The role of the left anterior temporal lobe in exception word reading: Reconciling patient and neuroimaging findings
Highlights
► We studied the role of the anterior temporal lobes in exception word (EW) reading. ► Atrophy in the left anterior temporal lobe was associated with surface dyslexia. ► The same region was activated during EW reading in healthy subjects. ► Patient and functional data consistently show the role of this region in EW reading.
Introduction
Major computational models of reading aloud indicate the existence of at least two different procedures by which the translation of an alphabetic letter string or word into a verbal pronunciation is successfully achieved. The first of these are subword processes, which employ regular orthography-to-phonology mappings and enable the pronunciation of words that the reader has never encountered or words without semantic representation. This latter type of stimulus is referred to as pseudowords in experimental investigations, e.g., ‘ked’ and ‘voot’. The second type of written word processing includes whole-word processes, which employ idiosyncratic item-specific information regarding the pronunciation of a particular word. This process would be indispensable for reading words with atypical spelling-to-sound correspondences (i.e., exception words such as ‘yacht’ and ‘colonel’). Although the basic distinction between subword and whole word processing is widely accepted, the nature of these procedures and their interactions with different components of the language processing system are still a matter of debate.
According to the dual route cascaded (DRC) model (Coltheart, 2006, Coltheart et al., 2001), these two procedures represent two distinct reading-specific pathways that function independently of one another and relatively independently from other components of language processing systems. Conversely, according to the connectionist triangle model (Patterson and Hodges, 1992, Seidenberg and McClelland, 1989), these two procedures are not specific to reading but rather determined by the modulation of connection weights between orthography, phonology and semantics within the language network architecture. In particular, readers would rely more on the activation of phonology from orthography during the reading of regular and pseudowords, whereas they would rely on the activation of phonology from semantics during less frequent exception word reading. Thus, it is not surprising and is indeed predicted by the connectionist triangle model that difficulties in reading exception words but not regular and pseudowords (i.e., surface dyslexia) are accompanied by semantic deficits (Brambati et al., 2009a, Jefferies et al., 2004, Patterson and Hodges, 1992, Woollams et al., 2007). Consistent with this viewpoint, the purest cases of surface dyslexia are reported in patients with semantic dementia (SD) (Brambati et al., 2009a, Gorno-Tempini et al., 2004a, Jefferies et al., 2004, Patterson and Hodges, 1992, Patterson et al., 2006, Wilson et al., 2009, Woollams et al., 2007).
Semantic dementia is a neurodegenerative disease characterised by a progressive loss of semantic knowledge (Gorno-Tempini et al., 2004a, Hodges et al., 1992, Neary et al., 1998, Snowden et al., 1989). Even at the earliest stages of the disease, most SD patients exhibit a reading profile that is characterised by surface dyslexia, a tendency to mispronounce exception words; this deficit is more severe for low-frequency words. In particular, exception word reading is characterised by “over-regularisation” errors (e.g., ‘sew’ is pronounced to rhyme with ‘sue’, and ‘pint’ with ‘mint’). This has been explained by an overreliance on the orthography-to-phonology mappings, which are used for the pronunciation of regular words and pseudowords. These patients are characterised by anterior temporal lobe (ATL) atrophy that is generally more severe in the left hemisphere (Gorno-Tempini et al., 2004a, Noppeney et al., 2007, Wilson et al., 2009). Also, the level of atrophy in the left anterior temporal lobe correlates with the severity of exception word reading difficulties (Brambati et al., 2009a).This evidence highlights the crucial role of the left ATL in the brain network mediating the reading of exception words. However, studies have yet to consistently confirm the involvement of the left ATL in whole-word reading processes in healthy subjects (Mechelli et al., 2005, Wilson et al., 2009). Several PET and fMRI studies of young, healthy subjects have attempted to map the brain regions implicated in whole-word reading processes by contrasting the activation pattern associated with exception and pseudoword reading. The results of these studies have consistently indicated the involvement of the pars triangularis of the left inferior frontal gyrus and the anterior portion of the left fusiform gyrus (Brunswick et al., 1999, Mechelli et al., 2005, Mechelli et al., 2003a, Price et al., 2003), but not of the left ATL (Binder et al., 2005, Mechelli et al., 2005, Wilson et al., 2009).
This discrepancy between patient and healthy subject data could be due to various methodological factors associated with the functional neuroimaging reading protocols used for healthy participants. First, it is notoriously difficult to obtain a strong BOLD fMRI signal from the ATLs. This is because of the proximity of the bone and air-filled sinuses in anterior temporal structures (Devlin et al., 2000, Visser et al., 2010). Second, some functional neuroimaging studies have employed mostly middle- to high-frequency words as stimuli, primarily representing highly familiar concepts. For instance, Mechelli et al. (2005) have used words with a mean frequency of 40.8 occurrences per million (range: 1 to 447). Using this middle range frequency they failed to find an ATL significant activation for exception word reading. Wilson et al. (2009) orthogonally manipulated word type (regular and exception words) and frequency (high and low). Low frequency items had a mean frequency (7.3 occurrences per million) comparable to that of the items used in the present study. With such a low frequency range for exception words Wilson and colleagues found a trend towards significance for the activation in the left pars triangularis but not in the ATL. However, the fact that these analyses were based on only 20 low frequency exception words, together with a different acquisition sequence, may explain their lack of power to detect a significant effect. Nonetheless, it is possible that the reading of higher frequency exception words can be achieved without the engagement of the left anterior lateral temporal cortex, which is generally more involved in the semantic processing of less prototypical stimuli (Brambati et al., 2006). This would also explain why high-frequency word reading is generally more preserved in SD patients (Jefferies et al., 2004, Patterson and Hodges, 1992, Woollams et al., 2007).
Alternatively, this inconsistency between data from patients and healthy subjects can be explained by the fact that the left ATL is not directly involved in the brain network mediating exception word reading. In this latter case, the association between ATL atrophy and the exception word reading difficulties observed in SD patients would reflect, for example, damage to a distributed network with connections that pass through the atrophy site rather than damage to a crucial region of the network itself. In this case, ATL would be indirectly implicated in the process and, for this reason, its lesion would impair exception word reading but it would not play a crucial role in the network for reading such words in healthy participants. Another possible explanation not addressed in the present study (Seghier et al., 2008, Woollams et al., 2007) proposes that the degree to which individual subjects use semantics during reading may vary. This could explain the inconsistencies found in some SD patients as well as normal readers.
To address this controversy, we present two studies with the following aims: (1) mapping the brain regions of grey matter (GM) atrophy using voxel-based morphometry in a patient with mild SD exhibiting a surface dyslexia profile; and (2) mapping the functional activation pattern associated with low-frequency exception word reading compared to pseudowords in young, healthy participants using functional magnetic resonance imaging (fMRI) acquisition parameters that were optimised to image the ATLs (Brambati et al., 2010).
If the left ATL is a crucial region for the brain network involved in exception word reading, the results from patient anatomical data and functional data from healthy subjects should converge. In particular, we should observe significant brain activation in healthy participants when comparing exception and pseudoword reading in the same left ATL regions that exhibit GM atrophy in the SD patient suffering from surface dyslexia.
Section snippets
Study One: grey matter atrophy in an SD patient suffering from surface dyslexia
The aim of the first study was to characterise the pattern of GM atrophy using voxel-based morphometry (VBM) in an SD patient exhibiting a reading pattern that can be characterised as “surface dyslexia” (Marshall and Newcombe, 1973). The patient presented with isolated difficulties in reading exception words, especially those with a low frequency. The patient reading assessment was performed using the experimental protocol detailed below.
Study Two: brain activation associated with exception word reading in young healthy subjects
The present study aimed to determine whether the left ATL is involved in the functional network mediating whole-word processes in the context of low-frequency exception word reading in young, healthy subjects. Specifically, we mapped the activation pattern associated with low-frequency exception word reading compared to the reading of pseudowords in young, healthy participants using fMRI imaging acquisition parameters that were optimised for imaging the ATLs. The same word stimuli used in the
General discussion
The present findings reconcile data from neuropsychological studies of SD patients and functional imaging of healthy subjects. In particular, we demonstrate that one of the regions of the left ATL that was most atrophic in the patient with SD who presented with surface dyslexia was significantly activated in healthy individuals engaged in an exception word reading task.
In Study One, we described the GM atrophy profile in an SD patient manifesting a surface dyslexia profile. The left ATL regions
Acknowledgments
This work was supported by the Québec Bio-Imaging Network and the Réseau Québécois de Recherche sur le Vieillissement. SMB is supported by a Chercheur boursier award from the Fonds de la recherche en santé du Québec (FRSQ). We would like to thank Valérie Dostie for technical support. The authors thank American Journal Experts (http://www.journalexperts.com/) for English editing.
References (59)
A fast diffeomorphic image registration algorithm
Neuroimage
(2007)- et al.
Voxel-based morphometry — the methods
Neuroimage
(2000) - et al.
Unified segmentation
Neuroimage
(2005) - et al.
Some neurophysiological constraints on models of word naming
Neuroimage
(2005) - et al.
The role of the left anterior temporal lobe in the semantic processing of famous faces
Neuroimage
(2010) - et al.
Reading disorders in primary progressive aphasia: a behavioral and neuroimaging study
Neuropsychologia
(2009) - et al.
Atrophy progression in semantic dementia with asymmetric temporal involvement: a tensor-based morphometry study
Neurobiol. Aging
(2009) - et al.
Unitary vs multiple semantics: PET studies of word and picture processing
Brain Lang.
(2004) - et al.
Optimized EPI for fMRI studies of the orbitofrontal cortex
Neuroimage
(2003) - et al.
Anatomic constraints on cognitive theories of category specificity
Neuroimage
(2002)
Susceptibility-induced loss of signal: comparing PET and fMRI on a semantic task
Neuroimage
A voxel-based morphometric study of ageing in 465 normal adult human brains
Neuroimage
Cognitive and behavioral profile in a case of right anterior temporal lobe neurodegeneration
Cortex
The impact of semantic memory impairment on spelling: evidence from semantic dementia
Neuropsychologia
Comparing event-related and epoch analysis in blocked design fMRI
Neuroimage
The assessment and analysis of handedness: the Edinburgh inventory
Neuropsychologia
Deterioration of word meaning: implications for reading
Neuropsychologia
Word and picture matching: a PET study of semantic category effects
Neuropsychologia
Normal and pathological reading: converging data from lesion and imaging studies
Neuroimage
Distributional assumptions in voxel-based morphometry
Neuroimage
Inter-subject variability in the use of two different neuronal networks for reading aloud familiar words
Neuroimage
Hemispheric dissociation in access to the human semantic system
Neuron
A quantitative evaluation of cross-participant registration techniques for MRI studies of the medial temporal lobe
Neuroimage
Familiarity Affects Neuroanatomical Concept Representation
Explicitly and implicit processing of words and pseudowords by adult developmental dyslexics
Brain
The neural correlates of verbal and nonverbal semantic processing deficits in neurodegenerative disease
Cogn. Behav. Neurol.
Acquired dyslexias and the computational modelling of reading
Cogn. Neuropsychol.
DRC: a dual route cascaded model of visual word recognition and reading aloud
Psychol. Rev.
The human brain: surface, three-dimensional sectional anatomy with MRI, and blood supply
Cited by (36)
The neural correlates of sub-lexical semantics and its integration with the lexical meaning in reading Chinese characters
2024, Journal of NeurolinguisticsDoes surface dyslexia/dysgraphia relate to semantic deficits in the semantic variant of primary progressive aphasia?
2019, NeuropsychologiaCitation Excerpt :However, future studies could explore the link between irregulars and semantic vs. lexical capacities in healthy participants by using more continuous markers/measures of reading/writing which might provide the necessary data variability as for example millisecond-measures of reading times for irregulars. Our findings are in line with previous studies showing the existence of surface dyslexia in sv-PPA (e.g., Patterson and Hodges, 1992; Caine et al., 2009; Brambati et al., 2009; Wilson et al., 2012), demonstrating poor performance on lexical decision in sv-PPA (Graham et al., 2000; Rogers et al., 2004; Benedet et al., 2006; Patterson et al., 2006; Jefferies et al., 2010), and suggesting lexical disorders in this PPA variant (Mesulam et al., 2013, 2014; Wilson et al., 2014, 2017; Boukadi et al., 2016). Our data extend and enrich previous evidence, and challenge several findings suggesting that surface dyslexia is specifically related to semantic breakdown (e.g., Macoir and Bernier, 2002; Brambati et al., 2009; Wilson et al., 2012).