Distinct unimodal and multimodal regions for word processing in the left temporal cortex
Introduction
In literate adults, the perception of a written word gives access, within half a second, to a wide variety of representations and processes ranging from orthography and phonology to semantics and articulation (Marinkovic et al., 2003). Accordingly, word reading is correlated with the activation of extensive bilateral cerebral networks, with left-sided predominance (for reviews, see for example, Fiez and Petersen, 1998, Jobard et al., 2003, Price et al., 2003a). However, the efficacy of all those processing stages first depends on the fast and parallel identification of strings of letters by the visual system, which may be seen as the gateway into the reading network (Besner, 1989, Nazir, 2000, Paap et al., 1984, Pelli et al., 2003). Among the visual areas that are consistently activated during reading, we proposed that a region of the left inferotemporal cortex plays a crucial role in this perceptual expertise (Cohen et al., 2000). On the basis of imaging and neuropsychological data, we suggested that this region (the visual word form area or VWFA) computes a representation of abstract letter identities from visual input, a representation invariant for irrelevant parameters such as size, location, font, or case (for reviews, see Cohen and Dehaene, in press, McCandliss et al., 2003).
There are still ongoing controversies about the functional properties of the VWFA, and particularly about its involvement in the processing of auditory or even Braille words (Büchel et al., 1998, Cohen and Dehaene, in press, Price et al., 2003b). In a recent study using functional magnetic resonance imaging (fMRI), we contrasted the processing of written and spoken words while subjects performed a same–different task on consecutive words (Dehaene et al., 2002). In agreement with our hypotheses, the left inferotemporal cortex was activated by visual words in every subject, while there was no activation by spoken words. However, a number of other studies have demonstrated left inferotemporal activations during the perception of auditory words (e.g., see Binder et al., 1996, Büchel et al., 1998, Buckner et al., 2000, Chee et al., 1999, Démonet et al., 1992, Démonet et al., 1994, D'Esposito et al., 1997, Giraud and Price, 2001, Perani et al., 1998, Pihlajamäki et al., 2000, Vandenberghe et al., 1996, Wise et al., 2000). As discussed in Cohen et al. (2002), this apparent discrepancy may result from at least two causes.
First, the left inferotemporal cortex may encompass several distinct areas involved in word processing, each area having distinctive patterns of activation to written or spoken words. Due to their close proximity, such regions may be difficult to distinguish, particularly when comparing the coordinates of activations across subjects and across studies, including group PET studies with a relatively low spatial resolution. Notwithstanding those methodological limitations, a review of ventral temporal activations suggests that words in nonvisual modalities yield activations more anterior (average y = −43) than the visual activations typical of the VWFA (average y = −60) (Cohen et al., 2002). Furthermore, the anterior activations are more sensitive to the semantic demands of the task, whereas posterior activations were observed even for visual pseudowords relative to random letter strings. Thus, the VWFA should possibly be distinguished from more anterior regions that are increasingly multimodal and engaged in semantic computations (for a convergent metaanalysis, see Jobard et al., 2003). Fine-grained spatial distinctions between inferotemporal regions with different modal selectivity have also been shown in studies of object perception (Amedi et al., 2001). To clarify this issue, one should resort to imaging techniques with a high spatial resolution and perform individual analyses to override the limitations due to interindividual anatomical variations (Sowell et al., 2002, Thompson et al., 1996).
A second source of difficulty in identifying the modality-dependent properties of inferotemporal cortex is that all visual regions, from area V1 to high-level inferotemporal cortex, can be activated in the absence of any visual input, depending on the requirements of the task (Pessoa et al., 2003, Somers et al., 1999). For instance, top-down activations have been evidenced in regions close to the VWFA during the mental imagery of faces or places (Ishai et al., 2000, O'Craven and Kanwisher, 2000). Similarly, the left inferotemporal cortex is activated when Japanese subjects write complex kanji ideograms or imagine doing so, a task with a strong visual component (Nakamura et al., 2000, Tokunaga et al., 1999). Such top-down influences may contribute to inferotemporal activations that are observed with auditory words, provided that the task involves some form of orthographic manipulation. For instance, Booth et al., 2002a, Booth et al., 2003 observed left inferotemporal activations when subjects performed a spelling task on auditory words, but not when they performed a rhyming task on the same stimuli (see also Burton et al., 2000). Thus, clarifying the role of task on the pattern of left inferotemporal activations requires that modality and task be independently manipulated.
We may now summarize the questions that are still open relative to the effect of word modality on left inferotemporal activations, distinguishing bottom-up from top-down influences. First, is there a unimodal visual area in the left inferotemporal region associated with visual word recognition (the VWFA)? Unimodality implies that this region should receive direct input from lower-level visual cortex, but not from equivalent auditory or tactile input systems. However this does not preclude some top-down activation by spoken words, but only when required by specific task demands. Second, if there is such a unimodal region, can it be distinguished from multimodal inferior temporal regions whose activation pattern would not crucially depend on input modality during word processing? Third, how are those areas affected by task demands, and particularly, can the VWFA be activated during auditory word processing if the task requires access to a visual or orthographic representation? Fourth, can equivalent areas be defined for spoken words?
To address those issues, we recorded brain activation with fMRI while manipulating stimulus modality (written vs. spoken words) and task demands (letter feature detection vs. phoneme detection). Letter feature detection (detecting whether a word contains a descending lowercase letter) was intended to activate visual or orthographic representations, even with auditory stimuli. Conversely, phoneme detection (detecting whether a word contains a specific phoneme) was intended to activate phonological representations even with written words.
We further combined this orthogonal task × modality design with the priming method (Naccache and Dehaene, 2001), by asking which areas are sensitive to word repetition and, if so, whether they are sensitive to cross-modal as well as to within-modality repetitions. On half the trials, the target word was a repetition of the previous word, either in the same modality or in a different modality. We expected that many brain areas would show a reduced activation on such repeated trials (repetition suppression) and that the pattern of repetition suppression would provide further confirmation of modality specificity: unimodal regions would be sensitive exclusively to word repetition within the corresponding modality, while multimodal regions may show comparable repetition effects within and between modalities.
Note, however, that our design investigated conscious rather than subliminal repetition priming. One drawback of conscious priming is that, once subjects become aware of the repetition, they may strategically alter their response decisions. In that respect, priming may help distinguish bottom-up from top-down activations. On any given trial, repeated words must follow essentially the same path as nonrepeated words in the visual and auditory cortices, at least up to the point where repetition is consciously detected. Subjects may then bypass the detection of the target phoneme or letter and directly trigger the same response as on the previous trial. This model of conscious priming would predict that top-down activations reflecting task-dependent processes should be reduced by repetition more than bottom-up activations chiefly depending on stimulus modality.
Section snippets
Subjects
Seventeen subjects (10 females, 7 males), aged 20–30 years, with university education, fully right-handed according to the Edinburgh Inventory, participated in the study. All were drug-free, had no neurological or psychiatric history, and had normal anatomical MRIs. All gave their written informed consent. The experiment was approved by the Ethical Committee of the Hôpital de Bicêtre.
Tasks and stimuli
Subjects received lists of randomly mixed auditory and visual words. Before each list, they were instructed to
Discussion
We will first summarize and sort out the main patterns of activation observed across the whole brain, successively considering the effects of modality, task, and word repetition and highlighting the parallels between activations and behavioral data. We will then focus on a more detailed discussion of activations in the left inferotemporal cortex. Finally, we will discuss methodological issues raised by the present study.
References (99)
- et al.
Human temporal-lobe response to vocal sounds
Brain Res., Cogn. Brain Res.
(2002) - et al.
Functional anatomy of intra- and cross-modal lexical tasks
NeuroImage
(2002) - et al.
On the status of object concepts in aphasia
Brain Lang.
(1997) - et al.
Specialization within the ventral stream: the case for the visual word form area
Neuroimage
(2004) Time-locked multiregional retroactivation: a systems-level proposal for the neural substrates of recall and recognition
Cognition
(1989)- et al.
Towards a cognitive neuroscience of consciousness: basic evidence and a workspace framework
Cognition
(2001) - et al.
A functional MRI study of mental image generation
Neuropsychologia
(1997) - et al.
Attention to single letters activates left extrastriate cortex
NeuroImage
(2004) - et al.
The planum temporale as a computational hub
Trends Neurosci.
(2002) - et al.
fMR-adaptation: a tool for studying the functional properties of human cortical neurons
Acta Psychol. (Amsterdam)
(2001)
Where is the /b/ in “absurde” [apsyrd]? It is in French listeners' minds
J. Mem. Lang.
Eccentricity bias as an organizing principle for human high-order object areas
Neuron
Large-scale mirror-symmetry organization of human occipito-temporal object areas
Neuron
Distributed neural systems for the generation of visual images
Neuron
Evaluation of the dual route theory of reading: a metaanalysis of 35 neuroimaging studies
NeuroImage
Feedforward, horizontal, and feedback processing in the visual cortex
Curr. Opin. Neurobiol.
The role of primary visual cortex (V1) in visual awareness
Vis. Res.
Spatiotemporal dynamics of modality-specific and supramodal word processing
Neuron
The visual word form area: expertise for reading in the fusiform gyrus
Trends Cogn. Sci.
Three distinct ventral occipitotemporal regions for reading and object naming
NeuroImage
Perception and awareness in phonological processing: the case of the phoneme
Cognition
Traces of print along the visual pathway
The myth of the visual word form area
Neuroimage
Normal and pathological reading: converging data from lesion and imaging studies
NeuroImage
Cortical localisation of the visual and auditory word form areas: a reconsideration of the evidence
Brain Lang.
Cognitive neuropsychological analysis and neuroanatomic correlates in a case of acute anomia
Brain Lang.
Disruption of posterior brain systems for reading in children with developmental dyslexia
Biol. Psychiatry
Neural systems for compensation and persistence: young adult outcome of childhood reading disability
Biol. Psychiatry
The influence of orthography on phonological representations in the lexicon
J. Mem. Lang.
Noun imageability and the temporal lobes
Neuropsychologia
Human extrastriate visual cortex and the perception of faces words numbers and colors
Cereb. Cortex
Electrophysiological studies of human face perception. I: potentials generated in occipitotemporal cortex by face and non-face stimuli
Cereb. Cortex
Visuo-haptic object-related activation in the ventral visual pathway
Nat. Neurosci.
A case of pure word deafness with autopsy
J. Nerv. Ment. Dis.
Adaptation to speaker's voice in right anterior temporal lobe
NeuroReport
On the role of outline shape and word-specific visual pattern in the identification of function words: NONE
Q. J. Exp. Psychol.
Function of the left planun temporale in auditory and linguistic processing
Brain
Human temporal lobe activation by speech and nonspeech sounds
Cereb. Cortex
Neural correlates of lexical access during visual word recognition
J. Cogn. Neurosci.
Modality independence of word comprehension
Hum. Brain Mapp.
Relation between brain activation and lexical performance
Hum. Brain Mapp.
Explicit and implicit processing of words and pseudowords by adult developmental dyslexics: a search for Wernicke's Wortschatz?
Brain
A multimodal language region in the ventral visual pathway
Nature
Functional MRI evidence for a role of frontal and inferior temporal cortex in amodal components of priming
Brain
The role of segmentation in phonological processing: an fMRI investigation
J. Cogn. Neurosci.
Auditory and visual word processing studied with fMRI
Hum. Brain Mapp.
The visual word form area: spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients
Brain
Language-specific tuning of visual cortex? Functional properties of the visual word form area
Brain
Visual word recognition in the left and right hemispheres: anatomical and functional correlates of peripheral alexias
Cereb. Cortex
Cited by (226)
Developing cortex is functionally pluripotent: Evidence from blindness
2024, Developmental Cognitive NeurosciencePoor reading is characterized by a more connected network with wrong hubs
2021, Brain and LanguageNeuroimaging of valence decisions in children and adults
2021, Developmental Cognitive Neuroscience