The three broad aims of this thesis, as laid out in the Introduction, were:
1) to address questions about the normal reading system that are key to understanding the nature of acquired reading disorders
2) to identify what reading disorders can tell us about the intact reading system
3) to test the efficacy of a reading training programme for patients with acquired reading disorders.
With regards to the first aim, the work presented in this thesis investigated the nature of the visual input system for reading (Chapter 3.1), the case for word-preferential processing in the left vOT cortex (Chapters 3.1, 3.2 and 3.5) and whether lateralisation of activity in the vOT cortex is sensitive to the spatial frequency content of visual stimuli (Chapter 3.5).
Chapter 3.1 demonstrated that the initial visual activation during word reading is contralateral to visual presentation, but that word-preferential activity in the vOT cortex is strictly left lateralised. For patients with an RHH, this means that activity in the right visual cortex has to be transferred via the splenium of the corpus callosum to the left vOT. Activity in the left vOT was demonstrated to be preferential for words over perceptually-matched stimuli, irrespective of the location of presentation.
Chapters 3.1 and 3.2 characterised the response profile of the left vOT cortex. The first study (using parafoveal stimulus presentation) showed a profile of words > false font > numbers. The second study aimed to replicate this finding using a central presentation location and a number reading task that was more closely matched to the other conditions. The profile observed was words = false font > numbers. Hence, it is unclear whether the left vOT cortex is preferential for words over false font stimuli, but it is evident that it is preferential for words or word-like stimuli over numbers. It is hard to
reconcile this finding with the LCD model of reading, which would predict that the left vOT cortex is specialised for word recognition, and should not show a differential response to different categories of non-word written stimuli such as false font and numbers. Instead, the results may reflect the impact of task-dependent modulation on the left vOT. According to this view, the reading task elicits top-down modulation from the left frontotemporal
language network onto the left vOT cortex. Number reading does not engage the same task-dependent state, as the ultimate aims of digit recognition are numerical rather than verbal. Further work is currently in progress to
discriminate between the effects of orthography type (words versus digits) and task (verbal versus numerical) on activity in the left vOT. If this work demonstrates that task-demands significantly affect activity in the left vOT, it would suggest that damage to the backward temporal-to-occipital
connections could play a role in the word recognition deficits in patients with PA, as well as the forward occipital-to-temporal connections.
Finally, Chapter 3.5 showed evidence suggesting that the lateralisation of word processing to the left vOT may be related to the spatial frequency content of written words. This work predicts that patients with left
occipitotemporal damage may have a general impairment in sensitivity to high spatial frequencies, which would fit in with the perceptual view of pure alexia.
To address the second aim, Chapters 3.3 and 3.4 presented behavioural and functional imaging data from a group of patients with acquired peripheral reading disorders. Chapter 3.3 showed that patients with damage to the left occipitotemporal cortex have impaired word, text and letter reading ability. A deficit at the level of single letters argues against the simultanagnostic view of PA. A domain-general perceptual deficit, or a domain-specific deficit at the level of single letters is more likely. Further work into the perceptual abilities of patients with left occipitotemporal damage is required to distinguish
between these two possibilities. If the perceptual view of PA was proven to be correct, this would also imply that the role of the left vOT cortex in
healthy volunteers is domain-general rather than being specialised for written words.
The structural lesion-symptom mapping reported in Chapter 3.3 indicated that damage to the left occipitotemporal white matter, in the vicinity of the ILF, is associated with slow word reading. The existing literature on the
structural pathology of PA (Damasio & Damasio, 1983; Binder & Mohr, 1992;
Leff et al. 2006) has relied on structural overlay mapping to demonstrate the areas most commonly damaged in patients with PA. The lesion-symptom mapping approach used here is an improvement on this methodology, as it takes into account the severity of the reading impairment, rather than using a binary classification. The dataset reported in Chapter 3.3 would benefit from the inclusion of more patients, and particularly patients with more severe reading disorders, before firm conclusions can be drawn about the critical lesion site for PA.
The final aim of this thesis was addressed by Chapters 3.3 and 3.4, which reported the behavioural and functional imaging results of a training study.
The first study tested the efficacy of an intensive, computer-based, whole-word recognition training protocol. As reported in Chapter 3.3, this training resulted in an average reduction in reading speed of roughly 110ms per word. However, this improvement was not statistically significant, and it did not persist until the follow-up reading assessment a few weeks after
cessation of the training period. The failure of the reading training was
reflected in the lack of positive training effects in the functional imaging data (Chapter 3.4). These findings suggest that the impairment caused by left occipitotemporal damage cannot be ameliorated by functional reorganisation and is insensitive to training – at least of the kind used in this study.
Regardless of whether the role of the vOT is reading-specific or domain-general, the data presented here demonstrates that it is critical for rapid whole-word recognition and cannot easily be replaced by compensatory activity elsewhere in the brain.