The investigative program

The investigation consisted of three studies, reported, respectively, in Chapters 2,

3 and 4.

1.3.2.1 Study 1

In the first study, reported in Chapter 2, three experiments (Experiments 1a, 1b

and 1c) were conducted. In the first experiment, the performance of adult

dyslexic participants on a simple version of the TOJ task was compared with that

of non-dyslexic participants; in the second, a within group design was used, in

which TOJ task performance by a heterogeneous group of reading impaired

adults was investigated; in the third, using the same TOJ task, the participants

were a similarly heterogeneous group of reading impaired children.

1.3.2.2 Study 2

The second study, reported in Chapter 3 consisted of two experiments

(Experiments 2a and 2b). It was hypothesised that impaired performance of dyslexic participants on the TOJ task may be due to deficits in orienting visual

27 atypical performance on a task designed to elicit peri-saccadic “spatial

compression”, a phenomenon whereby abrupt-onset probe stimuli presented

close in time to the onset of a saccadic eye movement are mislocalized in the

direction of the saccade target. It is thought that these peri-saccadic

mislocalizations may reflect predictive and post-dictive processes implicated in

maintaining spatial constancy across saccades (Lappe et al., 2000; Ross et al., 1997; Ross et al., 2001); specifically, it has been postulated that it reflects the

shifting of the receptive fields of parietal neurons involved in spatiotopic

mapping (Kusunoki and Goldberg, 2003).

In the Experiment 2a, dyslexic and non-dyslexic participants undertook a version

of the task in which the direction of the saccade required was predictable. In the

second, Experiment 2b, the paradigm was modified in such a way as to make the

required saccade direction unpredictable, and thus to eliminate any effects of

anticipatory covert attention. This modification also allowed saccade latencies to

be measured. Participants also undertook a version of the lateralized TOJ task,

so that performance on the two tasks could be correlated. In addition, measures

of real word and non-word reading efficiency were made, and symptoms of Attentional Deficit Disorder were assessed using The Brown Attention Deficit

Disorder Scales (Brown, 1996), in order to ascertain the extent to which variance

on the two tasks accounted for symptoms of Reading and/or Attentional Deficit

Disorder.

28 In the third and final study, comprising Experiment 3 and reported in Chapter 4,

a group of parents having at least one dyslexic child was recruited, giving a

sample of participants likely be at risk of dyslexia or related disorders. In

addition to a version of the lateralized TOJ task, each participant undertook the

Attentional Network Task (Fan et al., 2002), a task designed to measure the

efficiency of three postulated attentional networks: a right-lateralized “alerting” network implicated in sustained attention; an “orienting” network implicated in

directing attention to new locations; and an “executive” network implicated in

resolving conflict (Posner and Petersen, 1990). Again, measures of reading

impairment, and ADHD symptoms were obtained for each participant. These

were correlated with TOJ performance and with the three derived measures from

the ANT task representing, respectively, the efficiencies of the postulated

alerting, orienting and executive attentional networks, in order to ascertain the

proportion of variance in reading impairment, ADHD symptoms, and TOJ

performance attributable to variance in the efficiency of each of these three

networks.

1.3.2.4 Participants and sample design

In carrying out an investigation of this kind, a range of approaches can be taken

to the design of the participant samples. At its simplest, comparisons can be

made between the performance of people with dyslexia and that of people

without. However, there are at least three drawbacks with this approach. Firstly,

29 condition; a “between groups” study design may therefore artificially

dichotomize a continuous variable – reading efficiency. Secondly, dyslexia is

very broadly defined, and has a high rate of overlap with a number of other

disorders, including Dyspraxia; Attention Deficit and Hyperactivity Disorder,

Autistic Spectrum Disorder (ASD); and Specific Language Disorder (SLD)

(American Psychiatric Association, 1994; Kaplan et al., 2001; Ramus, 2004). Thirdly, a between-diagnostic groups approach may tend to suppress the

emergence of dimensions within disorder that extend into the normal range. Thus

while between-groups studies are potentially useful in identifying deficits that

may be associated with reading disorder and thus worth further investigation, any

between-group differences found may reflect neither proximal causes of reading

impairment nor dimensions of the underlying pathology, but simply conditions

that are more prevalent in one group than another.

An alternative approach, therefore, is to consider reading efficiency as a

continuum, and to try to delineate the factors that account for variance in reading

efficiency across the whole spectrum. However, a problem that arises with this

latter approach is that of finding a population in which reading efficiency, and factors associated with variance in reading efficiency, are likely to have a normal

distribution.

A number of sampling designs were therefore used in the course of the

investigation reported here, including between-group as well as within-group

30 (Experiments 1a and 2a) compared dyslexic with non-dyslexic participants, in

order to establish whether there was any overall difference between diagnostic

groups on the cognitive task in question. For the remainder of the experiments,

three approaches were taken to the issue of treating reading efficiency as

continuous rather than dichotomous. The first was to recruit a broad range of

reading impaired participants from the sub-clinically impaired to the severe (Experiments 1b and 1c, reported in Chapter 2). A drawback with this approach

is that at the severe end of the scale, reading ability may arise from compound

causes, while the range at the mild end will be restricted by diagnostic cut-off

criteria.

The challenge, therefore, was to recruit a sample in which the range of reading

efficiency included the normal range, without over-loading the study with

unimpaired participants; in other words to recruit a “dyslexia enriched” sample

with a range that included good readers. Thus, a second approach was to recruit

from a homogeneous population, such as a student body, and to invite

participation from both dyslexic and non-dyslexic students. Provided key

variables could be shown to have a normal distribution across the entire sample, logistic regression could be used to determine the extent to which the variables of

interest predicted recruitment group membership. This approach was used in

Experiment 2b, reported in Chapter 3.

A third approach was to identify a single population in which dyslexia is likely to

31 population. Because dyslexia has been shown to be heritable (Schulte-Korne,

2001), one candidate population is that consisting of adults who have a dyslexic

child. A sample recruited from such a population was anticipated to include both

unimpaired and impaired readers but have a mean degree of reading impairment

that is greater than that of the general population. This sampling methodology

was employed in the final study of the investigation, Experiment 3, reported in Chapter 4.