Impaired temporal contrast sensitivity in dyslexics is specific to retain-and-compare paradigms

Brain (2001), 124, 1381–1395

Impaired temporal contrast sensitivity in dyslexics is specific to retain-and-compare paradigms Gal Ben-Yehudah,1 Ella Sackett,2 Liat Malchi-Ginzberg2 and Merav Ahissar2 1Department

of Neurobiology and 2Department of Psychology, Hebrew University, Jerusalem, Israel

Correspondence to: Dr Merav Ahissar, Department of Psychology, Hebrew University, Jerusalem 91905, Israel E-mail: [email protected]

Summary Developmental dyslexia is a specific reading disability that affects 5–10% of the population. Recent studies have suggested that dyslexics may experience a deficit in the visual magnocellular pathway. The most extensively studied prediction deriving from this hypothesis is impaired contrast sensitivity to transient, low-luminance stimuli at low spatial frequencies. However, the findings are inconsistent across studies and even seemingly contradictory. In the present study, we administered several different paradigms for assessing temporal contrast sensitivity, and found both impaired and normal contrast sensitivity within the same group of

dyslexic participants. Under sequential presentation, in a temporal forced choice paradigm, dyslexics showed impaired sensitivity to both drifting and flickering gratings. However, under simultaneous presentation, with a spatial forced choice paradigm, dyslexics’ sensitivity did not differ from that of the controls. Within each paradigm, dyslexics’ sensitivity was poorer at higher temporal frequencies, consistent with the magnocellular hypothesis. These results suggest that a basic perceptual impairment in dyslexics may be their limited ability to retain-and-compare perceptual traces across brief intervals.

Keywords: dyslexia; contrast sensitivity; magnocellular; sequential presentation; temporal processing Abbreviations: ISI ⫽ inter-stimulus interval; LGN ⫽ lateral geniculate nucleus; S-FC ⫽ spatial forced choice; T-FC ⫽ temporal forced choice

Introduction Developmental dyslexia is a specific reading disability that affects 5–10% of the population (Shaywitz, 1998). Dyslexia is usually defined as low reading ability compared with that expected from general cognitive abilities, which cannot be explained by lack of education or emotional stress (Diagnostic and Statistical Manual of Mental Disorders; DSM-IV, 1994). Several recent studies have reported behavioural, imaging and EEG abnormalities that may reflect a deficit in dyslexics’ visual magnocellular pathway (for a review, see Stein and Walsh, 1997). Together with more direct anatomical evidence (Livingstone et al., 1991), they constitute the experimental basis for the ‘magnocellular hypothesis,’ described below. According to this hypothesis, the magnocellular system is impaired in a large proportion of dyslexics and this deficit contributes to their reading difficulties. The division between the magnocellular and parvocellular pathways in the visual system begins at the retina, where they differ in cell size (large versus small, respectively). Large compared with small retinal ganglion cells project to different layers in the lateral geniculate nucleus (LGN) (see Shapley and Perry, 1986). Projections from the LGN are still © Oxford University Press 2001

largely separate in the primary visual area (Shapley, 1990; Merigan and Maunsell, 1993). Some degree of segregation is retained in higher cortical visual areas, where magnocellular projections are more abundant in the dorsal stream and parvocellular projections are more abundant in the ventral stream (Ungerleider and Mishkin, 1982; Maunsell et al., 1990; Schiller and Logothetis, 1990; Maunsell, 1992). Selective lesions to the magnocellular layers of monkey LGN result in reduced contrast sensitivity to stimuli of low luminance and high temporal frequency (at or above 10 Hz; Merigan and Maunsell, 1990; Merigan et al., 1991a). Thus, detection of briefly presented stimuli, or stimuli that flicker or drift along the screen (i.e. transient stimuli), is impaired. On the other hand, no deficit in contrast sensitivity has been found for detecting stationary stimuli (i.e. sustained stimuli). In the spatial domain, impaired sensitivity is found mainly for low spatial frequencies (at or below 1 cycle per degree (c/°); see Skottun, 2000 for an evaluation of both spatial and temporal parameters). Several lines of research have reported findings that support a link between a magnocellular deficit and dyslexia. A post-

1382

G. Ben-Yehudah et al.

mortem study of dyslexics’ brains found that the number and size of neurones were reduced in the magnocellular layers of their LGNs (Livingstone et al., 1991). EEGs showed that responses to brief, low-luminance stimuli were reduced or delayed in dyslexics (Livingstone et al., 1991; Lehmkuhle, 1993). Psychophysical studies, beginning with that of Lovegrove and colleagues (Lovegrove et al., 1980), reported impaired contrast sensitivity in dyslexics. On the basis of the above, a magnocellular deficit is expected to induce a greater impairment in contrast sensitivity for higher (⬎10 Hz) temporal frequencies and lower (⬍1 c/°) spatial frequencies. Indeed several studies report findings supporting this prediction (Martin and Lovegrove, 1987; Evans et al., 1994, Flemingham and Jakobsen, 1995; Borsting et al., 1996; Ridder et al., 1997; Demb et al., 1998; Slaghuis and Ryan, 1999). However, other studies have not found any impairment using similar stimuli (Cornelissen et al., 1995; Gross-Glenn et al., 1995; Walther-Muller, 1995; Hayduk et al., 1996; Spinelli et al., 1997). Whereas the negative findings in some studies could be attributed to the use of high-luminance stimuli (Cornelissen et al., 1995; Gross-Glenn et al., 1995), and are thus not inconsistent with the magnocellular hypothesis, other results are hard to reconcile (Walther-Muller, 1995; Hayduk et al., 1996; Spinelli et al., 1997). Even more puzzling findings, from the perspective of the magnocellular deficit theory, are reports of impaired sensitivity in ranges attributed to the parvocellular pathway (Merigan et al., 1991b). These include reports of impaired sensitivity to sustained stimuli (Lovegrove et al., 1982; Cornelissen, 1993; Mason et al., 1993; Evans et al., 1994; Spinelli et al., 1997) with or without impaired sensitivity to transient stimuli. Other inconsistencies include findings that impaired detection of transient stimuli is more pronounced at the high spatial frequency range (Martin and Lovegrove, 1987, 1988). One interpretation which frequently is put forward to account for cross-study variability in contrast sensitivity is that different studies have assessed different types of dyslexia. Borsting and colleagues (in adults) and Slaghuis and Ryan (in children) (Borsting et al., 1996; Slaghuis and Ryan, 1999) found impaired contrast sensitivity only in one type of dyslexia; namely, those who express difficulties in both phonological abilities and in their ability to read and spell irregular words (dysphoneidetics, as defined by Boder, 1973). The estimated prevalence of this type depends on the assessment procedure (Manis et al., 1996), and ranges from 10% of the dyslexic population (Flynn and Boder, 1991) to a third and more (Castles and Coltheart, 1993). If the prevalence of dyslexics with an impaired magnocellular system is high, it is surprising that some studies have failed to detect such a deficit. One explanation is that the proportion of different types of dyslexia is language dependent, with perhaps fewer phonological difficulties in individuals speaking languages with easy and shallow phonetic rules, such as Italian or German (see Spinelli et al.,

1997; Walther-Muller, 1995, respectively). This suggestion is hard to test, since the proportion of various types of dyslexia depends on the specific tests used to assess these subtypes, which of course cannot be the same across different languages. In the present study, we too inquired whether dyslexics’ contrast sensitivity, or at least the sensitivity in a specific subtype of dyslexia, is indeed impaired in a manner consistent with a magnocellular deficit. However, being aware of interstudy differences, we also tried to resolve the source of previous discrepancies. Towards this goal, we recruited a group of adult dyslexics and systematically characterized their reading and cognitive abilities. In addition, we administered a variety of tests to assess contrast sensitivity, using different behavioural paradigms and stimulus conditions. In particular, we chose to replicate the experimental designs used in two previous studies, one conducted in Australia where a clear magnocellular-like deficit was found (Borsting et al., 1996), and the other in Italy (Spinelli et al., 1997) where no such deficit was found. Both groups of researchers attributed their conflicting results to sampling of different subtypes of dyslexia. Spinelli and colleagues suggested that since the orthography of Italian is shallow (i.e. regular grapheme– phoneme correspondence) whereas English has a deep orthography (i.e. irregular grapheme–phoneme correspondence), the proportion of Italian dyslexics experiencing phonological difficulties should be smaller. Since previous studies have reported impaired contrast sensitivity only in dyslexics with phonological difficulties, they state that it is not surprising that no such impairment was found in their Italian study. Variability in these results was attributed in part to differences in alphabetic code; therefore, we wondered whether the contrast sensitivity of Israeli dyslexics (Hebrew readers) would resemble the Italian or Australian pattern of results. Since Hebrew has both deep and shallow orthography, it is difficult to predict which pattern of results Hebrew readers would fit with best. There are two forms of written Hebrew. One, pointed Hebrew, contains a highly regular spelling–sound correspondence (vowels are indicated by diacritical marks). The other, unpointed Hebrew, has an irregular correspondence (vowel marks are omitted). Children are first taught to read pointed Hebrew. However, following increased exposure to print, they begin to read without the vowel marks. Most adults only read in the latter form. The subjects in this study were adults experienced with both forms of written Hebrew. Although the interpretation of these results in terms of different subtypes of dyslexia is plausible, there were also several methodological differences between the studies. One particularly striking difference concerns the behavioural paradigms. Each trial in the Australian study included two sequential intervals, and subjects were required to indicate which interval contained the stimulus (temporal forced choice, T-FC). In the Italian study, only one interval was used in

Contrast sensitivity in dyslexia each trial, and the subjects indicated whether the stimulus had been presented in the upper or lower part of the screen (spatial forced choice, S-FC). Thus, in the Australian study, subjects were asked implicitly to make temporal comparisons, whereas in the Italian study, comparisons were between different positions on the screen. Sequential comparisons may pose greater difficulties for dyslexics (e.g. Eden et al., 1995; Laasonen et al., 2000). Therefore, we also decided to examine the impact of temporal versus spatial comparisons on contrast sensitivity, by applying both experimental procedures in the same group of Hebrew readers. To summarize the experiments reported below, we replicated these two experimental designs and reproduced both sets of results, finding both magnocellular deficits (as in Borsting et al., 1996) and a lack of magnocellular deficits (as in Spinelli et al., 1997) within the same test population of Hebrew-speaking dyslexic adults. Thus, the difference in results could not be attributed to a different sampling of dyslexia subtypes. Rather, the differences were the outcome of applying different assessment procedures. When a T-FC paradigm was used, dyslexics’ detection was impaired. However, no impairment was detected in an S-FC paradigm using either sustained or transient stimuli. Within the T-FC paradigm, stimuli with higher temporal frequencies caused greater contrast sensitivity deficits, consistent with a magnocellular hypothesis. These findings suggest that the type of perceptual comparison required by the task is more critical to dyslexics’ performance than the specific properties of the stimuli used. The requirement to retain an accurate trace for subsequent comparison introduces a specific difficulty for dyslexics, even when explicit instructions involve only detection. Preliminary results of this study have been presented in abstract form (Ben-Yehudah et al., 2000).

Experiment 1: Contrast sensitivity in dyslexia Methods Subjects Our subjects were 38 reading-disabled adults (22 female, 16 male; mean age 22.3 ⫾ 4.2 years) and 42 normal readers (29 female, 13 male; mean age 22.5 ⫾ 4.1 years; see Table 1). All subjects were native Hebrew speakers and had normal or corrected-to-normal eyesight. Consent was obtained from all subjects before data collection commenced. Dyslexic subjects were referred to us by educators, parents or by selfreport, on the basis of a documented history of specific reading difficulties. Additional recruitment was through advertisements posted at the Hebrew University. As controls, we asked dyslexic participants to bring friends and/or spouses, thus ensuring a similar age group with a similar educational background. As part of the study, we administered reading, spelling and cognitive tests. Oral reading tests included lists of single words and non-words written with vowel marks, and an academic-level paragraph in which vowel marks were

1383

omitted. Orthographic skills were assessed using both a spelling and a lexical decision task between words and their pseudohomophones (Shalem and Lachmann, 1998). The Hebrew language consists of 22 letters, of which five letter pairs are homophones. All of the possible homophonic pairs were used in the lexical decision task. Phonological awareness was assessed separately using a spoonerism task (swapping the first phoneme in the first word with the first phoneme in the second word of an orally presented word pair). General cognitive skills were assessed with the Raven-SPM test (Raven et al., 1992) and four subtests (block design, digit symbol-coding, digit span and similarities) of the Wechsler Adult Intelligence Scale, WAIS-III (Wechsler, 1997). Dyslexics were significantly poorer than controls in all reading and spelling measures, both in accuracy and in rate, as shown in Table 1. Yet, while accuracy was variable within the dyslexic group, speed was consistently slower among the dyslexics, with almost no overlap with the control group (see Shaywitz, 1998). Interestingly, dyslexics’ reading rate was just as impaired for words as for non-words. Relative reading rate improved only when words were read in context, as part of a whole paragraph. Measures of short-term verbal memory (digit span), visual symbol memorization and eye–hand coordination (digit symbol-coding), known to be impaired specifically among dyslexics (Mishra et al., 1985; Swanson, 1994; Vargo et al., 1995), were significantly poorer in our dyslexic group compared with those of controls. General cognitive abilities typically used to match controls and dyslexics (block design and similarities, subtests of WAIS-III) did not differ between groups (Table 1). Although all the participants had fulfilled high school requirements, the dyslexics’ average performance on these subtests was somewhat poorer than that of the controls. Also, performance on the form completion test, as measured by the Standard Raven Progressive Matrices, was significantly poorer in the dyslexic group. This test may constitute a specific difficulty for some dyslexics, since verbalization of steps towards a solution substantially aids in solving the problems posed by the more difficult test items (Carpenter et al., 1990). Since we were interested in the relationship between cognitive abilities and contrast sensitivity, we did not exclude poorer performers from our test population. This aspect of our study is addressed in the next section.

Stimuli and experimental design We replicated, with minor differences, the stimuli and behavioural procedures of two previous studies that assessed contrast sensitivity in dyslexia. In one study (Borsting et al., 1996), whose equivalent in our replication was termed ‘T-Drift’ (T indicates a temporal judgement), a drifting vertical sinusoidal grating was used. Detection at several spatial frequencies (0.5, 1, 2, 4, 8 or 12 c/°) was assessed separately for 1 and 10 Hz drifting gratings (mean luminance 20.7 cd/m2). Observers sat 150 cm from the monitor, and the

1384

G. Ben-Yehudah et al. Table 1 Participant characteristics: mean (⫾ standard deviation)

(A) Reading measures Age (years) Reading accuracy (% errors)* Non-word Word Reading rate (words/min)† Non-word Word Paragraph‡ Spelling (no. of errors/24) Orthographic pattern recognition§ Duration (s) Accuracy (no. of errors/69) Phonemic awareness¶ Spoonerism (no. of errors/20) (B) Cognitive measures Raven-SPM (no. of errors/60)# WAIS-III (scaled scores/19)** Digit symbol-coding Digit span Block design Similarities

Dyslexic (n ⫽ 38)

Control (n ⫽ 42)

22.3 (4.2)

22.5 (4.1)

0.3

38.2 (18.3) 12.8 (9)

12.5 (10.2) 2.2 (3.6)

–7.7 –6.8

0.0001 0.0001

28.8 (12.4) 50.9 (21.5) 91.0 (25.4) 4.6 (4.7)

56.7 (13.4) 96.9 (21) 130.5 (15) 0.2 (1)

9.7 9.6 8.3 –5.6

0.0001 0.0000 0.0001 0.0001

130.9 (108.9) 2.3 (3.5)

51.7 (8.2) 0.1 (0.5)

–4.5 –3.8

0.0001 0.0006

7.6 (5.5)

3 (0.5)

–5.4

0.0001

8.1 (5.9)

4.4 (3.5)

–3.4

0.0012

8.5 (2.2) 7.6 (2.2) 11.4 (3.3) 13.4 (2.5)

11.8 (3.2) 11 (3.1) 12.5 (3.1) 13.6 (1.9)

4.8 5 1.4 0.3

0.0001 0.0000 n.s. n.s.

T

P value n.s.

Significance levels are for two-tailed t test for heteroscedastic samples. n.s. ⫽ not significant. *Reading accuracy for single words and non-words aloud; †speed of reading single words and non-words aloud; ‡speed of reading a paragraph aloud; §a lexical decision task between words and their pseudohomophones; ¶switching the first phoneme in the first word with the first phoneme in the second word in a word pair; #accuracy of selecting the correct missing section of a matrix; **four subtests of the WAIS-III test.

stimulus subtended 12.5° ⫻ 9°. Subjects made a twoalternative T-FC judgement, indicating which of the two trial intervals contained the stimulus (uniform mean luminance was used in the other interval). Each trial was composed of two 500 ms intervals, separated by a 500 ms inter-stimulus interval (ISI), during which the screen was of uniform mean luminance. Each interval was demarcated by a tone. In the second study (Spinelli et al., 1997), a small circular patch (2.5° diameter) containing a horizontal sinusoidal grating was presented 0.28° above or below the centre of the screen. Contrast sensitivity was assessed for both transient (flickering gratings) and sustained (stationary gratings) stimuli. In the transient condition, whose equivalent in our replication was termed ‘S-Flicker’ (S indicates spatial judgement), contrast sensitivity was determined for 0.5 c/° gratings, flickering at 5, 10, 17 or 25 Hz (square wave). In the sustained condition, termed here ‘S-Sustained’, sensitivity was determined for stationary gratings of 1, 2, 4, 8 and 16 c/°. Viewing distance was 150 cm (mean luminance 15.3 cd/m2). In both conditions, stimulus duration was 2 s and its appearance was demarcated by a tone. Subjects performed a two-alternative S-FC judgement, indicating whether the stimulus appeared in the upper or lower part of the screen. The parameters used in the three conditions of this

experiment (T-Drift, S-Flicker and S-Sustained) are summarized in Table 2. In each of these three conditions, the various spatial/ temporal frequencies were presented in mixed pseudorandom order. An adaptive staircase procedure was used to assess thresholds. Stimulus contrast was increased (1 dB) following an incorrect response, and decreased following two consecutive correct responses. The next trial was initiated 1 s after a response was given. If no response was given within 2 s following stimulus termination, the trial was ignored and the next trial was initiated. An experimental assessment was terminated when 50 trials were completed for each test frequency. Detection threshold (% contrast) was calculated by averaging the last five reversals. Data are presented in the conventional form plotting sensitivity (1/threshold). Significance levels of group differences between dyslexics and controls were assessed with planned comparisons using a two-tailed t test for independent samples. Differences were considered significant in Experiments 1 and 2 for P ⬍ 0.05, and marginally significant for 0.05 ⬍ P ⬍ 0.1. All stimuli were presented on a 17 inch Trinitron Multiscan II monitor, using a VSG graphics card (VSG software version 5.02, Cambridge Research Systems). Subjects responded using a response box (CB3, Cambridge Research Systems). All experiments were conducted in a dark room and began after a few minutes of dark adaptation.

Contrast sensitivity in dyslexia

1385

Table 2 Summary of test conditions Test condition Experiment1 T-Drift S-Flicker S-Sustained Experiment 2 T-Flicker T-Drift II TL-Drift S-Drift

Type of judgement

Spatial frequency (c/°)

Temporal frequency (Hz)

Stimulus duration; SI (ms)

Temporal Spatial Spatial

0.5, 1, 2, 4, 8, 12 0.5 1, 2, 4, 8, 16

1, 10 5, 10, 17, 25 –

500; 500 2000; – 2000; –

Temporal Temporal Temporal Spatial

0.5 0.5, 8 0.5 0.5, 8

10 10 10 –

1000; 500; 500; 500;

500 500 1000 –

T, a T-FC paradigm; S, an S-FC paradigm; T-Drift II, a second assessment of T-Drift; TL, a T-FC paradigm with a long ISI.

Results Dyslexics’ contrast sensitivity, as measured in the 1 and 10 Hz T-Drift conditions, was poorer than that of the control subjects (Fig. 1), consistent with the findings of Borsting and colleagues (Borsting et al., 1996). As expected from a magnocellular deficit, mostly marginal differences were found for the 1 Hz drift (Fig. 1A), whereas significant differences were found for the 10 Hz drift (Fig. 1B). A repeated measures ANOVA (analysis of variance) was used to examine the effect of subject group and spatial frequency on contrast sensitivity. In the 1 Hz drift, there was only a marginally significant effect for group [F(1,78) ⫽ 2.77, P ⫽ 0.1], whereas in the 10 Hz drift a highly significant group effect was found [F(1,78) ⫽ 9.38, P ⬍ 0.005]. No interaction was found between group and spatial frequency, for either the 1 or 10 Hz drift. As illustrated in Fig. 1B (upper graph), significant group differences were found for both low {0.5 c/° [t(78) ⫽ 2.34, P ⫽ 0.02] and 1 c/° [t(78) ⫽ 2.83, P ⬍ 0.01]} and high {8 c/° [t(78) ⫽ 3.03, P ⬍ 0.005] and 12 c/° [t(78) ⫽2.44, P ⫽ 0.02]} spatial frequencies, though not for intermediate ones. The log ratio between the contrast sensitivity of control and dyslexic subjects reveals that the largest quotient was for the highest spatial frequencies, as shown in Fig. 1B (lower graph). Note that the higher the spatial frequency, maintaining 10 Hz across spatial frequencies results in slower drift velocities (e.g. 1.25 °/s for 8 c/°). Dyslexics’ contrast sensitivity measured under the two spatial judgement conditions did not differ from that of controls, either for sustained (S-Sustained) or for transient (S-Flicker) stimuli, as shown in Fig. 2. This result is consistent with findings by Spinelli and colleagues (Spinelli et al., 1997). Interestingly, a marginally significant difference was found for transient stimuli at low and medium spatial frequencies {5 Hz [t(78) ⫽ 1.87, P ⫽ 0.065] and 17 Hz [t(78) ⫽ 1.72, P ⫽ 0.089]} respectively,. Thus, using the same population of control and dyslexic subjects, we found that dyslexics’ temporal contrast sensitivity was impaired in the T-Drift condition, but was

not impaired when the second transient condition, S-Flicker, was administered.

Experiment 2: Temporal versus spatial judgements Since the first experiment suggested that sequential comparisons may impede dyslexics’ performance, we designed a second experiment to pinpoint possible differences in the behavioural paradigms of the T-Drift and S-Flicker conditions. In the T-Drift condition, large full-screen drifting gratings were applied and subjects were asked to make a temporal judgement. In the S-Flicker condition, smaller (2.5° diameter) patches of flickering gratings were used, and the subjects were asked to make a spatial judgement. Although differences in methodologies spanned both stimuli and behavioural paradigms, we hypothesized that the behavioural paradigm was the crucial factor. Our hypothesis was based on a large body of literature indicating that dyslexics’ shortterm categorical and perceptual memory may be impaired (e.g. Vargo et al., 1995). We reasoned that if the particular difficulty for the dyslexic group in the original T-Drift condition stemmed from the requirement to retain-andcompare a perceptual trace, we should be able to eliminate the group effect if the T-FC judgement was changed to an S-FC judgement (S-Drift). Similarly, changing the S-Flicker condition to a T-FC judgement (T-Flicker) should reveal a group difference. Experiment 2 tested this hypothesis.

Methods Subjects who participated in the first experiment (lasting 4 h in two sessions) were invited for another test session (29 dyslexic and 34 control subjects returned). Four new conditions were designed, as summarized in Table 2. In one, the S-Flicker condition was adapted to a ‘T-Flicker’ condition. The stimulus was now a circular patch of 0.5 c/° horizontal grating, which appeared in the centre of screen at one of two successive intervals (the surrounding screen and the whole

1386

G. Ben-Yehudah et al.

Fig. 1 Contrast sensitivity (mean ⫾ standard error of the mean) of 42 control subjects (circles) and 38 dyslexics (triangles) for detecting drifting gratings, under T-FC conditions (T-Drift). (A) With 1 Hz drift, dyslexics are significantly less sensitive than control subjects at 0.5 and at 4 c/°. (B) For 10 Hz drift, dyslexics are significantly poorer than controls at both low (0.5 and 1 c/°) and high (8 and 12 c/°) spatial frequencies but not at intermediate spatial frequencies (upper plot). The log ratio between contrast sensitivity of control and dyslexic subjects is shown in the lower plot. *P ⬍ 0.05 using a t test; **P ⬍ 0.01.

screen during the ISI was set at the grating’s mean luminance). Detection was now measured only at 10 Hz. Each interval was presented for 1 s with a 500 ms ISI. Other properties of the stimulus were identical to those of the previous S-Flicker condition. The other three conditions were adapted from the T-Drift condition and included the following. (i) A replication of the previous 10 Hz T-Drift termed ‘T-Drift II’, using one high

Fig. 2 Contrast sensitivity (mean ⫾ standard error of the mean) of 42 control subjects (circles) and 38 dyslexics (triangles) for detecting gratings presented in a small circular patch, under S-FC conditions. (A) Detecting sustained gratings (S-Sustained), dyslexics and control subjects do not differ. (B) Detecting flickering gratings (S-Flicker, 0.5 c/°), dyslexics are marginally less sensitive than the control group for 5 and 17 Hz.

and one low spatial frequency (0.5 and 8 c/°). This replication simplified the previous condition, by using only two instead of six different spatial frequencies, thus ensuring that it was not the duration of the assessment (which is long in T-FC paradigms) that made it particularly difficult for some dyslexics. (ii) A variation on T-Drift II (0.5 c/°) which differed from it only in the ISI, which was prolonged from the previous 0.5 s to 1 s (termed ‘TL-Drift’, where L indicates a long ISI). This variation examined whether a longer interval between stimuli would decrease or increase dyslexics’ difficulties. If the source of dyslexics’ difficulties is the temporal proximity between stimuli, which does not enable

Contrast sensitivity in dyslexia

1387

Fig. 3 Contrast sensitivity measured under both T-FC and S-FC paradigms. Dyslexics’ performance (triangles) is similar under both conditions, whereas in the control group (circles) sensitivity improves under T-FC paradigms. Data are presented only for subjects who completed the second experiment (34 controls and 29 dyslexics). Both group averages ⫾ standard error of the mean (filled symbols) and the data for individual subjects (open symbols) are shown. (A) In the S-Flicker condition (0.5 c/°, 10 Hz), there is no difference between dyslexic and control subjects. In the temporal variation of this task (T-Flicker), there is a highly significant difference between the two groups. (B) In the 10 Hz T-Drift condition (0.5 c/°), there is a significant group difference when all the subjects are considered (see Fig. 1), but this difference is only marginally significant when only subjects who also participated in the second experiment are compared (T-Drift II). In the spatial variation of this condition (S-Drift), there is no difference between the groups. When the ISI in the temporal paradigm is increased to 1 s (TL-Drift), there is a highly significant difference between groups, due to the improved contrast sensitivity of the control group. *P ⬍ 0.05 using a t test; **P ⬍ 0.01.

sufficient processing time, then increasing the ISI should improve performance. However, if dyslexics’ difficulties stem from the specific requirements on visual perceptual memory, increasing the ISI may hinder their performance. (iii) A variation of T-Drift II termed ‘S-Drift’, in which subjects

made a spatial comparison, indicating whether the stimulus appeared in the upper or lower half of the screen (using 0.5 and 8 c/°). Stimulus duration was 500 ms. The remaining parameters and experimental procedure for these conditions were identical to those in the previous T-Drift condition.

1388

G. Ben-Yehudah et al.

Results In line with our hypothesis, changing the S-Flicker condition to the T-Flicker paradigm resulted in significantly poorer detection within the dyslexic as compared with the control group [t(55.5) ⫽ 2.8, P ⬍ 0.01]. The data collected at 0.5 c/° for these two conditions (for subjects who completed all contrast sensitivity assessments) are shown in Fig. 3A. As described above, when observers had to indicate whether the flickering gratings appeared in the upper or lower part of the screen, no difference was found between controls and dyslexics. However, in a temporal judgement paradigm, using very similar stimuli, a highly significant difference was found between these same subject groups. This difference stems mainly from the fact that most control subjects performed better in the T-Flicker paradigm than in the S-Flicker paradigm. Dyslexics’ sensitivity was similar in both paradigms. The group difference remains significant [t(58) ⫽ 2.14, P ⫽ 0.03] even when the three control subjects with the highest contrast sensitivity were removed from the analysis. Similarly, when the 10 Hz T-Drift condition was adapted to the S-Drift paradigm, no difference was found between the control and dyslexic groups, as illustrated in Fig. 3B. The replication of the previous T-Drift paradigm (T-Drift II) at 0.5 c/° yielded a marginally significant difference [t(61) ⫽ 1.96, P ⫽ 0.055]; note that only some of the participants in the first two sessions returned for an additional session. This difference became highly significant in the TL-Drift condition, where the ISI was prolonged to 1 s [t(60) ⫽ 2.68, P ⬍ 0.01]. Here too, extending the ISI improved detection of most controls {the averages of 567.4 and 645.4 for the SDrift and TL-Drift, respectively, differ significantly [t(29) ⫽ 2.44, P ⫽ 0.02]}, but had no similar effect on dyslexics’ contrast sensitivity. Changing from an S-FC paradigm to a T-FC paradigm had a differential effect on performance for dyslexic and control subjects. Figure 4 shows, for each participant, the log ratio between contrast sensitivity in T-FC compared with S-FC, for the Drift and Flicker conditions. In these conditions, control subjects tended to perform better in the T-FC paradigms, with an average ratio (temporal/spatial) of 1.2 and 1.5, respectively. This group effect was highly significant in the Flicker condition [t(61) ⫽ 3.12, P ⬍ 0.005], where dyslexic participants were scattered with an average ratio near zero (no difference). The majority of dyslexics did not utilize the advantage afforded to most control participants by the T-FC paradigms. Interestingly, this advantage was significant even though stimulus duration was quite long in both paradigms (2 s for S-Flicker and 1 s for T-Flicker). The results of our modifications of the S-FC and T-FC paradigms show that the differences in patterns of results reported by Borsting and colleagues and by Spinelli and colleagues (Borsting et al., 1996; Spinelli et al., 1997) can be accounted for by the difference in the behavioural paradigms used. In our replication, the different findings

Fig. 4 The log ratio between contrast sensitivities measured in T-FC and in S-FC paradigms for individual subjects. Ratios are shown for both the Drift (temporal-long/spatial) and the Flicker (temporal/spatial) conditions for 0.5 c/° modulated at 10 Hz. Performance is shown both for individual subjects (open symbols) and for group averages ⫾ standard error of the mean (filled symbols). **P ⬍ 0.005 using a t test.

did not arise from different populations. Using flickering compared with drifting gratings was not crucial either. The main factor affecting group differences was the relative advantage of normal readers, as compared with dyslexics, in T-FC paradigms. Even though only detection was required explicitly in this paradigm, an accurate comparison improved thresholds. Since sensitivity in the T-FC paradigm improves when stimuli are retained and compared accurately, dyslexics are faced with a particular disadvantage.

Drifting compared with flickering gratings Do the same individuals have difficulties on both the drifting and flickering gratings? Figure 5A illustrates Spearman’s (rank) correlation between sensitivities measured in T-Flicker and T-Drift II, administered in the last test session. In both cases, observers made a T-FC judgement, indicating which interval contained the 0.5 c/° grating modulated at 10 Hz. This correlation was highly significant (Spearman, r ⫽ 0.4, P ⬍ 0.001). Typically, observers were either good or poor detectors in both paradigms, although a few dyslexic and a few control subjects show discrepancies, performing either T-Flicker or T-Drift II substantially better. On the other hand, the sensitivities measured in the S-Flicker and T-Flicker conditions, illustrated in Fig. 5B, are not correlated (Spearman, r ⫽ 0.03, n.s.). Both paradigms used very similar stimuli (although not completely overlapping spatially), but in one condition subjects made an S-FC judgement, while, in the other, subjects made a T-FC judgement. Rank correlations calculated separately for each group

Contrast sensitivity in dyslexia

1389

of the dyslexic group is an outcome of their inability to utilize the advantage afforded to most controls in a sequential stimulus presentation. While the structure of the paradigm is important, stimulus characteristics are also relevant. For example, as shown in Fig. 5C, even within a single assessment condition (T-Drift II), the correlation between detection at 0.5 and at 8 c/° was nearly zero (Spearman, r ⫽ 0.11, n.s.), both when all of the subjects were considered and when data for each group were analysed separately. Interestingly, detection at 0.5 c/° is the exception, since detection thresholds at the higher spatial frequencies (1 c/° and higher) were significantly intercorrelated.

The dependence of sensitivity on subject selection To what extent did our subject selection criteria (or sampling bias) affect our findings? Since this was one of the main issues in this study, we characterized our participants’ cognitive and reading abilities, enabling examination of the relationships between each of the following factors and contrast sensitivity.

Specific type of reading difficulty

Fig. 5 Correlations between sensitivities measured under different stimuli and behavioural conditions in the last test session. (A) Sensitivities measured under the two T-FC paradigms using drifting (T-Drift II) and flickering gratings (T-Flicker) are significantly correlated [Spearman (rank), r ⫽ 0.4, P ⬍ 0.001]. Only a few individuals differ substantially in their relative performance on these two tasks. (B) The correlation between sensitivity measured in the S-Flicker and T-Flicker conditions is nearly zero although stimulus characteristics are almost the same. (C) Detection for the lowest spatial frequency (0.5 c/°) is not correlated with detection of high spatial frequencies although both were tested within the same assessment using the same procedure. A higher rank order indicates poorer performance in all plots.

yielded a similar trend of results for the control group (T-Flicker and T-Drift II: r ⫽ 0.45, P ⬍ 0.01; T-Flicker and S-Flicker: n.s.), whereas neither correlation was significant for the dyslexic group. This difference between groups is consistent with the observation that the relative impairment

Previous studies (Borsting et al., 1996; Ridder et al., 1997; Spinelli et al., 1997; Slaghuis and Ryan, 1999) suggested that only one subtype of dyslexics, i.e. dysphoneidetics, who have difficulties in reading non-words, reading irregular words and in spelling (Flynn and Boder, 1991), are consistently poorer in their temporal contrast sensitivity. Dyslexics who have specific difficulties in reading non-words (dysphonetics) have a broad range of contrast sensitivities, and dyslexics who have specific difficulties in reading and spelling irregular words (dyseidetics) have normal contrast sensitivity. Applying this division to subtypes of Israeli dyslexics was difficult, since there are no formal tests for assessing these subtypes in Hebrew. Adapting the concept underlying Boder’s division (Boder, 1971, 1973) into Hebrew, we defined subtypes according to the discrepancy between a person’s ability in reading non-words and his/her orthographic skills (assessed by spelling and the lexical decision task). Two standard deviations between these measures was set as a threshold for determining a large discrepancy. We thus defined three subtypes according to their relative pattern of reading non-words compared with orthographic abilities, and a fourth subtype as having only mild difficulties (within 2 SD from controls in both reading and orthographic abilities). Table 3 shows reading and cognitive scores of the different dyslexia subtypes which we defined. Contrast sensitivity did not differ consistently across these subgroups. For example, in the 10 Hz T-Drift condition, a repeated measures ANOVA (analysis of variance) that examined the effect of dyslexia subtype and spatial frequency on contrast sensitivity showed

1390

G. Ben-Yehudah et al. Table 3 Dyslexic subtype characteristics: mean (⫾ standard deviation)

(A) Reading measures Age (years) Reading accuracy (% errors) Non-word Word Reading rate (words/min) Non-word Word Paragraph Spelling (no of errors/24) Orthographic pattern recognition Duration (s) Accuracy (no. of errors/69) Phonemic awareness Spoonerism (no. of errors/20) (B) Cognitive measures Raven-SPM (no. of errors/60) WAIS (scaled scores/19) Digit symbol-coding Digit span Block design Similarities

Dysphonetic Dyseidetic (n ⫽ 8) (n ⫽ 12)

Dysphoneidetic (n ⫽ 9)

Mild deficits (n ⫽ 9)

22.4 (5.8)

22 (3.8)

23.1 (4.4)

21.9 (3.7)

46.5 (20.1) 16 (7.7)

38.7 (16.6) 14.5 (10.7)

46.6 (14.8) 15.9 (6.2)

21.9 (12) 4.3 (3.4)

18.7 (4.1) 44.4 (12.7) 87.2 (17.5) 1.9 (1.4)

31.2 (10) 38.7 (8.2) 79.3 (13.3) 8.9 (4.8)

22.4 (5.7) 46.7 (14.4) 86.9 (36.1) 5.6 (3.6)

40.5 (14.6) 77.1 (25.2) 114.2 (18.3) 0.3 (0.5)

85 (10.9) 0.5 (0.5)

228.6 (153.5) 103.2 (25.3) 5.2 (4.9) 1.9 (1.4)

69.2 (12.9) 0.2 (0.4)

5.9 (3)

7.7 (5.7)

11.9 (6.9)

4.8 (2.6)

13.4 (7.2)

7.7 (4.7)

7 (6.1)

5 (2.7)

7.5 (2.2) 6.8 (2.1) 9 (3.3) 11.3 (2.1)

9 (2.5) 7.4 (1.6) 10.6 (3.2) 12.7 (3)

8.6 (2.2) 7.7 (2.7) 12.6 (2.7) 14.3 (1.4)

9 (2.1) 8.7 (2.5) 13 (3.2) 15.3 (1.4)

See footnotes to Table 1.

a significant effect for spatial frequency [F(5,30) ⫽ 130.1, P ⬍ 0.001], but no effect for dyslexia subtype [F(3,34) ⫽ 0.95, P ⫽ 0.43] or for the interaction between these factors. Figure 6 illustrates the mean contrast sensitivity for these subgroups in the 10 Hz T-Drift condition (experiment 1) and in the various T-FC conditions tested in the second experiment. Dysphonetics had both the lowest non-word reading rates and the poorest cognitive abilities, yet their contrast sensitivity was not lower than that of the other subtypes. In fact, contrast sensitivity of three of the four subgroups (dysphonetics, dyseidetics and dyslexics with mild reading deficits) was poorer than that of controls in 10 Hz T-Drift, as illustrated in Fig. 6A. Only dysphoneidetics’ contrast sensitivity did not differ from that of controls. However, they were poorer than controls in detecting flickering gratings, as illustrated in Fig. 6B. Thus, subtyping dyslexics according to their specific pattern of phonological and orthographic difficulties did not yield any distinct population with or without difficulties in detecting transient gratings. Applying other related criteria for defining dysphonetics and dyseidetics yielded similar negative results.

Gustafson and Samuelsson, 1999), contrast sensitivity may still differ between these groups. We thus compared the temporal contrast sensitivity of the individuals in our test population, who would be defined as dyslexics by the most stringent intelligence criteria (0–4 errors in the RavenSPM test, 12 dyslexics and 24 controls), with individuals with reading difficulties and average cognitive abilities (5– 9 errors in the Raven-SPM test, 14 dyslexics and 18 controls). Each reading-disabled subgroup was compared with their age- and intelligence-matched controls. Poor readers with above average intelligence measures (dyslexics, given traditional definitions) were significantly less sensitive than their matched control subgroup in the temporaljudgement conditions {[t(16) ⫽3.52, P ⬍ 0.005], [t(16) ⫽ 2.89, P ⬍ 0.01]; respectively, T-Flicker and TL-Drift}. Interestingly, individuals with reading difficulties and average cognitive abilities did not have poorer contrast sensitivity than their matched control group. Contrast sensitivity of the remaining individuals with reading difficulties (12 participants with the lowest Raven-SPM scores) was poor, with respect to both the control subgroups and the other dyslexic subgroups for most of the transient stimuli, though not for the sustained stimuli.

Cognitive abilities This distinction is related to traditional definitions of dyslexics (DSM-IV, 1994) having at least average intelligence compared with other poor readers who have additional cognitive difficulties. While this definition has been challenged recently since both groups suffer from poor phonological abilities (Siegel, 1992; Lyon, 1995;

Severity of reading difficulty Contrast sensitivity was not correlated with reading scores, either within the control or within the dyslexic group (e.g. Spearman’s correlation between reading accuracy of nonwords and contrast sensitivity in 10 Hz T-Drift was not significant for any spatial frequency; in all conditions r ⬍ 0.16

Contrast sensitivity in dyslexia

1391

and r ⬍ 0.15, for the control and dyslexic groups, respectively). In fact, as illustrated in Fig. 6, dyslexics who currently experience only mild reading deficits were significantly poorer than controls in detecting both drifting and flickering temporally judged gratings.

Discussion Summary of results In the first experiment, we replicated the experimental design of two previous studies and reproduced their differing results. Using drifting gratings presented in a T-FC paradigm (replicating Borsting et al., 1996), dyslexics’ contrast sensitivity was poorer than that of controls. Consistent with the magnocellular hypothesis, dyslexics’ relative sensitivity was poorer for 10 Hz compared with 1 Hz drift. However, when flickering gratings at various temporal frequencies were used in an S-FC paradigm (replicating Spinelli et al., 1997), no difference was found between the two groups. In the second experiment, we tested the hypothesis that the difference in relative sensitivities stemmed from dyslexics’ specific difficulties in retaining and comparing accurate visual traces. We thus designed a T-FC version of the flickering gratings previously administered in an S-FC paradigm, and an S-FC version of the drifting gratings previously administered in a T-FC paradigm. Indeed, when tested in a T-FC paradigm, dyslexics were poorer in detecting both the drifting and flickering gratings. Poor temporal contrast sensitivity for drifting and flickering gratings was not related to more severe reading deficits, or to lower cognitive abilities. Neither could we relate poor sensitivity to a specific pattern of reading difficulties, phonological or orthographic.

Relation to previous studies assessing contrast sensitivity in dyslexia A central issue addressed in the current study was the extent to which inter-study variability can be accounted for by differences in behavioural methodologies, an aspect that previously was ignored. We found that the specific methodology applied to assess a fundamental perceptual property was more crucial than differences in sampling the dyslexic population. In retrospect, the finding that paradigms involving temporal comparisons are difficult for dyslexics may not be surprising. However, methodologies are often left unattended. The present findings stress the complexities of the psychophysical trait. Optimal performance, even on the simplest task, involves incredibly complex coordinated neuronal activity. Thus, when a single assessment procedure is used, reduced sensitivity to the stimuli and reduced ability to perform on a given test paradigm are confounded. To what extent do our findings account for the various results reported in the literature? Our results are consistent with findings by several authors (Martin and Lovegrove, 1987; Demb et al., 1998; Slaghuis and Ryan, 1999), who

Fig. 6 Contrast sensitivity of controls (circles) and the different subtypes of dyslexics (as defined by Table 3). All subtype groups showed poorer performance than that of controls. (A) In the 10 Hz T-Drift condition, dyseidetics (square), dysphonetics (triangle) and dyslexics with mild reading deficits (diamond) were significantly less sensitive than the control group. However, dysphoneidetics’ (cross) contrast sensitivity did not differ from that of controls. (B) In the T-FC conditions tested in the last session, all dyslexic subtypes were poorer than the control group in detecting both drifting and flickering gratings.

applied a T-FC paradigm and found inter-group differences (though Demb and colleagues, comparing small groups, found only marginally significant differences). However, the different pattern of results in some studies cannot be accounted for solely on the basis of the behavioural paradigm used. For example, Walther-Muller and colleagues also applied a T-FC paradigm (1 and 12 c/° gratings modulated at 16.8 Hz), but did not find a group effect (Walther-Muller et al., 1995). Felmingham and Jakobson used an S-FC

1392

G. Ben-Yehudah et al.

paradigm (2 c/° gratings flickering at 5–25 Hz) and did find a significant inter-group difference, which increased with increasing temporal frequencies (Felmingham and Jakobson, 1995). Whether these inter-study differences stem from sampling different types of dyslexia or from additional methodological differences that we did not examine cannot be determined at this point.

Contrast sensitivity as a probe of a magnocellular deficit Within a given behavioural paradigm, dyslexics’ contrast sensitivity tended to be poorer for higher temporal frequencies, in agreement with the magnocellular hypothesis. When a T-FC paradigm was applied, significant group differences were found for the 10 Hz but not the 1 Hz T-Drift condition. Although no significant group difference was found for the spatial judgement tasks, a marginally significant group difference was found in the S-Flicker condition (for 5 and 17 Hz), whereas in the S-Sustained condition the two groups did not differ. However, the dependence on spatial frequencies was not monotonic. Within the 10 Hz T-Drift paradigm, poorer sensitivity was found for low (0.5 and 1 c/°) and high (8 and 12 c/°) but not for intermediate spatial frequencies. This non-monotonic dependence deviates from the magnocellular hypothesis, which predicts that dyslexics’ difficulties will increase consistently when the spatial frequency of the stimulus is decreased. Comparing behavioural paradigms, we found that dyslexics were particularly impaired when assessed in a T-FC design. The implication of this finding for the magnocellular hypothesis is not straightforward due to our limited understanding of the magnocellular role in perceptual memory. Monkey studies ablating magnocellular layers of the LGN that report substantial deficits in temporal contrast sensitivity used either a T-FC (Merigan and Maunsell, 1990) or an S-FC (Schiller et al., 1990) paradigm, but did not compare their relative effects. Although higher visual areas along the dorsal stream play a role in retaining accurate perceptual traces (e.g. Bisley and Pasternak, 2000), the contribution of magnocellular projections to perceptual memory was not assessed separately. Taken together, our results are consistent with a loose magnocellular hypothesis, which emphasizes a special deficit in dyslexics’ ability to retain accurate perceptual traces across brief intervals.

Dyslexia and perceptual memory In a detection task, the subject does not need to compare the sequentially presented stimuli in order to achieve reasonable performance. If a stimulus is detected, it need not be compared with another one. Indeed, reasonable detection was achieved by all our participants. However, a two-alternative forced

choice paradigm affords more sensitive thresholds, gained by accurate comparisons between consecutive traces when lower contrasts are presented and the stimulus is not clearly detected. The significant inter-group differences found under these conditions, despite substantial inter-group overlap, resulted from two factors. First, no dyslexic individual was very good in the conditions requiring sequential comparisons. Secondly, dyslexics’ data points were clustered around lower levels than those of controls. An interesting point is that all the stimulus durations enabled eye movements. Thus, spatial judgements may also have involved temporal comparisons. In the S-FC paradigm, subjects had a ‘second chance’ to examine the stimulus, since the duration of the presentation permitted ‘going back’ to the location previously examined. In a sequential presentation (i.e. T-FC paradigm), however, by the time the second stimulus was presented, the first was no longer visually available. Consequently, subjects had to rely on their initial perceptual trace in deciding which interval contained the stimulus. The mechanisms underlying the ability to compare sequentially presented stimuli accurately are far from understood. In order to compare the response to an incoming stimulus with the response to a previous stimulus, the visual system needs to have a ‘memory cell’ for each ‘perception cell’ (for a review, see Magnussen, 2000). It is not clear what components of the perceptual memory are retained in the purely perceptual areas and what components are held in higher areas, such as the frontal cortex (Haxby et al., 2000). Recent evidence from motion tasks (Bisley and Pasternak, 2000) suggests that both high and low level areas are involved in retaining and comparing the perceptual trace. Thus, the behavioural deficits revealed in this study may be the outcome of an impairment at any site within these large-scale neuronal loops.

Dyslexia and the hypothesis of ‘fast temporal processing’ A prominent theory proposed to describe dyslexics’ perceptual deficits is the ‘fast temporal processing’ hypothesis (Tallal, 1980). This hypothesis, originally based on psychoacoustic results, asserts that dyslexics have difficulties in correctly identifying rapid streams of brief signals. The magnocellular hypothesis is a concrete suggestion for how a deficit in ‘fast temporal processing’ may be manifested in the visual system (Stein and Walsh, 1997). In the current study, we examined detection and not stimulus identification. We found that dyslexics’ detection of high temporal frequencies (25 Hz) was unimpaired, when tested in an S-FC paradigm. Interestingly, these results are consistent with psychoacoustic results when temporal resolution was the only factor measured. Thus, for example, the minimal duration of silence needed for a ‘gap’ to be detected in a continuous noise was not longer in dyslexics

Contrast sensitivity in dyslexia (McAnally and Stein, 1996; Ahissar et al., 2000). Difficulties were observed when responses depended on the quality of the perceived stimulus and not when subjects had to detect its presence. Several studies have suggested that the quality of both visual (Lovegrove and Brown, 1978; Baddock and Lovegrove, 1981; Di Lollo et al., 1983) and auditory (Helenius et al., 1999) stimuli is more degraded for dyslexics than for controls when brief ISIs are used. Nevertheless, deficits in the quality of the perceived stimulus (especially when it has to be retained and compared) do not seem to be specific to the range of tens of milliseconds. Deficits in processing both auditory (e.g. Ahissar et al., 2000) and visual (e.g. Hari et al., 1999) stimuli have also been found when several hundreds of milliseconds separated sequentially presented stimuli. In the current study too, deficits were found for both 0.5 and 1 s ISIs (using 0.5 s stimulus durations) when retaining an accurate trace was important for optimal performance (e.g. in a T-FC paradigm). The finding that the quality of the perceived stimulus is impaired for both the range of tens and hundreds of milliseconds can be interpreted in two different ways. One interpretation, consistent with a fast temporal processing deficit, is that hundreds of milliseconds are still too brief. In this case, a further increase in duration and/or interval (to ⬎1 s) might eliminate dyslexics’ difficulties. The second interpretation is that perceptual memory is mainly impaired. Here, longer ISIs should degrade dyslexics’ performance further. Dissociating these alternatives, by a systematic assessment of different ISIs, will be the focus of our future research. In summary, this study shows that dyslexic adults have difficulty in detecting temporally modulated gratings when stimuli are presented sequentially. Limited ability to retainand-compare perceptual traces may be functionally related to poor reading. Reading in this respect is a special visual task, in which the integration of sequentially identified elements is crucial for success and, as such, its failure may be an outcome of the same underlying deficits.

Acknowledgements We wish to thank Avital Deutch for permission to use the tests she designed for reading single words and non-words, Ilana Ben-Dror and Sharon Peleg for the spoonerism test they designed, and Shaul Hochstein for helpful comments on the manuscript. This study was supported by grants from the Israel Foundation Trustees, The Institute for Psychobiology in Israel and the Israeli Science Foundation.

References Ahissar M, Protopapas A, Reid M, Merzenich MM. Auditory processing parallels reading abilities in adults. Proc Natl Acad Sci USA 2000; 97: 6832–7.

1393

American Psychiatric Association. Diagnostic and statistical manual of mental disorders. (DSM-IV). 4th edn. Washington (DC): American Psychiatric Association; 1994. Badcock D, Lovegrove W. The effects of contrast, stimulus duration, and spatial frequency on visible persistence in normal and specifically disabled readers. J Exp Psychol Hum Percept Perform 1981; 7: 495–505. Ben-Yehudah G, Sackett E, Malchi-Ginzberg L, Ahissar M. Contrast sensitivity in dyslexia [abstract]. Perception 2000; 29 Suppl: 36. Bisley JW, Pasternak T. The multiple roles of visual cortical areas MT/MST in remembering the direction of visual motion. Cereb Cortex 2000; 10: 1053–65. Boder E. Developmental dyslexia: prevailing diagnostic concepts and a new diagnostic approach. In: Myklebust HR, editor. Progress in learning disabilities, Vol. II. New York: Grune and Stratton; 1971. p. 293–321. Boder E. Developmental dyslexia: a diagnostic approach based on three atypical reading–spelling patterns. Dev Med Child Neurol 1973; 15: 663–87. Borsting E, Ridder WH 3rd, Dudeck K, Kelley C, Matsui L, Motoyama J. The presence of a magnocellular defect depends on the type of dyslexia. Vision Res 1996; 36: 1047–53. Carpenter PA, Just MA, Shell P. What one intelligence test measures: a theoretical account of the processing in the Raven Progressive Matrices Test. Psychol Rev 1990; 97: 404–31. Castles A, Coltheart M. Varieties of developmental dyslexia. Cognition 1993; 47: 149–80. Cornelissen P. Fixation, contrast sensitivity and children’s reading. In: Wright SF, Groner R, editors. Facets of dyslexia and its remediation. Amsterdam: Elsevier; 1993. p. 139–62. Cornelissen P, Richardson A, Mason A, Fowler S, Stein J. Contrast sensitivity and coherent motion detection measured at photopic luminance levels in dyslexics and controls. Vision Res 1995; 35: 1483–94. Demb JB, Boynton GM, Best M, Heeger DJ. Psychophysical evidence for a magnocellular pathway deficit in dyslexia. Vision Res 1998; 38: 1555–9. Di Lollo V, Hanson D, McIntyre JS. Initial stages of visual information processing in dyslexia. J Exp Psychol Hum Percept Perform 1983; 9: 923–35. Eden GF, Stein JF, Wood HM, Wood FB. Temporal and spatial processing in reading disabled and normal children. Cortex 1995; 31: 451–68. Evans BJW, Drasdo N, Richards IL. An investigation of some sensory and refractive visual factors in dyslexia. Vision Res 1994; 34: 1913–26. Felmingham KL, Jakobson LS. Visual and visuomotor performance in dyslexic children. Exp Brain Res 1995; 106: 467–74. Flynn JM, Boder E. Clinical and electrophysiological correlates of dysphonetic and dyseidetic dyslexia. In: Stein JF, editor. Vision and visual dyslexia. Boca Raton (FL): CRC Press; 1991. p. 121–31.

1394

G. Ben-Yehudah et al.

Gross-Glenn K, Skottun BC, Glenn W, Kushch A, Lingua R, Dunbar M, et al. Contrast sensitivity in dyslexia. Vis Neurosci 1995; 12: 153–63. Gustafson S, Samuelsson S. Intelligence and dyslexia: implications for diagnosis and intervention. Scand J Psychol 1999; 40: 127–34. Hari R, Valta M, Uutela K. Prolonged attentional dwell time in dyslexic adults. Neurosci Lett 1999; 271: 202–4. Haxby JV, Petit L, Ungerleider LG, Courtney SM. Distinguishing the functional roles of multiple regions in distributed neural systems for visual working memory. [Review]. Neuroimage 2000; 11: 380–91. Hayduk S, Bruck M, Cavanagh P. Low-level visual processing skills of adults and children with dyslexia. Cogn Neuropsychol 1996; 13: 975–1015.

Maunsell JH, Nealey TA, DePriest DD. Magnocellular and parvocellular contributions to responses in the middle temporal visual area (MT) of the macaque monkey. J Neurosci 1990; 10: 3323–34. McAnally KI, Stein JF. Auditory temporal coding in dyslexia. Proc R Soc Lond B Biol Sci 1996; 263: 961–5. Merigan WH, Maunsell JH. Macaque vision after magnocellular lateral geniculate lesions. Vis Neurosci 1990; 5: 347–52. Merigan WH, Maunsell JH. How parallel are the primate visual pathways? [Review]. Annu Rev Neurosci 1993; 16: 369–402. Merigan WH, Byrne CE, Maunsell JH. Does primate motion perception depend on the magnocellular pathway? J Neurosci 1991a; 11: 3422–9.

Helenius P, Uutela K, Hari R. Auditory stream segregation in dyslexic adults. Brain 1999; 122: 907–13.

Merigan WH, Katz LM, Maunsell JH. The effects of parvocellular lateral geniculate lesions on the acuity and contrast sensitivity of macaque monkeys. J Neurosci 1991b; 11: 994–1001.

Laasonen M, Tomma-Halme J, Lahti-Nuuttila P, Service E, Virsu V. Rate of information segregation in developmentally dyslexic children. Brain Lang 2000; 75: 66–81.

Mishra S, Ferguson B, King P. Research with the Wechsler Digit Span subtest: implications for assessment. Sch Psychol Rev 1985; 14: 37–47.

Lehmkuhle S, Garzia RP, Turner L, Hash T, Baro JA. A defective visual pathway in children with reading disability. N Engl J Med 1993; 328: 989–96.

Raven JC, Court JH, Raven J. Manual for Raven’s Progressive Matrices and Vocabulary Scales. Oxford: Oxford Psychologists Press; 1992.

Livingstone MS, Rosen GD, Drislane FW, Galaburda AM. Physiological and anatomical evidence for a magnocellular defect in developmental dyslexia. Proc Natl Acad Sci USA 1991; 88: 7943–7.

Ridder WH, Borsting E, Cooper M, Mcneel B, Huang E. Not all dyslexics are created equal. Optom Vis Sci 1997; 74: 99–104.

Lovegrove W, Brown C. Development of information processing in normal and disabled readers. Percept Mot Skills 1978; 46: 1047–54. Lovegrove WJ, Bowling A, Badcock D, Blackwood M. Specific reading disability: differences in contrast sensitivity as a function of spatial frequency. Science 1980; 210: 439–40. Lovegrove W, Martin F, Bowling A, Blackwood M, Badcock D, Paxton S. Contrast sensitivity functions and specific reading disability. Neuropsychologia 1982; 20: 309–15.

Schiller PH, Logothetis NK. The color-opponent and broad-band channels of the primate visual system. [Review]. Trends Neurosci 1990; 13: 392–8. Schiller PH, Logothetis NK, Charles ER. Functions of the colouropponent and broad-band channels of the visual system. Nature 1990; 343: 68–70. Shalem Z, Lachman D. Diagnostic Battery for Reading Process in Hebrew. Nitzan: The Israeli Association for the Advancement of Children with Learning Disabilities; 1998. Shapley R. Visual sensitivity and parallel retinocortical channels. [Review]. Annu Rev Psychol 1990; 41: 635–58.

Lyon GR. Toward a definition of dyslexia. Ann Dyslexia 1995; 45: 3–27.

Shapley R, Perry VH. Cat and monkey retinal ganglion cells and their visual functional roles. Trends Neurosci 1986; 9: 229–35.

Magnussen S. Low-level memory processes in vision. [Review]. Trends Neurosci 2000; 23: 247–51.

Shaywitz SE. Dyslexia. [Review]. N Engl J Med 1998; 338: 307–12.

Manis FR, Seidenberg MS, Doi LM, McBride-Chang C, Petersen A. On the bases of two subtypes of developmental [corrected] dyslexia. Cognition 1996; 58: 157–95. Martin F, Lovegrove W. Flicker contrast sensitivity in normal and specifically disabled readers. Perception 1987; 16: 215–21.

Siegel LS. An evaluation of the discrepancy definition of dyslexia. J Learn Disabil 1992; 25: 618–29. Skottun BC. The magnocellular deficit theory of dyslexia: the evidence from contrast sensitivity. [Review]. Vision Res 2000; 40: 111–27.

Martin F, Lovegrove WJ. Uniform-field flicker masking in control and specifically-disabled readers. Perception 1988; 17: 203–14.

Slaghuis WL, Ryan JF. Spatio-temporal contrast sensitivity, coherent motion, and visible persistence in developmental dyslexia. Vision Res 1999; 39: 651–68.

Mason A, Cornelissen P, Fowler S, Stein J. Contrast sensitivity, ocular dominance and specific reading disability. Clin Vis Sci 1993; 8: 345–53.

Spinelli D, Angelelli P, De Luca M, Di Pace E, Judica A, Zoccolotti P. Developmental surface dyslexia is not associated with deficits in the transient visual system. Neuroreport 1997; 8: 1807–12.

Maunsell JH. Functional visual streams. [Review]. Curr Opin Neurobiol 1992; 2: 506–10.

Stein J, Walsh V. To see but not to read; the magnocellular theory of dyslexia. Trends Neurosci 1997; 20: 147–52.

Contrast sensitivity in dyslexia

1395

Swanson HL. Short-term memory and working memory: do both contribute to our understanding of academic achievement in children and adults with learning disabilities? J Learn Disabil 1994; 27: 34–50.

R scores in the diagnosis of dyslexia in children. Percept Mot Skills 1995; 80: 1219–29.

Tallal P. Auditory temporal perception, phonics, and reading disabilities in children. Brain Lang 1980; 9: 182–98.

Wechsler D. Wechsler Adult Intelligence Scale: WAIS-III: administration and scoring manual. San Antonio: Psychological Corporation; 1997.

Ungerleider LG, Mishkin M. Two cortical visual systems. In: Ingle DJ, Goodale MA, Mansfield RJ, editors. Analysis of visual behavior. Cambridge (MA): MIT Press; 1982. p. 549–85. Vargo FE, Grosser GS, Spafford CS. Digit span and other WISC-

Walther-Muller PU. Is there a deficit of early vision in dyslexia? Perception 1995; 24: 919–36.

Received September 11, 2000. Revised January 11, 2001. Accepted March 3, 2001