Temporal acuity in developmental dyslexia across the life span : Tactile, auditory, visual, and crossmodal estimations

(1)

dyslexia across the life span:

Tactile, auditory, visual, and crossmodal estimations

Marja Laasonen

Department of Psychology, University of Helsinki, Finland Finnish Graduate School of Neuroscience - FGSN

Academic dissertation to be publicly discussed, by due permission of the Faculty of Arts at the University of Helsinki in auditorium XII,

on the 13th of November, 2002 at 12 o´clock.

(2)

University of Helsinki Finland

Docent Elisabet Service, Ph.D.

Department of Psychology University of Helsinki

Finland

Reviewers: Docent Teija Kujala, Ph.D.

Department of Psychology University of Helsinki

Finland

Professor Pekka Niemi, Ph.D.

Department of Psychology University of Turku

Finland

Opponent: Professor Timo Ahonen, Ph.D.

Department of Psychology University of Jyväskylä

Finland

ISBN 952-91-5153-5 (nid.)

ISBN 952-10-0731-1 (PDF) (http://ethesis.helsinki.fi/)

Cover design: Ulla Peltonen

Layout: Jaana Lindholm

Cover photograph: Rami Salle Multiprint Oy, Helsinki 2002

(3)

Temporal acuity in developmental dyslexia across the life span:

Tactile, auditory, visual, and crossmodal estimations

Marja Laasonen

University of Helsinki, FIN

Developmental dyslexia is a specific problem of learning to read, often accompanied by writing and spelling difficulties. Dyslexia presumably results from impaired phonological processing, which is either a primary cause, or reflects a more basic deficit, such as impaired perceptual temporal processing. The prevalence of the temporal impairment in dyslexia is not known. It is also unclear whether or not the impairment can be observed at all ages, from childhood to adulthood, or how the difficulties change with age. Further, the relationships between temporal processing, phonological processing, and reading in different language environments have not been resolved.

We assessed cognitive temporal acuity in tactile, auditory, and visual modalities, and their crossmodal combinations (audiotac- tile, visuotactile, audiovisual) in a series of four studies. Two methods were used. In the Temporal Order Judgment (TOJ) method, the participant estimated the order of two short stimuli. In the phase difference detection (Temporal Processing Acuity, TPA) method, the participant judged whether the same short stimuli, now in two parallel trains, were simultaneous or nonsimultane-

(4)

years).

The results of studies I-IV showed that dyslexic children and adults were impaired as compared to fluent readers when assessed with the unimodal and crossmodal TPA and TOJ (the latter assessed only for adults) methods. The temporal acuity impairment was general: it concerned all modalities, their combinations, and assessment methods. Control analyses showed that nontemporal aspects could not explain the impairment. Further- more, temporal acuity was related to phonological processing but not directly to reading. In study IV, increasing adult age was found to decrease unimodal temporal acuity in both reading groups. The decline was more noticeable in the case of dyslexic readers, as compared to the fluent readers when assessed with the TPA method. The reading-related tasks did not indicate similar age-related accelerated decline.

In summary, temporal impairment in dyslexia was general and occurred at every age investigated. The impairment did not improve with increasing age but, on the contrary, aggravated.

Not every dyslexic reader was impaired, and therefore, temporal acuity impairment alone is not a sufficient general cause of developmental dyslexia.

(5)

The thesis is based on the following publications which are referred to in the text by Roman numerals I-IV.

I

Laasonen M, Tomma-Halme J, Lahti-Nuuttila P, Service E, and Virsu V (2000) Rate of information segregation in developmentally dyslexic children, Brain and Language, 75(1), 66-81.

II

Laasonen M, Service E, and Virsu V (2001) Temporal order and processing acuity of visual, auditory, and tactile perception in developmentally dyslexic young adults, Cognitive, Affective, & Behavioral Neuroscience, 1(4), 394-410.

III

Laasonen M, Service E, and Virsu V (2002) Crossmodal temporal order and processing acuity in developmentally dyslexic young adults, Brain and Language, 80(3), 340-354.

IV

Laasonen M, Lahti-Nuuttila P, and Virsu V (2002) Developmentally impaired processing speed decreases more than normally with age, NeuroReport, 13(9),1111-1113.

(6)

2AFC Two alternative forced choice

CV Consonant vowel

ERP Event-related potential FM Frequency modulation

fMRI Functional magnetic resonance imaging Hz Hertz

IQ Intelligence quotient ISI Inter stimulus interval

LGN Lateral geniculate nucleus of thalamus MEG Magnetoencephalography

MGN Medial geniculate nucleus of thalamus MMN Mismatch negativity

MT+ Motion complex

PIQ Performance intelligence quotient QTL Quantitative trait loci

RAS Rapid alternating stimulus naming SD Standard deviation

SEM Standard error of the mean SLI Specific language impairment SOA Stimulus onset asynchrony TOJ Temporal order judgment TPA Temporal processing acuity VIQ Verbal intelligence quotient V1 Primary visual cortex V5/MT Middle temporal area

WAIS-R Wechsler Adult Intelligence Scale Revised WISC-R Wechsler Intelligence Scale for Children Revised WMS-R Wechsler Memory Scale Revised

(7)

1 Introduction

10

1.1 Definition of developmental dyslexia 10

1.2 Frequency 11

1.2.1 Identification method 11

1.2.2 Gender 12

1.3 Relationship to other specific learning difficulties 12

1.4 Development of fluent and nonfluent reading 14

1.4.1 Phonological processing 15

1.4.2 Fluent and nonfluent reading 16

1.4.3 Effect of orthography 18

1.4.4 Adult developmental dyslexia 18

1.4.5 Subtypes of developmental dyslexia 19

1.5 Biological basis of developmental dyslexia 20

1.5.1 Genetic background 20

1.5.2 Anatomical findings and the magnocellular system 22 1.5.2.1 Anatomical findings related to the magnocellular system 23 1.5.2.2 Anatomical findings related to language 25 1.5.3 Magnocellular hypotheses of developmental dyslexia 26 1.5.3.1 Lovegrove & Breitmeyer: Saccadic suppression 26 1.5.3.2 Stein: The magnocellular theory of developmental dyslexia 27

1.5.3.3 Vidyasagar: Attentional spotlight 27

1.5.3.4 Hari: Sluggish Attentional Shifting 28

1.5.3.5 Nicolson & Fawcett: Cerebellar dysfuncion 28 1.5.3.6 Galaburda: Cortical-phonological and thalamic-temporal impairments 29

1.6 Temporal processing and acuity 30

1.6.1 Levels and modalities of processing 30

1.6.2 Cognitive models of simultaneity/nonsimultaneity and order judgment 31 1.6.3 Localization and time-windows of temporal processing 33

1.6.4 Development and decline 34

1.7 Temporal processing in developmental dyslexia 36

1.7.1 Auditory temporal processing 36

1.7.1.1 Interaural temporal processing 36

1.7.1.2 Stimulus individuation 37

1.7.1.3 Temporally modulated tones 37

(8)

1.7.1.6 Order judgment 40

1.7.1.7 Nontemporal tasks 41

1.7.2 Tactile temporal processing 42

1.7.3 Visual temporal processing 42

1.7.3.1 Stimulus individuation 43

1.7.3.2 Order judgment 44

1.7.3.3 ‘Magnosensitive’ tasks 44

1.7.4 Crossmodal temporal processing 48

1.8 Aims of the present studies 49

2 Methods

51

2.1 Participants 51

2.1.1 Children (Study I) 52

2.1.2 Young Adults (Studies II and III) 52

2.1.3 Older Adults (Study IV) 52

2.2 Psychometric methods 53

2.2.1 Wechsler Intelligence Scale Revised subtests 53

2.2.2 Criteria for dyslexia 54

2.2.2.1 Children 54

2.2.2.2 Young adults 55

2.2.2.3 Older adults 55

2.2.3 Reading-related tasks 55

2.2.3.1 Children 55

2.2.3.2 Young and older adults 56

2.3 Temporal acuity tasks 58

2.3.1 Temporal order judgment (TOJ) 60

2.3.2 Temporal processing acuity (TPA) 61

2.3.3 Modalities 62

2.3.4 Exceptions for the children 63

2.4 Statistical analyses 65

3 Results

66

3.1 Reading-related tasks (Studies I-IV) 66

3.2 Group differences in temporal acuity (Studies I-IV) 68 3.2.1 Independence of the crossmodal and unimodal temporal

acuity impairments (Studies I and III) 71

(9)

3.5 Relationships between temporal acuity, intelligence, phonological processing,

and reading (Studies I-IV) 77

3.6 Classification and individuals 82

3.6.1 Individuals with poor temporal acuity 84

3.6.1.1 Dyslexic readers with very poor temporal acuity 84

3.6.1.2 Misclassified fluent readers 88

3.7 Methodological considerations 90

3.7.1 Standard deviation of responses in the temporal acuity tasks 90

3.7.2 Response probability curves 91

3.7.3 Response reliability 91

3.7.4 Threshold reliability 92

3.7.5 Response latency 92

3.7.6 Effect of memory 93

3.7.7 Effect of intelligence 93

4 Discussion

94

4.1 Group differences in different modalities and temporal acuity tasks 94

4.1.1 Windows of temporal integration 101

4.2 Methodological considerations 102

4.3 Effect of adult age 104

4.4 Correlations between temporal acuity tasks 106

4.5 Relationships between temporal acuity and reading-related tasks 108

4.5.1 Language, modalities, and time 109

4.5.2 Temporal acuity rehabilitation and training 112

4.5.2.1 Specific language impairment 112

4.5.2.2 Developmental dyslexia 113

4.6 Neural disorganization hypothesis 115

Acknowledgements

117

References

117

Original publications

149

(10)

1 Introduction

1.1 Definition of developmental dyslexia

Developmental dyslexia has been defined in 1968, by the World Federation of Neurology, as a disorder manifested by difficulty in learning to read despite conventional instruction, adequate intelligence, and sociocultural opportunity (Critchley, 1970). This definition concentrates only on reading and has been extended by the Orton Dyslexia Society to a specific problem in reading, often accompanied by difficulties of acquiring proficiency in writing and spelling (Orton Dyslexia Society, 1994). Developmen- tal dyslexia is therefore considered to be one of the specific learning difficulties as opposed to the more generalized developmental disorders. For example, general reading backwardness, or garden-variety poor reading, is a term for describing people who have reading difficulties, but also more generalized learning problems. Developmental dyslexia can further be differentiated from acquired dyslexia, which occurs abruptly or gradually, after possibly normal development in reading and writing, in neural in- sults, for instance, cerebral strokes or growing tumors. In this introduction, (developmental) dyslexia is used to refer to the specific developmental learning difficulty , if not otherwise determined.

The difficulties of developmentally dyslexic readers are gen- erally assumed to result from difficulties in language processing. It is not known, however, whether these language-related difficulties alone are the primary cause of dyslexia, since dyslexic readers often have symptoms that extend beyond linguistic

(11)

tasks. Research findings have led to hypotheses that perhaps it is not linguistic impairments but other, more generalized or primary ones that could cause dyslexia. One such candidate is a difficulty in perceptual temporal processing, which could reflect impaired processing within the neural system specialized for processing rapidly changing input information.

In this introduction, the varying symptoms of dyslexia and its relations to other learning difficulties are presented. The developmental pathways to fluent and nonfluent reading are described together with the suggested biological basis for developmental dyslexia. Finally, hypotheses, results, and open questions in the research area of temporal processing are reviewed. In a series of four studies (I-IV), temporal processing (cognitive temporal acuity) was investigated with an aim to clarify some of the problems, that is, the generality (modalities, estimation methods) of the possible temporal processing impairment in developmental dyslexia, its relation to reading-related processes, and how it is manifested at different ages.

1.2 Frequency

1.2.1 Identification method

The assessment methods and criteria for diagnosing dyslexia vary, and with them the specific estimates of its frequency. Recently, in a large (5718, 5-19 year-old children) population-based birth co- hort study, Katusic and her colleagues (Katusic, Colligan, Bar- baresi, Schaid, & Jacobsen, 2001) assessed the incidence of dyslexia (number of new cases occurring in a population during a specific time-interval) depending on the identification method.

Regression-based estimates predict a person’s reading performance based on her/his Intelligence Quotient (IQ), or other com- parable indicator. The diagnosis is based on a pre-set criterion for discrepancy between the predicted and the real reading performance. Discrepancy-based estimates assess the difference between the participant’s IQ and reading. Again, a pre-set difference results in diagnosis. In the case of low-achievement meth-

(12)

od, the diagnosis is given based on poor reading performance alone. When regression methods were used, fewer children were diagnosed (from 5.3%) than in the case of discrepancy-based or low-achievement methods (up to 11.8%) (Katusic et al., 2001).

These numbers are, however, in line with previous estimates of up to 5% of school-aged children having severe reading difficulties, up to 15% having moderate, and as many as 20% having mild difficulties (Hulme, 1987; Poussu-Olli, 1993; Stein & Walsh, 1997). In Finland the frequency of dyslexia corresponds roughly to these international estimates (6%, Lyytinen, Leinonen, Niku- la, Aro, & Leiwo, 1995; Poussu-Olli, 1993).

1.2.2 Gender

It has been argued that developmental dyslexia is either more frequent among males than females, or that the pronounced male prevalence (proportion of the entire population found to have a condition at a certain point in time) results from biased participant selection (Shaywitz, Shaywitz, Fletcher, & Escobar, 1990).

In an earlier epidemiologic sample (Shaywitz et al., 1990), the male to female ratios differed from each other slightly (males, second grade: 8.7% and third grade: 9%; females, second grade:

6.9% and third grade: 6%), although this was not statistically significant with research-identified participants in the second and third grades. In school-identified dyslexics, the prevalence among males was statistically significant (males: 13.6% and 10%; females:

3.2% and 4.2%, in the second and third grades, respectively). In more recent studies, this pronounced male incidence has been replicated (Flannery, Liederman, Daly, & Schultz, 2000) and males have had a greater relative risk of incidence (varying from 1.98 to 2.98) with every identification method, both in research and school-identified samples (Katusic et al., 2001).

1.3 Relationship to other specific learning difficulties

Developmental dyslexia has been shown to be comorbid with various other specific learning difficulties. Comorbidity refers to

(13)

the co-occurrence of conditions that are believed to be separate.

Attention deficit disorder with or without hyperactivity, with its prevalence of up to 10% of school-aged children (Scahill &

Schwab-Stone, 2000), is one of the most common co-morbid conditions for dyslexia or for learning difficulties in general [estimates from 10% (Halperin, Gittelman, Klein, & Rudel, 1984) to over 90% (Silver, 1981)]. This co-occurrence has partly been explained by shared genetic influences (Light, Pennington, Gilger,

& DeFries, 1995; Stevenson, Pennington, Gilger, DeFries, & Gil- lis, 1993), so that only certain features of attention deficit disorder, for instance inattention, would be related to developmental dyslexia (Gilger, Pennington, & DeFries, 1992; Willcutt, Penning- ton, & DeFries, 2000). The co-occurrence of developmental dys- calculia (difficulties in arithmetic concepts or operations) and dyslexia has been shown to be perhaps as common as specific difficulties in arithmetic [prevalence up to 6% of school population (Shalev, Auerbach, Manor, & Gross-Tsur, 2000; Shalev &

Gross-Tsur, 2001)] or reading alone (Badian & Ghublikian, 1983;

Fletcher & Loveland, 1986; Lewis, Hitch, & Walker, 1994; Nor- man & Zigmond, 1980). The comorbidity has been explained in part by genetic influences (Knopik, Alarcon, & DeFries, 1997;

Knopik & DeFries, 1999; Light & DeFries, 1995). Developmental dyspraxia (motor and coordination difficulties) causes difficulties in up to 5% of children (Henderson & Hall, 1982; Sovik & Mae- land, 1986), but over 20% of developmentally dyslexic children may have dyspraxic features, for instance, motor clumsiness (Maeland & Sovik, 1993). Dyslexic children often also have difficulties in complex motor tasks, such as nonverbal choice reaction time (Nicolson & Fawcett, 1994) or bimanual asynchronous hand movements (Wolff, Michel, Ovrut, & Drake, 1990).

Specific language impairment (SLI) refers to a delay or difficul- ties in language acquisition in combination with normal nonverbal intelligence. It is estimated to affect up to 10% of preschool- age children (Rapin, 1998). However, over half of the children with SLI may also have reading difficulties (Tomblin, Zhang, Buckwalter, & Catts, 2000) and developmentally dyslexic readers have been shown to have difficulties in various aspects of

(14)

language, for instance, speech and phoneme perception (Chiappe, Chiappe, & Siegel, 2001; Joanisse, Manis, Keating, & Seidenberg, 2000), phonological processing (Bradley & Bryant, 1978, 1983), and morphology (Joanisse et al., 2000). Early language impairment is also a risk factor for later reading disability (Bishop &

Adams, 1990; Snowling, Bishop, & Stothard, 2000). Accordingly, over 50% of developmentally dyslexic readers could equally well be classified into the group of children with SLI, and vice versa (McArthur, Hogben, Edwards, Heath, & Mengler, 2000).

The view that developmental dyslexia is strictly independent of other specific learning difficulties may, therefore, be questioned, in at least some cases. Developmentally dyslexic readers often have features of the other specific learning difficulties. The specific learning difficulties seem to accumulate, and they can in some cases even be explained by shared genetic influences. Fur- ther, the classification is sometimes unreliable.

1.4 Development of fluent and nonfluent reading

The cause and origin of developmental dyslexia are not fully known. The prevailing suggestions may be divided into three possibilities. Firstly, developmental dyslexia may result directly from difficulties in language (Vellutino, 1978), and phonological processing (Bradley & Bryant, 1978, 1983; Vellutino, 1978; see Snowling, 2000 for a review). The other possible difficulties, for instance, in motor functions and perception, are assumed to be independent of developmental dyslexia and not causal for the true linguistic difficulties (Mody, Studdert-Kennedy, & Brady, 1997; Shaywitz et al., 1998). Secondly, a more basic deficit, often suggested to be impaired perceptual processing, may be sufficient to deteriorate orthographic and phonological representations and processing (e.g. Lovegrove, 1993; Tallal, 1980), and therefore reading acquisition (Farmer & Klein, 1995; Galaburda & Liv- ingstone, 1993; Galaburda, Menard, & Rosen, 1994; Livingstone, Rosen, Drislane, & Galaburda, 1991; Lovegrove, 1993; Stein &

Walsh, 1997; Tallal, 1980; Tallal, Merzenich, Miller, & Jenkins, 1998). Thirdly, the possibly co-existing perceptual and linguistic

(15)

difficulties are assumed to reflect a common neural background.

Therefore, the perceptual problems alone do not cause dyslexia, although are perhaps indicative of the underlying causal neural dysfunction.

1.4.1 Phonological processing

Phonological processing (Torgesen, Wagner, & Rashotte, 1994) indicates the processing of phonological or sound structure of oral and written language. It may be divided into three sub-processes relevant for reading acquisition. The first, phonological aware- ness, refers to the awareness of the sound structure of spoken language, that is, the awareness of syllables, onsets (part preced- ing the vowel) and rimes (vowel and the following consonants) within syllables, and phonemes (the smallest contrasting unit of sound) (Swan & Goswami, 1997; Witton et al., 1998). It can be assessed with, for instance, tasks of phoneme identification, iso- lation, and blending. Phonological awareness is a prerequisite and a predictor of emerging reading ability, beginning to arise only shortly before the children actually learn to read (Bradley &

Bryant, 1978, 1983; Goswami, 1993; Goswami, 1999). Delays and difficulties in the development of phonological awareness are related to developmental dyslexia (Bradley & Bryant, 1978).

The second phonological sub-process (Torgesen, Wagner, &

Rashotte, 1994) is phonological memory or coding, which refers to the coding and temporary storage of sound-based representations. It is called the phonological/articulatory loop in the working memory framework of Baddeley and Hitch (1986). The codes or representations for the short-term phonological storage are the phonological properties of the stimuli (Torgesen, Wagner, &

Rashotte, 1994). Therefore, impaired performance is assumed to reflect imperfect representations of phonology, for instance pho- netically underspecified representations of phonemes. Phonolog- ical memory is assessed with memory span tasks that require verbal short-term retention of relatively meaningless sequences (e.g. digits, pseudowords). Fluent reading development can be predicted by earlier phonological memory performance (Jorm,

(16)

Share, Maclean, & Matthews, 1986), and developmentally dyslexic children often have difficulties in tasks tapping phonological memory (Baddeley, 1986).

The third sub-process of phonological processing (Torgesen, Wagner, & Rashotte, 1994) is the rapid access of phonological infor- mation stored in long-term memory. The efficiency of the access to these phonological codes is related to the extent to which phonological information is used in reading, that is, the conversion between graphemes (the smallest contrasting unit of writing representing one phoneme) and phonemes. Rapid access to phonological information is assessed with rapid naming tasks (Denckla & Rudel, 1976). The speed in naming matrices of sym- bols or objects before school predicts early reading acquisition (Bowers, Steffy, & Tate, 1988; Wolf, 1991), and dyslexic readers often have difficulties in such rapid naming tasks (Kinsbourne, Rufo, Gamzu, Palmer, & Berliner, 1991).

1.4.2 Fluent and nonfluent reading

The process of learning to read (in English) is often described as evolving through stages (Ehri & Wilce, 1985; Frith, 1985; Marsh, Friedman, Welch, & Desberg, 1981). The model of Uta Frith (1985) divides the process into three stages. In the first stage, logograph- ic, children use visual cues, for instance, word length or other salient features, to access the meaning of a limited number of words. The second stage, alphabetic, necessitates phonological awareness. This stage involves acquiring the conversion rules between phonemes and graphemes. Children learn these corre- spondences when beginning to write, and thereafter apply the rules when reading familiar and unfamiliar words as well as pseudowords. Gradually, the speed of grapheme-phoneme conversion in reading accelerates and becomes more automatic. At the third, orthographic stage, the words are again processed as entities based on morphemes (the smallest meaningful units) or even larger wholes, without the need for grapheme-phoneme conversion.

(17)

In the well-practiced adult age, word recognition and reading are suggested to follow two basic alternative routes in the so- called dual-route model of reading (Coltheart, 1978; Coltheart, Curtis, Atkins, & Haller, 1993). These routes correspond approximately to the alphabetic and orthographic levels of reading acquisition presented above (Frith, 1985). Familiar words are suggested to be recognized based on their whole word or morphe- mic form. Unfamiliar words are suggested to be read with the aid of grapheme-phoneme conversion.

The more recent connectionist (or parallel distributed processing) models (McClelland, 1988; Plaut, McClelland, Seidenberg,

& Patterson, 1996; Seidenberg & McClelland, 1989) suggest that the representation of a specific word consists of a distributed parallel activity pattern of multiple simple elements. The process of learning to read is explained by the gradual and corrective learning of associations between input patterns (single letters/

letter clusters, letter sequences making up whole words) and output patterns (phonemes/phonetic features, phonological syllables, morphemes, words). More frequent words and words with frequent letter-sound mappings are learnt more easily due to the excessive amount of rehearsal and the ability to generalize from one event to another. The mapping between orthographic input and semantics hastens the learning of irregular word reading (Plaut et al., 1996).

Developmental dyslexia is suggested to result from difficulties in phonological processing and hence in acquiring the grapheme-phoneme correspondence rules (Bradley & Bryant, 1978, 1983). In the presented models, this would indicate not reaching the alphabetic stage (Frith, 1985), or difficulties in mapping between orthographical input and phonological output patterns (Snowling et al., 2000). In less severe cases of dyslexia, the transi- tion from the alphabetic to the orthographic stage would be dis- turbed. Dyslexic readers are, however, suggested to be able to employ the orthographic-semantic route in reading, thus somewhat compensating for their phonological difficulties (Snowling et al., 2000).

(18)

1.4.3 Effect of orthography

Phonological awareness is suggested to develop similarly, independently of the language environment (Goswami, 1999): first, the child acquires the awareness of syllables, then of sub-syllabic components (e.g. onsets and rimes in English), and last of phonemes. Also, independent of language, difficulties in phonological processing seem to be one of the core deficits of developmental dyslexia (Paulesu et al., 2001). Languages vary greatly in their complexity of grapheme-phoneme correspondence, however. Finnish has a two-way shallow orthography. Every letter corresponds to one pronunciation and every phoneme has only one corresponding letter (or in one case a bigram). In languages with relatively regular grapheme-to-phoneme conversion in reading, it is easier to become aware of the phonemic units and learn to read at an early age, also for dyslexic readers, as compared to for example English with its ambiguous letter-sound rules (Cos- su, 1999; Wimmer, Landerl, & Frith, 1999). Further, in the case of the relatively regular languages, the dyslexic readers are more accurate in grapheme-phoneme conversion and read at a faster rate than the readers whose languages have deep orthographies (Harris & Hatano, 1999; Paulesu et al., 2001). However, even in the case of the relatively regular languages, learning to read is slower for dyslexic compared to fluent readers. The ultimate reading rate is also slower and the performance more error-prone (Wimmer et al., 1999). Interestingly, the dyslexic readers’ phonological difficulties in languages with relatively regular grapheme- to-phoneme conversion can be observed more easily in some other contexts, for example, in verbal spoonerisms (Wimmer et al., 1999) or in the impaired awareness of phoneme quantity in Finn- ish.

1.4.4 Adult developmental dyslexia

Most readers with a childhood history of developmental dyslexia have persistent reading difficulties also in adult age (Scarbor- ough, 1984). This contradicts the suggestion that developmental dyslexia reflects a mere lag in development. Some adult dyslexic

(19)

readers, with severely impaired phonological skills, may however reach high levels of literacy by using various compensatory strategies (Daryn, 2000). Thus, impaired phonological processing seems to be the core deficit also in adult developmental dyslexia (Felton, Naylor, & Wood, 1990). Dyslexic adults have difficulties in various aspects of phonological processing, for example in tasks of phonological awareness, verbal short-term memory and learning, verbal fluency, rapid naming, and nonword reading (Felton et al., 1990; Kinsbourne et al., 1991; Pennington, Van Orden, Smith, Green, & Haith, 1990; Scarborough, 1984).

1.4.5 Subtypes of developmental dyslexia

Developmental dyslexia has sometimes been seen as a heterogeneous group of conditions, which could be divided into subtypes, most often three (Boder, 1973; Castles & Coltheart, 1993). The first group includes those who have difficulties in phonological processing and grapheme-phoneme conversion. This corresponds roughly to difficulties in advancing to the alphabetic stage of Uta Firth (1985) or acquired phonological dyslexia, in which the grapheme-phoneme route, and hence nonword reading, is impaired. The second group comprises those who have difficulties in sight vocabulary. This corresponds to difficulties in proceed- ing to the orthographic stage of reading (Frith, 1985) or having features of acquired surface dyslexia, in which whole word processing, and hence irregular word reading, is impaired (Cas- tles & Coltheart, 1993). The third group is a combination of the first two, having difficulties in both processes.

Boder (1973) divided dyslexic readers into groups based on their writing errors. The majority belonged to the dysphonetic group who had difficulties in grapheme-phoneme conversion and were prone to process the words as wholes, relying on their sight vocabulary. A smaller group of dyseidetic readers had emphasized difficulties in visuospatial functions and therefore difficulties in acquiring sight vocabulary. Mixed, dysphoneidetic readers had difficulties in both areas, and therefore the least positive prognosis. This early classification is presented here because it

(20)

has been used in later research where it has been validated psy- chophysiologically both with visual and auditory material (Borst- ing et al., 1996; Cohen, Hynd, & Hugdahl, 1992; Flynn & Deer- ing, 1989; Fried, Tanguay, Boder, Doubleday, & Greensite, 1981).

Castles and Coltheart (1993) used a regression equation, based on age-matched fluent reader performance, and were able to divide up to 85% of their dyslexic readers into two groups: those who were proportionally poorer in either reading pseudowords or irregular words, phonological and surface dyslexics respectively. However, when the same method was applied to dyslexic and reading-age-matched fluent readers (Manis et al., 1996), it resulted in fewer participants falling into specific subtypes. Based on the classification, surface dyslexics were claimed to suffer only from a developmental lag, whereas phonological dyslexics were thought to have a specific learning difficulty.

There are additional difficulties in attempting to divide developmental dyslexia into subtypes and at the moment there is no classification which would be the agreed one. The criteria for participant selection may bias the ratios between different sub- groups (Aaron, Joshi, & Williams, 1999), as do the statistical methods selected (Cestnick, 1998). Subtyping may also result in over- simplified classes, which are not sufficient to explain the whole spectrum of difficulties in developmental dyslexia or classify every impaired reader (Manis, Seidenberg, Doi, McBride-Chang,

& Petersen, 1996).

1.5 Biological basis of developmental dyslexia 1.5.1 Genetic background

A number of studies suggest that there are genetic elements that contribute to developmental dyslexia. Twin studies have shown that monozygotic twins have a higher concordance rate for dyslexia, where both twins share the same trait, than dizygotic twins (DeFries & Alarcón, 1996). Further, the heritability of reading impairment, the proportion of variation attributable to genetic

(21)

factors, has been estimated to be approximately 50% (DeFries &

Gillis, 1993).

With various linkage analysis approaches the probability of a certain genotype and phenotype occurring together by chance is assessed. The phenotypes in these analyses are considered to be either in the more traditional approach qualitative, where the trait is dichotomized into those affected and those unaffected, or quantitative in the more recent Quantitative Trait Loci (QTL) approaches, where the trait is a measurable continuum. No single gene explaining the dyslexic symptoms across families has been implicated, but positions for multiple loci have been mapped re- peatedly. Developmental dyslexia is most probably genetically complex and heterogeneous, as the dyslexic features themselves are considerably varying and diverse. In specific families, alleles at a single locus might be crucial, however.

The long arm of chromosome 15 (15q21, DYX1) has been linked to (and associated with, p < 0.006) dyslexia (Fulker et al., 1991;

Morris et al., 2000; Nopola-Hemmi et al., 2000; Smith, Kimber- ling, & Pennington, 1991). Linkages have been found more specifically with single-word reading (Grigorenko et al., 1997) and spelling (in a shallow orthography) (Schülte-Körne et al., 1998), but not with phonological awareness (Grigorenko et al., 1997).

Conflicting evidence exists, however, for nonsignificant linkage in the older studies (Bisgaard, Eiberg, Moller, Niebuhr, & Mohr, 1987; Rabin et al., 1993).

The short arm of chromosome 6 (6p21.3, DYX2) has been re- peatedly linked to dyslexia (e.g. Smith et al., 1991). The linkage has been found to be stronger with more severe reading difficulties, as assessed with (Peabody Individual Assessment Test) reading recognition, comprehension, and spelling (Cardon et al., 1994).

Further, linkages with specific features/phenotypes of reading disability have been shown: phonological awareness (Gayan et al., 1999; Grigorenko et al., 1997); accuracy and speed of irregular, (pseudo)homophone, and nonword reading (Fisher et al., 1999; Gayan et al., 1999; Grigorenko, Wood, Meyer, & Pauls, 2000), vocabulary and spelling (Grigorenko et al., 2000), and naming speed (Grigorenko et al., 2001). However, conflicting evidence

(22)

exists indicating the importance of different analysis methods and criteria for dyslexia (Field & Kaplan, 1998; Petryshen, Kap- lan, Liu, & Field, 2000; Schülte-Körne et al., 1998). The long arm of chromosome 6 (6q11.2-12, DYX4) may also be implicated in developmental dyslexia (Petryshen et al., 2001).

The short arm of chromosome 2 (2p15-16, DYX3) has been linked to dyslexia (Francks et al., 2001). In a single multigenera- tional Norwegian family, a linkage was found as assessed with an emphasis on phonological processing difficulties (Fagerheim et al., 1999).

In addition, preliminary results suggest loci on other chromosomes. Chromosome 1 has been linked to dyslexia (Rabin et al., 1993), more specifically to difficulties in phonological awareness, nonword reading, and naming speed (Grigorenko et al., 2001). In a single German family, a balanced translocation of chromosomes 1 and 2 (1p22;2q31) has been linked to retarded speech and spelling (Froster, Schülte-Körne, Hebebrand, & Remschmidt, 1993). Chromosome 3 has recently been linked to dyslexia in a four-generation family by a Finnish group (Nopola-Hemmi et al., 2001). The linkage was found to exist for multiple dyslexic features: phonological awareness, rapid naming, and verbal short- term memory. Also, single preliminary findings have been reported for chromosomes 4, 9, 13, and 18 (Fisher & Smiths, 2001).

Therefore, multiple different loci seem to be linked to various aspects of developmental dyslexia, evidence for chromosomes 15, 6, and 2 being the strongest. However, the suggestion that a specific feature of dyslexia would be linked to a certain chromosome, for instance, phonological processing to chromosome 6 and orthographic skills to chromosome 15, is far from unequivocal (Warnke, 1999).

1.5.2 Anatomical findings and the magnocellular system

Developmentally dyslexic readers differ from fluent readers in various aspects of cerebral anatomy. Some of the findings have been taken as evidence of impaired magnocellular system and functioning (Stein & Walsh, 1997). The distinction between two

(23)

neural subsystems, magnocellular and parvocellular, originates from vision research on cats and primates, but is considered to be an applicable framework for humans as well. M/magno/transient system has high sensitivity for temporal information, lumi- nance and contrast, but poor spatial resolution and chromatic selectivity. P/parvo/sustained system, on the other hand, has high sensitivity for spatial information and color, but poorer temporal resolution and contrast sensitivity (Merigan & Maunsell, 1993).

This division into systems specialized in processing rapidly changing and more static input information has been generalized from vision to other modalities and beyond them, for instance, to motor functions (Lovegrove, 1993; Merigan & Maun- sell, 1993; Stein & Walsh, 1997).

1.5.2.1 Anatomical findings related to the magnocellular system

Galaburda and his colleagues have shown thalamic abnormalities implicating the magnocellular system in developmental dyslexia post mortem, especially in the geniculate nuclei. The thalamic medial geniculate nuclei (MGN) and lateral geniculate nuclei (LGN) are part of perceptual pathways leading to primary sensory cortices. In the MGN (Galaburda et al., 1994), mainly related to auditory processing, neurons of dyslexic readers were smaller on the left side, compared to the right. With controls, no such asymmetry was observed. Also, the left MGN of the dyslexic readers contained more small and less large cells compared to the controls. In the LGN (Livingstone et al., 1991), focused on visual processing, the cells in the magnocellular (large cell) layer, but not in the parvocellular (small cell) layer, were smaller in the case of dyslexic readers, compared to fluent readers.

Cortical abnormalities have also been observed. At the primary visual areas, paralleling the thalamic findings, the cell size of dyslexic and fluent readers has been found to differ (Jenner, Rosen, & Galaburda, 1999). The fluent readers’ neurons were larger in the left hemisphere, whereas with the dyslexic readers, no such asymmetry was observed. However, cells in the layers with

(24)

thalamic magnocellular input were not consistently different in the case of the dyslexic readers, compared to the fluent readers.

Compared to the controls, the dyslexic readers also exhibit more cortical disorganization, that is, ectopias (intrusions of cells from another layer), dysplasias (within-layer cell disorganization), and glial scars in border-zone areas, the latest indicating a vascular, possibly immunological etiology (Galaburda & Kemper, 1979;

Galaburda, Sherman, Rosen, Aboitiz, & Geschwind, 1985; Hum- phreys, Kaufmann, & Galaburda, 1990). In males the ectopias were located predominantly in the left perisylvian region, temporal lobe, and inferior frontal cortex (Galaburda & Kemper, 1979;

Galaburda et al., 1985). In females the ectopias were fewer in number and not lateralized (Humphreys et al., 1990). The amount of cortical scarring was, however, larger in females (Galaburda

& Kemper, 1979; Galaburda et al., 1985; Humphreys et al., 1990).

These results have been taken as evidence of abnormal cortical development, during or after neuronal migration, that is, the 6th gestational month or after that (Galaburda et al., 1985; Humphreys et al., 1990). The gender differences are explained by the earlier maturation of the female central nervous system. The time of the possible insult could be the same, but the consequences interact with gender (Humphreys et al., 1990). The subjects in the studies of Galaburda and colleagues cannot, however, be described as pure dyslexics, as they suffered from various other medical conditions, for instance, head traumas, inflammations, and subarach- noid hemorrhages (Galaburda et al., 1985; Humphreys et al., 1990). In addition, the control groups were less than perfectly matched, for instance, in age (Galaburda et al., 1994).

Cerebellum, ‘the head-ganglion of the magnocellular system’

(Stein, 2001), has recently been linked to developmental dyslexia (reduction of gray matter in semilunar lobules: Brown et al., 2001;

altered patterns of cell density, smaller number of larger neurons: Rae et al., 1998). The biochemical differences in the left tem- poro-parietal lobe and the right cerebellum of in vivo adult dyslexic men, as compared to fluent readers in the study of Rae and her colleagues (Rae et al., 1998), have been interpreted to indicate developmental abnormalities in the cell density of areas re-

(25)

lated to the magnocellular system. It has been suggested, however, that perhaps only a certain dyslexic subtype would be related to cerebellar impairments (phonological dyslexia: Leonard et al., 2001).

1.5.2.2 Anatomical findings related to language

Other cerebral areas, not explicitly related to the magnocellular framework, but more so to language and reading and therefore presumably to the linguistic difficulties in dyslexia, have also been investigated, the planum temporale perhaps the most intensive- ly. This area has been linked to features like handedness (Beaton, 1997; Shapleske et al., 1999), gender (Good et al., 2001; Shapleske et al., 1999), language lateralization (Beaton, 1997), as well as phonological processing and reading (Dalby, Elbro, & Stodkilde- Jorgensen, 1998; Heiervang et al., 2000; Larsen, Høien, Lundberg,

& Ødegaard, 1990). Planum temporale, part of the superior bank of the superior temporal gyrus, situated posteriorly to the primary auditory cortex, has been shown to be leftwards asymmetrical in about 80% of fluent readers (Shapleske, Rossell, Wood- ruff, & David, 1999). The asymmetry is observed already approximately during the 30th gestational week (Chi, Dooling, & Gilles, 1977; Wada, Clarke, & Hamm, 1975; Witelson & Pallie, 1973), com- parable to or after the time of neuronal migration implicated above. The plana in child and adult dyslexic readers have devi- ated from this pattern and have often been shown to be symmet- rical (postmortem: Galaburda & Kemper, 1979; Galaburda et al., 1985; Humphreys et al., 1990; in vivo: e.g. Larsen et al., 1990;

Rumsey et al., 1986) or reversely asymmetrical (Hynd, Semrud- Clikeman, Lorys, Novey, & Eliopulos, 1990). Conflicting evidence also exists (no group differences: e.g. Robichon, Levrier, Farnari- er, & Habib, 2000; Rumsey et al., 1997; Schultz et al., 1994; other differences perhaps more emphasized: e.g. Heiervang et al., 2000).

Other deviations from the expected cerebral asymmetry among dyslexic readers have been found, for instance, in posterior parts of the inferior frontal gyrus or Broca’s area, an area often associated with language and speech output (Hynd et al., 1990; Robi-

(26)

chon et al., 2000), and the angular gyrus region, an area associated with reading, writing, and grapheme-phoneme conversion (Duara et al., 1991).

The corpus callosum, the bundle of fibers connecting the cerebral hemispheres, has also been implicated in developmental dyslexia. The often found larger posterior callosal area in dyslexic readers (Duara et al., 1991; Robichon & Habib, 1998; Rum- sey et al., 1996) is suggested to reflect differences in the pathways connecting the angular gyri bilaterally (Duara et al., 1991).

Accordingly, callosal morphology has correlated with phonological processing (Robichon & Habib, 1998). Conflicting evidence for the size difference exists, however (e.g. Larsen, Høien, &

Ødegaard, 1992).

1.5.3 Magnocellular hypotheses of developmental dyslexia

Various hypotheses have been presented concerning the possible causal role of the magnocellular system in the difficulties of developmentally dyslexic readers. What these hypotheses have in common is the notion of impaired temporal processing, which either causes the dyslexic reading difficulties (either via affect- ing the visual or linguistic processes) or only co-occurs with them.

1.5.3.1 Lovegrove & Breitmeyer: Saccadic suppression

Based on visual psychophysical contrast sensitivity tasks, Love- grove and his colleagues (Lovegrove, Bowling, Badcock, & Black- wood, 1980) suggested a deficit in the development of the transient (that is magnocellular) visual system in dyslexia. Reading was thought to be impaired, because during a saccade the slowed transient system would not suppress the sustained (parvocellular) system, that is, erase the image of a previous ‘sustained’ fixation (Breitmeyer, 1980; Lovegrove, Martin, & Slaghuis, 1986;

Lovegrove, Garzia, & Nicholson, 1990). As a result, the succeed- ing images would blend. It was later shown, however, that during a saccade, the previous activity of the transient, not the sus-

(27)

tained, system is inhibited (Burr, Morrone, & Ross, 1994). Love- grove (1993) has later hypothesized a more general sensory timing problem behind the difficulties in developmental dyslexia.

1.5.3.2 Stein: The magnocellular theory of developmental dyslexia

Stein and his colleagues (Stein, 2001; Stein & Walsh, 1997) have broadened the original hypotheses by speculating more explicitly that the processing of ‘fast temporal information’ is impaired in developmental dyslexia, and is causally related to the dyslexic difficulties. They suggested that similar divisions, as found in the visual modality, also exists in the auditory and tactile modalities between the neural systems specialized in fast and slow stimulus processing (Stein & Walsh, 1997). This parallels the earlier results and suggestions of Tallal and her colleagues (Tallal, 1980), who showed that auditory temporal processing is impaired in developmental dyslexia. Stein generalizes the temporal impairment further to encompass processing beyond perceptual and brain areas outside those solely focusing on perceptual processing, however related to the magnocellular system (e.g. cerebellum) (Stein, 2001). In addition to temporal processing difficulties, Stein has also suggested that, resulting from the magnocellular impairment, some dyslexic readers have difficulties in sta- bilizing binocular fixation and therefore letters appear to move around (Stein & Walsh, 1997). This is supported by their finding that reading difficulties are at least partly relieved when one eye of such a dyslexic reader is occluded (Stein, Richardson, & Fowl- er, 2000).

1.5.3.3 Vidyasagar: Attentional spotlight

Vidyasagar has proposed a causal role of impaired magnocellular mechanisms and attention in developmental dyslexia (Vid- yasagar, 1999). The magnocellular system has emphasized trans- mission times and spatial coding properties of object locations.

It is therefore suggested to give rise to the so-called attentional

(28)

spotlight. This pre-attentive, parallel processing, would guide the parvocellular system to focus selectively on a specific area or areas for serial search. Magnocellular impairment could therefore also lead to symptoms of impaired parvocellular functioning. In reading, the communication between these two systems would enable the ordering (magno-based) of identified letters or words (parvo-based). When learning to read, the accumulative training of the sequential movement of the attentional spotlight is suggested to be perhaps the most important element. Other modalities besides vision could also be affected. This could result from a modality-specific or a more generalized impairment in the various magnocellular systems. Alternatively, impaired visual magnocellular input could affect parietal, possibly multimodal, spatial representation system and therefore cause difficulties also in other modalities.

1.5.3.4 Hari: Sluggish Attentional Shifting

Hari and her colleagues (Hari & Renvall, 2001) have presented a magnocellular hypothesis concentrating on attention as well. She proposes that dyslexic readers have difficulties in engaging and disengaging attention, that is, they would suffer from ‘sluggish attentional shifting’. This would result from a weakness of attentional capture system, which is supported by, for instance, parietal functions. Because of this ‘sluggish attentional shifting’, processing of rapid stimulus sequences would be impaired, and the chunks in which input information is processed would be prolonged, possibly in every perceptual modality. This sluggish- ness is suggested to distort motor sequencing too, and perhaps cortical representation of stimulus categories, for instance, phonemes.

1.5.3.5 Nicolson & Fawcett: Cerebellar dysfuncion

Nicolson, Fawcett and Dean (2001) claim that a cerebellar dysfunction could cause the impairments observed in 80% of dys-

(29)

lexic readers. Their hypothesis in not ‘magnocellular’, but is in- troduced here because of its association to the hypotheses of Stein (2001). Directly, cerebellar dysfunction could cause motor difficulties and therefore the poor quality of handwriting and ham- pered articulation observed in dyslexic readers. Indirectly, the cerebellar dysfunction could cause difficulties in sensory feed- back processing, because articulation would require more processing capacity. Slower articulation speed would also result in less effective working memory processing and therefore difficulties in language acquisition. The defective articulatory representations could again lead to poor sensitivity to the sound structure of spoken language, that is, poor phonological awareness.

Difficulties in spelling could be explained with over-effort in reading, poor phonological awareness, and impaired automatization of skills.

1.5.3.6 Galaburda: Cortical-phonological and thalamic- temporal impairments

Galaburda and his colleagues (Galaburda et al., 2001) have suggested that difficulties in linguistic and temporal processing could perhaps be somewhat independent of each other. In their rodent model, induced cortical injuries led, in male rats only, to secondary thalamic MGN changes (greater number of small and fewer large cells). Temporal processing impairment was observed only in the case of animals with these secondary thalamic abnormalities. Impairments in tasks of more ‘cognitive’ nature (e.g. learning) were related to primary cortical injuries. Accordingly, they suggest that also in humans cortical lesions would lead to difficulties in higher cognitive functioning, for instance, impaired phonological processing, but not directly to temporal processing difficulties. In some cases, in addition to the cortical areas, the thalamic magnocellular areas would be implicated and this sub- cortical abnormality would be related to impaired temporal processing.

(30)

1.6 Temporal processing and acuity 1.6.1 Levels and modalities of processing

The speed or capacity of processing rapidly changing perceptual information is suggested to form the basis for many if not most higher cognitive functions (Cerella & Hale, 1994; Kail & Salthouse, 1994). Temporal processing is not, however, a unitary function (Farmer & Klein, 1995). Different aspects of temporal processing are suggested to form a hierarchy, beginning from the simpler stimulus (detection and) individuation, and leading to the more complex temporal estimations of cognitively separated perceptual entities. The simpler estimations in this hierarchy are suggested to be prerequisites for processing at other levels, and therefore the different aspect of temporal processing require different cognitive processes (Farmer & Klein, 1995; Hirsh & Sherrick, 1961;

Jaskowski, 1991). In this introductory review, temporal processing is referred to as a general concept, but a distinction is made between two areas of research. The first focuses on the temporal modifications within a stimulus or stimulus sequence, often at very high frequencies, and its effects on the quality of perception. The other focuses on the question of how the temporal relations between stimuli that have already been processed and perceived as discrete perceptual events are estimated. The latter aspect, cognitive temporal acuity, is the focus of the present series of studies (I-IV).

In stimulus individuation, the observer assesses whether one or more stimuli have been presented (Farmer & Klein, 1995). This class comprises fusion tasks in which the minimum interval required between two identical stimuli in order for them to be perceived as two is assessed. In gap detection tasks, the minimum length of a gap causing a perception of discontinuation in a long stimulus is determined. In integration tasks, the minimum length of a separating interval at which two nonidentical stimuli are perceived as separate in contrast to integrated is determined.

The next level of temporal processing requires that the stimuli that are compared are perceived and identified as discrete per-

(31)

ceptual entities (individuated). Only after this can their explicit temporal comparisons be made. This means that the stimuli have to differ along some continuum, which gives them an amodal (duration, location) or modality-specific identity (e.g. pitch in audition, color in vision) (Farmer & Klein, 1995). At this level the assessments of stimulus simultaneity and nonsimultaneity may be differentiated, since the psychometric functions in their assessments are identical in form, but shifted in time (Allan, 1975). The simultaneity/nonsimultaneity assessment is further considered to be a prerequisite for temporal order judgment (TOJ) (Hirsh &

Sherrick, 1961; Jaskowski, 1991; Stelmach & Herdman, 1991): order judgment may follow only the assessment of stimulus nonsimultaneity.

1.6.2 Cognitive models of simultaneity/nonsimultaneity and order judgment

Several models and their modifications have been proposed for the assessment of simultaneity/nonsimultaneity and the temporal order of discrete perceptual events (cognitive temporal acuity). In the majority of them, stimulus processing proceeds through independent or interacting channels to a central timing mechanism for a decision analysis (e.g. Pastore, 1983; Sternberg & Knoll, 1973).

Stimulus progress. The presentation modality may affect the stimulus progress. The reaction times for auditory stimuli are often shorter than those for visual stimuli, and therefore the auditory stimuli have to follow the visual stimuli in order for the two modalities to be perceived as simultaneous (e.g. Jaskowski, Jaroszyk, & Hojan Jezierska, 1990; see however Rutschmann &

Link, 1964). However, in addition to stimulus modality, factors like stimulus duration and intensity may also affect the progress (Jaskowski, 1991). Further, the acuity in assessing nonsimultaneity within a modality has been suggested to vary between modalities, the visual modality having the poorest temporal acuity, the tactile one the second poorest, and the auditory one having

(32)

the best temporal acuity (Hirsh & Sherrick, 1961). This kind of variability in the sensory arrival times or temporal acuity affects the thresholds (Rutschmann, 1973).

Decision mechanism. In the presented models, a hypothetical central timing mechanism is postulated. Its exact structure and functions are not, however, agreed on. The arrival of the first stimulus to the central decision mechanism may trigger a moment of duration, within which additional arriving input stimuli are considered simultaneous (triggered-moment hypotheses; see for a review Ulrich, 1987). The central timing mechanism may alternatively sample the incoming information into discrete input- independent perceptual moments (Stroud, 1955). Again input stimuli arriving within the same perceptual moment are considered simultaneous and input stimuli arriving at different perceptual moments nonsimultaneous. In attention-switching models (Kristofferson, 1970), the capacity of the central timing mechanism is limited by its ability to attend to only one input channel at a time, and by the limited and input-independent time-points at which attention can be switched between the input channels.

If the second stimulus arrives before its channel can be attended to, it is perceived as simultaneous with the first one.

The TOJ is expected to require more time and cognitive re- sources, for instance, memory, than simultaneity/nonsimultaneity assessments (Jaskowski, 1991). The threshold for order judgment has been originally suggested to be about 20 ms independent of modality or modality combination (Hirsh & Sherrick, 1961).

This would suggest a time organizing system that is both independent of the perceptual systems and central to all of them. In later studies, the thresholds in different modalities have been shown to vary from this constant value (Swisher & Hirsh, 1972).

In addition to the factors presented above, for instance the allo- cation of attention, the assessed aspect of the presentation (stimulus onset/offset), the number of presentation possibilities (two possibilities: which one first/three possibilities: simultaneous or which one first) also affect the thresholds (Pastore, 1983; Stel- mach & Herdman, 1991; Ulrich, 1987).

(33)

1.6.3 Localization and time-windows of temporal processing

The presented models of cognitive psychology have not been validated with psychophysiological methods. Of the levels of temporal processing, stimulus individuation has been studied the most.

Temporal processing is observed and required throughout the (central) nervous system. In the case of auditory modality, for example, the auditory nerve fibers lock and fire to the temporal phase of low frequency sound stimuli of up to as high as 5000 hertz (Hz) (Hanekom & Kruger, 2001). At the level of the brainstem, in the superior olivary, binaural temporal integration and input comparison occur at the level of tens or hundreds of mi- croseconds (Grothe, 2000). This makes, for instance, horizontal sound localization possible. At the level of the auditory cortex, temporal processing is also extremely refined. The mismatch between a standard and a deviant tone can be detected for temporal manipulations of less than 10 ms (Desjardins, Trainor, Hevenor, & Polak, 1999), a response which is localized at the primary auditory cortex (Tervaniemi, 2001). Gap detection for as short as 1-ms interruptions has also been localized to the primary auditory cortex (Rupp et al., 2000).

In addition to temporal segregation, stimulus integration is also essential, for instance, for object perception/recognition, localization, and partitioning the input information in order to decrease the processing load. In the case of auditory modality, for example, integration is observed already below the level of cerebral cortex, in the brainstem (Grothe, 2000). At the level of the primary auditory cortex, sounds following in rapid succes- sion are suggested to be represented neurally as a single unit (Tervaniemi, Saarinen, Paavilainen, Danilova, & Näätänen, 1994).

This window of temporal integration has been shown to span up to approximately 150 ms from the first stimulus onset (Tervanie- mi et al., 1994; Yabe, Tervaniemi, Reinikainen, & Näätänen, 1997;

Yabe et al., 1998). Crossmodal temporal integration also emerges below the level of the cerebral cortex. A midbrain structure, the

(34)

superior colliculus, includes multisensory neurons for visual, auditory, and somatosensory input (Stein, 1998) and is function- ally related to, for instance, orientation (Wallace, Meredith, &

Stein, 1998). The thalamus has also been associated with crossmodal temporal processing (Galaburda et al., 2001) and temporal integration (Stein & Meredith, 1993). It has been suggested that at the cortical level, crossmodal (temporal) integration emerges already on the secondary sensory cortices (Kaas & Hackett, 1998; Lutkenhoner, Lammertmann, Simoes, & Hari, 2002) perhaps via their interconnections, or that it possibly involves association areas, such as the prefrontal, frontal, parietal, and superi- or temporal cortices (Lewkowicz, 2000). The temporal window of integration in crossmodal (audiovisual) comparisons has been suggested to be up to approximately 150 ms, or perhaps more, depending on the presentation order of the modalities (Lewko- wicz, 1996).

1.6.4 Development and decline

Temporal processing is not a stable feature, but in the case of most modalities and their combinations it seems to follow an in- verted U-shaped function of early development and later decline.

Temporal processing improves in the developing auditory system. This is suggested to point to both peripheral and central mechanisms (Grose, Hall, & Gibbs, 1993; Werner, 1996). In stimulus individuation, for instance, gap detection improves rapidly during the first months of life (Werner, Marean, Halpin, Spetner,

& Gillenwater, 1992), that is, infants require shorter gaps for per- ceiving the discontinuation as they age. The adult stage is ap- proached gradually (Wightman, Allen, Dolan, Kistler, & Jamie- son, 1989) and reached in the early adolescent years (Irwin, Ball, Kay, Stillman, & Rosser, 1985). Children’s results in simultaneity/

nonsimultaneity assessments qualitatively parallel that of adults’, but they have poorer temporal values than adults (Jusczyk, Piso- ni, Walley, & Murray, 1980). Similarly, judging the temporal order of two events ameliorates during development (Brannan & Wil- liams, 1988). In the aging auditory system, temporal processing

(35)

is affected independently of peripheral hearing loss (Fitzgibbons

& Gordon-Salant, 1996). In stimulus individuation, for instance, gap detection deteriorates with various types of stimuli (wide band noise: He, Horwitz, Dubno, & Mills, 1999; noise bursts: Snell, 1997;

simple tones: Moore, Peters, & Glasberg, 1992; Schneider & Ham- stra, 1999; pips: Schneider, Pichora-Fuller, Kowalchuk, & Lamb, 1994). In fusion tasks, older adults perceive two tones as one at longer intervals in between, compared to younger adults (Robin

& Royer, 1989). Order judgment processing also declines with age (Fitzgibbons & Gordon-Salant, 1998; Gordon-Salant & Fitzgib- bons, 1999; Trainor & Trehub, 1989).

Visual temporal processing has also been shown to vary ac- cording to age. Interestingly, temporal processing in contrast sensitivity tasks has been shown to improve during development, adult values being reached earlier for high temporal frequencies (20 and 30 Hz), and a few years later for slightly lower temporal frequencies (5 and 10 Hz) (Ellemberg, Lewis, Hong Liu, & Mau- rer, 1999). Accordingly, it has been suggested that perhaps the magnocellular system would show ‘precocious’ development compared to the parvocellular system (Dobkins, Anderson, & Lia, 1999). With increasing adult age, stimulus individuation in fu- sion tasks becomes poorer (Walsh & Thompson, 1978). Older adults’ Order judgment is also poorer than that of younger adults (Di Lollo, Arnett, & Kruk, 1982; Schieber & Kline, 1982). Similar- ly, it has been suggested that during development, the magnocellular neural system is affected by aging, and perhaps even more than the parvocellular (Justino, Kergoat, & Kergoat, 2001; Spear, 1993; Sturr, Van Orden, & Taub, 1987).

Aging observers have also shown reduced tactile temporal processing. In stimulus individuation tasks, fusion thresholds (Petrosino & Fucci, 1989) and gap detection with vibratory stimuli are affected (Van Doren, Gescheider, & Verrillo, 1990). Simultane- ity/nonsimultaneity assessments also become poorer with increas- ing age (Axelrod, Thompson, & Cohen, 1968; Brown & Sains- bury, 2000).

The development and decline of crossmodal temporal processing has been investigated very little. In audiovisual assessments,

(36)

infants at the age of three months are suggested to be able to detect the asynchrony between auditory and visual temporal input (Bahrick, 1988). However, children have been shown to have a wider window of temporal integration for audiovisual input compared to adults (Lewkowicz, 1996).

1.7 Temporal processing in developmental dyslexia

Temporal processing assessed with nonverbal stimuli, and its association with language and reading, has been investigated widely with developmentally dyslexic readers. The aim has often been to resolve the debate whether the dyslexic difficulties are purely linguistic, a result of temporal processing impairment, or if they only co-exist with them. In the following section the major findings in this research area are presented. In some in- stances, studies of children with SLI are described for comparison or in order to compensate the lack of results on developmental dyslexia.

1.7.1 Auditory temporal processing

Both developmentally dyslexic children and adults have been found to have difficulties in various aspects of auditory temporal processing, from stimulus individuation, processing of temporally modulated tones and tone sequences, to the estimations of order. However, dyslexic readers also seem to have difficulties in nontemporal aspects of auditory processing, for instance, in pitch perception.

1.7.1.1 Interaural temporal processing

In a dichotic pitch identification task, dyslexic children have been poorer compared to fluently reading controls. This task is suggested to require the use of interaural time difference information (Dougherty, Cynader, Bjornson, Edgell, & Giaschi, 1998). In binaural masking level difference task the difference in detection, when the tone in noise is the same in both ears or phase-shifted

(37)

between the ears, is assessed. No differences have been found between dyslexic readers and fluent adult readers, whose age and IQ matched those of the dyslexic readers, when a 200 Hz tone in 1000 Hz lowpass noise (Hill et al., 1999), was presented.

However, with a higher tone of 1000 Hz in 500-1500 Hz noise the reading groups, whose age, IQ, education, and hearing sensitivity matched, were found to differ from each other (McAnally &

Stein, 1996).

1.7.1.2 Stimulus individuation

Developmentally dyslexic children in age groups of 7, 8, and 9 years have had poorer pure tone fusion thresholds compared to their age-matched controls (McCroskey & Kidder, 1980). Dyslex- ic adolescents were similarly found to differ from controls whose age/reading-level and IQ matched those of the dyslexics in a click fusion task (Farmer & Klein, 1993). In gap detection, adolescents with developmental dyslexia required a longer gap in 1-2000 Hz bursts than their age-matched and IQ-matched controls (Schülte- Körne, Deimel, Bartling, & Remschmidt, 1999) However, this difference was not significant. In line with this result, dyslexic adults have not been found to differ from fluent readers in detecting gaps in white noise stimuli, or in wide-band noise with frequency gaps, although again, the fluent readers seemed to perform slightly better (Ahissar, Protopapas, Reid, & Merzenich, 2000;

controls matched on age, IQ, education, and hearing sensitivity:

McAnally & Stein, 1996). The differences between children and adults or types of tasks are difficult to interpret because of differences in stimulus complexity.

1.7.1.3 Temporally modulated tones

In frequency modulation (FM) tasks, the carrier frequency (pitch) of a sound is varied at predetermined temporal rates in order to determine a detection threshold for the pitch change. In the case of 10-year-old fluent readers, a 2 Hz FM threshold correlated with phonological processing and reading and predicted them, but a

(38)

higher 240 Hz modulation threshold did not (carrier frequency 1000 Hz) (Talcott et al., 1999; Talcott, Witton et al., 2000). In the case of SLI children, a trend for poorer frequency modulation threshold (carrier frequency 1000 Hz, modulated at 2 and 20 Hz) has been found compared control children who matched them in age, IQ, and gender. The threshold did not, however, correlate with pseudoword repetition/reading (Bishop, Carlyon, Deeks, &

Bishop, 1999). Also, in an auditory evoked response study with frequency modulated tones, 6-11-year-old SLI children with re- ceptive impairment had deviating responses (smaller ampli- tudes), when compared to adults, controls, and children with expressive SLI (Stefanatos, 1993). Adult dyslexic readers have also been found to be less sensitive, that is, require larger modulation, compared to fluent readers who matched them in age and performance intelligence quotient (PIQ), when low modulation rates of 2 and 40 Hz (carrier frequency 500 Hz) were presented, but not with the higher rate of 240 Hz (carrier frequency 1000 Hz) (Witton et al., 1998). Sometimes the reading group differences in FM detection have not been as strong, and they have been statistically significant only when the outliers have been discard- ed (carrier frequency 1000 Hz, modulation 2.5 Hz FM) (Hill, Bai- ley, Griffiths, & Snowling, 1999).

In amplitude modulation (AM) tasks, the strength (amplitude) of a sound is varied at predetermined temporal rates in order to determine a detection threshold for the strength change. Devel- opmentally dyslexic adults have been reported to differ from controls, who matched them in age and hearing sensitivity, in responses following amplitude modulation (carrier frequency 400 Hz, modulated at 20, 40, 60, and 80 Hz). The dyslexic readers’

evoked response (amplitude modulation following response, AMFR) amplitude was lower than that of the controls, but no differences were observed in the latency (McAnally, Hansen, Cornelissen, & Stein, 1997).

1.7.1.4 Temporal illusions

Developmentally dyslexic adults have performed more poorly