of performance on a nonsense word test of non-words that followed regular grapheme-phoneme rules (Lovegrove et al., 1986).

READING DISABILITY

297

Consequently it is likely that many disabled readers can have both a visual processing deficit and some form of language difficulty.

Conclusions Research conducted over a number of years has demonstrated a transient system deficit in a majority of specifically disabled readers tested. Differences have been found between normal and disabled readers in measurements of visible persistence, pattern contrast sensitivity, and temporal or flicker contrast sensitivity. The foveal temporal processing of disabled readers has been found to be sluggish compared to that of normal readers. Peripheral visual processing has been found to be characterized by a lack of inhibitory processes. Differences in the temporal processing of normal and disabled readers surface in the processing of both simple stimuli and words. The primary deficit appears to be in the response properties of the transient subsystem of the visual system, although the response of the sustained system may be affected through a different pattern of interactions with the transient system over time. Measures of sustained system function that presumably do not involve a dynamic interaction between transient and sustained systems fail to demonstrate visual processing differences between normal and disabled readers. The temporal processing differences between the groups have been found to have consequences for the processing of words and for reading performance. Image blurring and color have the effect of reestablishing a normal pattern of visual processing in disabled readers. A transient system deficit was found in approximately 75% of the disabled readers tested, seems to be present before reading instruction, and predicts reading ability reasonably well.

References Badcock, D. and Lovegrove. W. (1981). The effects of contrast, stimulus duration, and spatial frequency on visible persistence in normal and specifically disabled readers. Journal of Experimental Psychology: Human Perception and Performance. 7 . 495-505. Benton, A. (1962). Dyslexia in relation to form perception and directional sense. In J. Money (Ed.), Reading Disability: Progress and Research Needs in Dysleuh. Baltimore: Johns Hopkins Press. Benton. A41975). Developmental dyslexia: Neurological aspects. In W.J. Freidman (Ed.). Advances in Neurology. Vol 1 . Current Reviews of Higher Order Nervous System Dysfunction. New York: Raven Press. Bowen. R.. Pokorny, J. and Cacciato. D. (1977). Metacontrast masking depends on luminance transients. Vision Research, 17, 971-975. Bowling, A.. Lovegrove. W. and Mapperson, B. (1979). The effect of spatial frequency and contrast on visible persistence. Perception. 8 , 529-539. Brannan, J.R. and Williams, M. (1987). Allocation of visual attention in good and poor readers. Perception and Psychophysics. 41. 23-28. Brannan, J.R. and Williams, M. (1988a). The effects of age and reading ability on flicker thresholds. Clfnical Vision Sciences, 3(2).137-142. Brannan, J.R. and Williams, M. (1988b). Developmental versus sensory deficit effects on perceptual processing in the reading disabled. Perception and Psychophysics. 44(5). 437-444.

298

CHAPTER 8

Breitmeyer. B. (1975). Simple reaction time a s a measure of the temporal response properties of transient and sustained channels. Vision Research 15. 1411-1412. Breitmeyer, B. (1980). Unmasking visual masking: A look a t the "why" behind the veil of the "how." Psychological Review. 87. 52-69. Breitmeyer, B. (1983). Sensory masking, persistence, and enhancement in visual exploration and reading. In Rayner K. (Ed.), Eye Movements in Reading: Perceptual and Language Processes. New York: Academic Press. Breitmeyer. B. (1984). Visual Masking: An Interactive Approach. Oxford: Oxford University Press. Breitmeyer. B. and Ganz. L. (1976). Implications of sustained and transient channels for theories of visual pattern masking, saccadic suppression, and information processing. Psychological Review, 83, 1-36. Breitmeyer. B. and Ganz. L. (1977). Temporal studies with flashed gratings: Inferences about human transient and sustained channels. Vision Research, 17, 861-865. Breitmeyer, B.. Levi, D.M. and Harwerth. R.S. (1981). Flicker-masking in spatial vision. Vision Research, 21, 1377-1385. Breitmeyer, B. and Valberg, A. (1979). Local, foveal, inhibitory effects of global, peripheral excitation. Science. 203, 463-465. Breitmeyer. B. and Williams, M. (1990). Effects of isoluminantbackground color on metacontrast and stroboscopic motion: Interactions between sustained (P) and transient (M) channels. Vision Research. Burbeck, C.A. (1981). Criterion-free pattern and flicker thresholds. Journal of the Optical Society of America, 71. 1343-1350. Critchely. M. (1964). Developmental Dyslexia. Heinemann. London. Dreher, B., Fukuda, Y. and Rodieck, R. (1976). Identification, classification and anatomical segregation of cells with X-like and Y-like properties in the lateral geniculate nucleus of old-world primates. Journal of Physiology, 258, 433-452. Fletcher, J. and Satz, P. (1979a). Unitary deficits hypothesis of reading disability: Has Vellutino led u s astray'? Journal of Learning Disabilities. 12(3).155-159. Fletcher, J. and Satz, P. (1979b). Has Vellutino led u s astray? A rejoinder to a reply. Journal of Learning Disabilities 12. 168-171. Foster, D. (1978). Interactions between blue- and red-sensitive colour mechanisms in metacontrast masking. Vision Research, 19, 92 1-931. Gelger. G. and Lettvin. J . (1987). Peripheral vision in persons with dyslexia. New England Journal of Medicine, 316. 1238-1243. Ikeda, H. and Wright, M. (1972). Receptive field organization of "sustained" and "transient" retinal ganglion cells which subserve different functional roles. Journal of Physiology, 227. 769-800. Johnston, J.C. and McClelland. J.L. (1973). Visual factors in word perception. Perception and Psychophysics. 14. 365-370. Kolers. P. and Rosner. B. (1960). On visual masking (metacontrast): Dichoptic observations. American Journal of Psychology, 73, 2-2 1. Kruger, J. (1977). Stimulus dependent color specificity of monkey lateral geniculate neurones. Experimental Brain Research, 30, 297-3 11.

READING DISABILITY

299

Kulikowski, J.J. and Tolhurst. D.J.(1973). Psychophysical evidence for sustained and transient detectors in human vision. Journal of Physiology London, 232, 149-162. Lennie, P. (1980). Parallel visual pathways: A review. Vkbn Research, 20. 561-594. Livingstone. M.S. and Hubel, D.H. (1987). Psychophysical evidence for separate channels for the perception of form, color, movement, and depth. Journal of Neuroscience, 7 , 3416-3468. Livingstone. M . S . and Hubel. D.H. (1988). Segregation of form, color. movement and depth: Anatomy. Physiology and perception. Science. 240. 740-749. Lovegrove. W., Billing, G. and Slaghuis. W. (1978). Processing of visual contour orientation information in normal and disabled reading children. Cortex, 14, 268-278. Lovegrove, W., Bowling, A., Badcock, D., and Blackwood, M. (1980a). Specific reading disability: differences in contrast sensitivity a s a function of spatial frequency. Science. 210, 439-440. Lovegrove, W.. Bowling, A.. Slaghuis, W.. Geeves, E.. and Neison. P. (1986b). Contrast sensitivity scores of pre-readers as predictors of reading ability. Perception and Psychophysics. 40, 440-444. Lovegrove, W. and Brown, C. (1978). Development of information processing in normal and disabled readers. Perceptual and Motor Skills, 46, 1047-1054. Lovegrove. W., Heddle, M and Slaghuis, W. (1980b). Reading disability: spatial frequency specific deficits in visual information store. Neuropsychologica. 18. 111-115. Lovegrove. W. and Macfarlane, T. (1990). How can we help SRDs in learning to read? Unpublished honors thesis. Lovegrove, W., Martin, F., Bowling, A., Blackwood, M., Badcock, D. and Paxton, S. (19821. Contrast sensitivity functions and specific reading disability. Neuropsychologica. 20. 309-315. Lovegrove. W.. Martin, F. and Slaghuis, W. (1986). A theoretical and Experimental case for a visual deficit in specific reading disability. Cognitive Neuropsychology, 3. 225-267. Martin, F. and Lovegrove, W. (1984). The effects of field size and luminance on contrast sensitivity differences between specifically reading disabled and normal children. Neuropsychologica. 22, 73-77. Martin, F. and Lovegrove, W. (1987). Flicker contrast sensitivity in normal and specifically disabled readers. Perception, 16. 215-221. Matin. E. (1974). Saccadic suppression: A review and analysis. Psychological Bulletin, 81. 899-915. Matin, E. (1975). The two-transient (masking) paradigm. Psychological Review. 82, 451-461. Matin. E.. Clymer. A. and Matin, L. (1972). Metacontrast and saccadic suppression. Science, 178, 179-182. Matthews, M., Weisstein. N. and Williams, A. (1974). Masking of letter features does not remove the word-superiority effect. Bulletin of the Psychonomic Society. 4, 262. Maunsell. J.H.R. (1987). Physiological evidence for two visual subsystems. In L. Vaina (Ed.). Matters of Intelligence. Amsterdam: Reidel. May, G., Wil1iams.M. and Dun1ap.W. (1988). Temporal order judgements in good and poor readers. Neuropsychologica, 26(6). 917-924.

300

CHAPTER 8

Meyer, G. and Maguire. W. (1977). Spatial frequency and the mediation of short-term visual storage. Science, 198, 524-525. Pirozzolo, F. (1978). The neuropsychology of developmental reading disorders. New York: Praeger. Rayner, K and McConkie, G. (1976). What guides a reader's eye movements? Vision Research, 16. 829-837. Reicher, G.M. (1968). Perceptual recognition a s a function of meaningfulness of stimulus material. Journal of Experimental Psychology. 61, 275-280. Schiller. P. and Malpeli. J. (1978). Functional specificity of lateral geniculate nucleus laminae of the rhesus monkey. Journal of Neurophysiology, 41, 766-767. Singer, W. and Bedworth, N. (1973). Inhibitory interaction between X and Y units in cat lateral geniculate nucleus. Brain Research, 4 8 . 291 -307. Slaghuis. W. and Lovegrove. W. (1984). Flicker masking of spatial frequency dependent visible persistence and reading disability. Perception, 13. 527-534. Slaghuis, W. and Lovegrove, W. (1985). Spatial-frequency mediated visible persistence and specific reading disability. Brain and Cognition, 4. 238-240. Solman. R., Dain, S. and Keech. 5. (1990). Colour mediated contrast sensitivity in disabled readers. Submitted. Stanley, G. (1975). Visual memory processes in dyslexia. In D. Deutsch and J . Deutsch (Eds.). Short-term memory. New York: Academic Press. Stanley, G. and Hall, R. (1973a). A comparison of dyslexics and normals in recalling letter arrays after brief presentations. British Journal Experimental Psychology, 43, 301-304. Stanley, G. and Hall, R. (1973b). Short-term visual information processing in dyslexia. Child Development, 44. 841-844. Stewart, A. and Purcell. D. (1970). U-shaped masking functions in visual backward masking: Effects of target configuration and retinal position. Perception and Psychophysics. 7 , 253-256. Stone J., Dreher B. and Leventhal A.G. (1978). Hierarchical and parallel mechanisms in the organization of the visual cortex. Brain Research Review, 1, 345-354. Tolhurst, D.J. (1973). Separate channels for the analysis of the shape and movement of a moving visual stimulus. Journal of Physiology London. 231, 395-402. Tolhurst, D.J. (1975). Reaction times in the detection of gratings by human observers: A probabilistic mechanism. Vision Research, 15, 1143-1148. Vasselev. A. and Miltov. D. (1976). Perception time and spatial frequency. Vision Research, 16, 28-32. Vellutino, F.. Steger. J.. DeSetto. L. and Phillips, F. (1975a). Immediate and delayed recognition of visual stimuli in poor and normal readers. Journal of Experimental Child Psychology, 223-232. Vellutino, F.. Steger, J., Kaman. M. and DeSetto. L. (1975b). Visual form perception in deficient and normal readers as a function of age and orthographic-linguistic familiarity. Cortex, 11. 22-30. Vellutino, F. (1977). Alternative conceptualizations of dyslexia: evidence in support of verbal-deficit hypothesis. Harvard Educational Review, 47. 334-354.

READING DISABILITY

301

Vellutino. F. (1978a). The validity of perceptual deficit explanations of reading disability: A reply to Fletcher and Satz. Journal o f k a r n i n g Disabilities, 12(3),160-167. Vellutino, F. (1978b). Dyslexia: Theory and Research. London: M.I.T. Press. Vellutino, F. (1987). Dyslexia. Scienti& American, 256(31. 34-4 1. Walters, J . (1970). Metacontrast: The effects obtained with consecutively presented concentric disks and rings of different wavelengths. American Journal of Optometry, 46, 634-638. Watson, A.B. and Nachmias, J. (1977). Patterns of temporal interaction in the detection of gratings. Vision Research, 17. 193-202. Weisstein, N. (1968). A Rashevsky-Landahl neural net: Simulation of metacontrast. Psychological Review, 75. 484-521. Weisstein. N. (1972). Metacontrast. In D. Jameson and L.M. Hurvich (Eds.). Handbook of sensory physiology, Vol. 7/4. Visual Psychophysics, pp. 233-72. Weisstein, N. and Harris, C. (1974). Visual detection of line segments: an object-superiority effect. Science, 186. 752-755. Weisstein. N.. Ozog. G. and Szoc, R. (1975). A comparison and elaboration of two models of metacontrast. Psychological Review, 82, 325-342. Wheeler, D.D. (1970). Processes of word recognition. Cognitive PSyChOlogy, 1, 58-65. Williams, M. and Bologna, N. (1985). Perceptual grouping in good and poor readers. Perception and Psychophysics, 33,367-374. Williams, M.C.. Brannan, J.R.. and Lartigue, E. K. (1987). Visual search in good and poor readers. Clinical Vision Sciences, 1. 367-371. Williams, M., Breitmeyer, B. and Lovegrove. W. (1990). Metacontrast with masks varying in spatial frequency and wavelength. Submitted to Vision Research Williams, M.. Faucheux, A. and LeCluyse. K. (1990).The time course of the processing of stimuli with different wavelengths in normal and disabled readers. Submitted. Williams, M.. LeCluyse. K. and Bologna, N. (1990).Masking by light as a measure of visual integration time and persistence in normal and disabled readers. Clinical Vision Sciences. Williams, M. and LeCluyse. K. (1990). Perceptual consequences of a temporal processing deficit in reading disabled children. Journal of the American Optometric Association, 61, 1 11- 121. Williams, M.. LeCluyse. K. and Faucheux. A. (1990). Effects of image blurring and color on reading performance in normal and disabled readers. In preparation. Williams, M.. Molinet, K. and LeCluyse, K. (1989). Visual masking as a measure of temporal processing in normal and disabled readers. Clinical Vision Sciences, 4(2),137-144. Williams, M. and Weisstein, N. (1980). Perceptual grouping produces spatial-frequency specific effects on metacontrast. Investigative Opthalmology and Visual Science, 165. Williams, M. and Weisstein. N. (1981). Spatial frequency response and perceived depth in the time-course of object-superiority. Vision Research, 21,631-646.

302

CHAPTER 8

Williams, M. and Weisstein. N (1984). The effect of perceived depth on metacontrast functions. Vision Research, 24(10). 1278-1282. Williams, M., Weisstein, N. and LeCluyse, K. (1990). The time course of the processing of words in normal and disabled readers. Submitted.