DiscussionThe myth of the visual word form area
Introduction
Reading involves access to prior knowledge about familiar letter combinations, their meanings and sounds. The visual word form is the combination of letters that make up the word. It is an abstract representation because it does not depend on the perceptual dimensions of the stimulus such as the location, color, size, or font of the letters. There are several ways that representations of visual word forms might be implemented at a cognitive or neural level. Some cognitive models of reading include a “visual word form system” or “orthographic input lexicon” as a specific functional module Marshall and Newcombe 1973, Patterson and Shewell 1987, while others claim that knowledge of familiar letter combinations arises from interactions between orthographic, semantic, and phonological processing without explicit word form representations Plaut et al 1996, Seidenberg and McClelland 1989. Thus it is of particular theoretical interest that Cohen and colleagues (2002) have recently claimed that visual word form representations: “are subtended by a restricted patch of left-hemispheric fusiform cortex [average Talarach coordinates x = −43, y = −54, z = −12] which is reproducibly activated by reading” (p. 1054).
In a series of functional neuroimaging experiments, Cohen et al 2000, Cohen et al 2002 had subjects silently read visual words briefly presented to either the left or right visual hemifield. Written words relative to fixation activated many frontal, temporal, and parietal areas irrespective of the hemifield of presentation (Cohen et al., 2000). The authors, however, focused on a left lateralised midfusiform activation (x = −42, y = −57, z = −6; Z = 8.49) because it was the only area to lie in the occipitotemporal cortex where lesions give rise to pure alexia. Subsequently, Cohen et al. (2002) showed that the same area was more responsive to written words than consonant letter strings (x = −43, y = −54, z = −12; t = 3.42). Along with other evidence that the same area is activated irrespective of typographical case (Dehaene et al., 2001; see also Polk and Farah, 2002) and that it does not respond to same–different matching of auditory words or pseudowords (Dehaene et al., 2002), the authors have labeled this left midfusiform region the “visual word form area” (VWFA). Moreover, they claim it “therefore appears as a unimodal area, at least as far as word perception is concerned” (Cohen et al., 2002, p. 1055).
Potential objections to this hypothesis are considered by Cohen et al. (2002). Specifically, they note that:
… in some studies, words in the auditory or tactile modality have been shown to induce left fusiform activation, which should not be expected if this region was purely devoted to visual processing Binder et al 1996, Buchel et al 1998, Buckner et al 2000, Chee et al 1999, Demonet et al 1994, Demonet et al 1992, D’Esposito et al 1997, Giraud and Price 2001, Perani et al 1998, Pihlajamaki et al 2000, Vandenberghe et al 1996, Wise et al 2000.
They go on to claim, however, that previous studies that demonstrated fusiform activation for stimuli other than visual words activated a fusiform region that was anterior to their proposed visual word form area. According to their review of the literature, fusiform activation for nonvisual stimuli had an average anterior–posterior coordinate of y = −43 (range −54 to −32). In contrast, studies of visual words activate a more posterior region with an average anterior–posterior coordinate of y = −60 (range −43 to −70). Cohen et al. (2002) therefore propose that there are several distinct areas in the left midfusiform gyrus and the “VWFA proper would occupy the middle portion of the left fusiform gyrus, with Talairach co-orindates (TC) close to y = −60, while increasingly more abstract and supramodal representations would occupy its more anterior sectors” (p. 1054). Even if nonvisual word stimuli do activate the “VWFA proper,” Cohen et al. (2002) suggest that this could be due to “top-down” processing such as mental imagery activating visual word forms. As we argue below, however, these arguments are not consistent with either the neuropsychological nor functional imaging evidence.
Section snippets
Neuropsychological evidence
In support of their hypothesis, Cohen et al. (2002) claim that the left midfusiform area that they refer to as the VWFA “seems to be the critical lesion site for pure alexia, a unimodal deficit of word reading with sparing of writing and of auditory word comprehension Beversdorf et al 1997, Binder and Mohr 1992, Damasio and Damasio 1983, Dejerine 1892, Leff et al 2001” (p. 1055). A closer inspection of this claim, however, reveals that there are no neuropsychological studies that have
Functional imaging evidence
In the following section we report data from several previously published imaging studies to demonstrate that, in addition to reading, the precise region of the left mid fusiform gyrus identified by Cohen and colleagues as the VWFA is engaged in tasks without any visual word form component. We begin by illustrating activation for reading that identifies the left midfusiform area referred to by Cohen and colleagues. We then demonstrate that activation in this region is not specific to visual
Reading activation
The top row of Fig. 1 (see also Table 1) illustrates activation when normal subjects read visual words relative to rest (yellow) and relative to false fonts (red) as reported in Brunswick et al. (1999). The crosshairs indicate the peak coordinates for the VWFA (x = −42, y = −57, z = −15) according to Cohen et al. (2002). As can be seen, the only left lateralized region of the occipitotemporal cortex to be activated for words relative to false fonts corresponds exactly to the left midfusiform
Object recognition/naming
The same area, however, is also engaged when subjects name, view, or generate verbs to pictures of objects Bookheimer et al 1995, Etard et al 2000, Murtha et al 1999 yet these tasks do not entail visual word form processing. Fig. 1 (row 2) illustrates the activation found when subjects named pictures of objects relative to saying “OK” to the same pictures (Moore and Price, 1999). This effect cannot be due to “implicit” activation of visual word forms when subjects name pictures because
Visual form processing
Martin and Chao (2001) have suggested that midfusiform activation is driven by object form. Thus one possibility is that the visual word form area is not specific to orthography but is also involved in processing object structure more generally. However, this “visual form” hypothesis cannot be the complete explanation. For instance, activation is also observed when subjects name the color of meaningless visual gratings relative to saying “OK” to the same gratings, where visual input is matched
Naming
The association of the left midfusiform area with naming has previously been noted in lesion and electrical stimulation studies Burnstine et al 1990, Foundas et al 1998, Luders et al 1986, Luders et al 1991, Raymer et al 1997. The problem with this account is that functional imaging activation in the left midfusiform area is not specific to naming. For instance, it is also seen when subjects use a manipulandum to make manual “twist” or “pour” action responses to pictures of familiar objects
Visual processing
Finally, activation in the left midfusiform area is not even limited to visual processing. It is also observed when subjects hear, repeat, and think about the meaning of auditory words (Price et al., 2002, see Fig. 1, row 8); hear words and make rhyming (e.g., soap-hope) judgments Booth et al 2002a, Booth et al 2002b, see Table 2); hear definitions of objects and make semantic decisions (Thompson-Schill et al., 1999, see Table 2) or imagine the object (Mellet et al., 1998, see Table 2); and
Acknowledgements
This work was funded by the Wellcome Trust. We thank Karl Friston, Eraldo Paulesu, David Green, and two anonymous reviewers for their helpful comments.
References (65)
- et al.
Visual complexity in letter-by-letter reading: “pure” alexia is not pure
Neuropsychologia
(1998) - et al.
Functional anatomy of intra- and cross-modal lexical tasks
Neuroimage
(2002) - et al.
Differential activation of right and left posterior sylvian regions by semantic and phonological tasksa positron emission tomography study
Neurosci. Lett.
(1994) - et al.
A functional MRI study of mental image generation
Neuropsychologia
(1997) - et al.
Recovery from alexia without agraphiareport of an autopsy
Cortex
(1979) - et al.
Semantic memory and the brainstructure and processes
Curr. Opin. Neurobiol.
(2001) - et al.
Three distinct ventral occipitotemporal regions for reading and object naming
Neuroimage
(1999) - et al.
Cognitive conjunctiona new approach to brain activation experiments
Neuroimage
(1997) - et al.
Cognitive neuropsychological analysis and neuroanatomic correlates in a case of acute anomia
Brain Lang.
(1997) - et al.
A neural basis for category and modality specificity of semantic knowledge
Neuropsychologia
(1999)