Abstract

Reading requires the neural integration of visual word form information that is split between our retinal hemifields. We examined multiple visual cortical areas involved in this process by measuring fMRI responses while observers viewed words that changed or repeated in one or both hemifields. We were specifically interested in identifying brain areas that exhibit decreased fMRI responses as a result of repeated versus changing visual word form information in each visual hemifield. Our method yielded highly significant effects of word repetition in a previously reported visual word form area (VWFA) in occipitotemporal cortex, which represents hemifield-split words as whole units. We also identified a more posterior occipital word form area (OWFA), which represents word form information in the right and left hemifields independently and is thus both functionally and anatomically distinct from the VWFA. Both the VWFA and the OWFA were left-lateralized in our study and strikingly symmetric in anatomical location relative to known face-selective visual cortical areas in the right hemisphere. Our findings are consistent with the observation that category-selective visual areas come in pairs and support the view that neural mechanisms in left visual cortex—especially those that evolved to support the visual processing of faces—are developmentally malleable and become incorporated into a left-lateralized visual word form network that supports rapid word recognition and reading.

INTRODUCTION

Reading relies on our ability to rapidly recognize words. Each time we fixate a word, its comprising letters become split at the vertical midline between our right and left visual hemifields. A consequence of hemifield-splitting is that visual word form information in each hemifield is routed directly from the back of the eye to primary visual cortex in either the left or right hemisphere (Lavidor & Walsh, 2004; Shillcock, Ellison, & Monaghan, 2000) and must therefore be neurally reunited through interhemispheric communication (Ellis & Brysbaert, 2010; Brysbaert, 2004). Despite considerable progress in understanding interhemispheric transfer of visual information in the human brain (Berlucchi, 2014), the neural basis of hemifield-split word integration is not known. We used fMRI to dissociate neural populations that represent hemifield-split visual word form information as a coherent whole from those that represent partial word form information in each hemifield.

Many previous fMRI studies of word recognition have shown evidence of a highly specialized visual word form area (VWFA) in left occiptotemporal cortex. Neurons in the VWFA support rapid word recognition by representing words independent of retinal location, size, and other word-specific properties such as letter case and font (Dehaene & Cohen, 2011; Cohen & Dehaene, 2004; Dehaene et al., 2004; McCandliss, Cohen, & Dehaene, 2003; Cohen et al., 2000, 2002). Neurons in the VWFA bind individual letters into whole words (Glezer, Jiang, & Riesenhuber, 2009; Dehaene et al., 2004) but also show selectivity for nonword symbols (Muayqil, Davies-Thompson, & Barton, 2015; Shin et al., 2008). Even if VWFA neurons are not specialized for reading per se (Vogel, Petersen, & Schlaggar, 2014; Price & Devlin, 2003, 2004), VWFA neurons exhibit both anatomical and functional properties that would be useful for the integration of hemifield-split words into neurally merged orthographic representations (Glezer, Kim, Rule, Jiang, & Riesenhuber, 2015; Bouhali et al., 2014; Vogel, Petersen, & Schlaggar, 2012; Molko et al., 2002). We therefore expected the VWFA in left occipitotemporal cortex (OT) to be a primary cortical locus of hemifield-split word integration. We also expected to find an additional more posterior site of hemifield-split word integration. This conjecture was based on the observation that category-selective visual cortical regions generally come in anterior–posterior pairs (Taylor & Downing, 2011; Schwarzlose, Swisher, Dang, & Kanwisher, 2008), such as the fusiform face area (FFA; Kanwisher, McDermott, & Chun, 1997) and the occipital face area (OFA; Gauthier et al., 2000), both of which are frequently implicated in fMRI studies of face recognition. One view is that the OFA generates an initial representation of face features or parts that are subsequently integrated with respect to spatial configuration in the FFA (Pitcher, Walsh, & Duchaine, 2011; Liu, Harris, & Kanwisher, 2010; Pitcher, Walsh, Yovel, & Duchaine, 2007). We therefore hypothesized the existence of a secondary VWFA—a more posterior occipital word form area (OWFA) in the left hemisphere—that works together with the VWFA to represent hemifield-split letter strings as whole words. Furthermore, given that word recognition relies on neurons that evolved to support face recognition (Ventura, 2014; Dehaene & Cohen, 2007), we predicted the anatomical location of the prospective OWFA in the left hemisphere to be somewhat mirror-symmetric to that of the OFA in the right hemisphere, as is the case for the VWFA and FFA.

We presented observers with hemifield-split words that either repeated or changed in one of four possible ways: (1) the whole word repeated, (2) the whole word (all letters) changed, (3) the letters in the left (but not the right) half of the word changed, or (4) the letters in the right (but not the left) half of the word changed. The general logic of our main experiment was based on an expected reduction of visual cortical fMRI responses to repeated stimuli, which is sometimes referred to as fMRI “adaptation” or “repetition suppression” (Malach, 2012; Grill-Spector, Henson, & Martin, 2006; Krekelberg, Boynton, & van Wezel, 2006). We expected the VWFA to show decreased fMRI responses to whole-word repetitions (Condition 1) relative to both whole-word changes (Condition 2) and half-word changes (Conditions 3 and 4) if it represents hemifield-split words as a coherent visual word form. In contrast, based on our expectation of a more posterior OWFA, we predicted the observation of an OWFA that would show decreased fMRI responses to half-word repetitions in both the whole-word repetition condition (Condition 1) and the half-word change condition (Conditions 3 and 4) relative to whole-word changes (Condition 2), which would mean that the OWFA represents each hemifield-split portion of a word more independently of the other than the VWFA.

METHODS

Participants

We scanned 18 healthy right-handed participants with normal or corrected-to-normal vision (11 women, 7 men; aged 21–27 years); each individual participated in 2 hr of scanning. All participants were literate native English speakers who had previous experience as participants in fMRI experiments that required steady visual fixation. All participants were recruited from the University of Western Ontario (London, Ontario, Canada), and all consent forms and experimental procedures described in these forms were approved by the University of Western Ontario's research ethics board.

fMRI Data Acquisition

Imaging was conducted at the Robarts Research Institute (London, Ontario, Canada) using a 3-T Siemens Tim MAGNETOM Trio imaging system. BOLD data were collected using T2*-weighted interleaved, single segment, EPI, PAT = 2, and a 32-channel head coil (Siemens, Erlangen, Germany). Foam padding was used to reduce head motion. Functional data were aligned to high-resolution anatomical images obtained using a 3-D T1 MPRAGE sequence (echo time [TE] = 2.98 msec; repetition time [TR] = 2300 msec; inversion time = 900 msec; flip angle = 9°; 192 contiguous 1 mm slices; field of view = 240 × 256 mm2). Each functional volume included 33 contiguous slices. Scanning parameters for obtaining functional data with full coverage of OT: TE = 30; TR = 2 sec (single shot); flip angle = 90°; field of view = 148 × 148 mm2; 2 × 2 × 2 mm3 voxel size. Each run of the main experiment included 204 volumes. Face area (OFA) localizer runs included 36 slices for 192 volumes (TE = 29, TR = 2 sec). It should be noted that the coverage of traditional nonvisual language areas anterior to the VWFA (e.g., inferior frontal gyrus) and also more anterior face processing areas was not complete in all participants. Interpretation of our statistics should therefore not be used to make claims regarding the degree of involvement of the visual system in our experiment as compared with nonvisual language areas.

Data Analysis

Data were preprocessed and analyzed using BrainVoyager QX 2.1 (BVQX; Brain Innovation, Maastricht, The Netherlands). We performed corrections for slice scan time, head motion (always <2 mm), and low-frequency artifactual drift (linear trend removal and high pass filter of 3 cycles/run); each functional volume for a given participant was aligned to the functional volume collected closest in time to the anatomical volume. Functional data were superimposed on anatomical brain images, aligned on the AC–PC line, and transformed into Talairach (Talairach & Tournoux, 1988) space and coregistered with the anatomical image for each participant. Talairach transformation was performed using standard BVQX procedures (Goebel, 1996). The hemispheres were segmented at the gray/white matter boundary, and the resultant cortical sheet was then reconstructed, inflated, and flattened for functional data analyses and visualization.

For all group analyses, functional data were spatially smoothed using a Gaussian kernel of 8 mm (FWHM). Group data were then analyzed using a random effects general linear model. For our main experiment, predictors were generated using rectangular wave functions (with a value of 1 for l volume = 2 sec when the action was initiated at the onset of the intertrial interval and a value of 0 for the remainder of the trial) that were convolved with a hemodynamic response function (Boynton, Engel, Glover, & Heeger, 1996). The data were processed using a percentage signal change transformation.

Main Experiment

Figure 1A shows the four conditions and sequence of stimuli within a block. Observers fixated on a small (∼0.05°) fixation dot in the center of the projection screen (seen via mirror at viewing distance of 15 cm) while viewing common four-letter English words presented sequentially at a rate of 1 Hz. Participants were asked to read silently each presentation of a word.

Figure 1. 

The four conditions used in the main experiment (Different, Same, Left Δ, Right Δ) are shown in (A). Sample within-block stimulus sequences are also shown for the Different (B), Left Δ (C) and Right Δ (D) conditions.

Figure 1. 

The four conditions used in the main experiment (Different, Same, Left Δ, Right Δ) are shown in (A). Sample within-block stimulus sequences are also shown for the Different (B), Left Δ (C) and Right Δ (D) conditions.

All word stimuli (∼5° × 1.5°) were composed of four capital letters (all in Arial font). We used four-letter words, which are optimally “split” between hemifields near the center of the word during normal reading (Shillcock et al., 2000), whereas longer words are typically fixated more to the leftmost portion of the word (Rayner, 1998). Additionally, four-letter words elicit greater neural activity than words with fewer letters and equivalent neural activity to words with a greater number of letters (Cornelissen, Tarkiainen, Helenius, & Salmelin, 2003; Tarkiainen, Helenius, Hansen, Cornelissen, & Salmelin, 1999). All word stimuli were positioned such that half of each word was located on either side of the fixation point.

As in other fMRI studies of word viewing (Harris, Rice, Young, & Andrews, 2015; Chee, Venkatraman, Westphal, & Siong, 2003), we used a block design (12 sec per block with 32 blocks in each run; eight blocks for each of the four conditions). Within each block, successive words were either all different (Different), all the same (Same), differed in the left two letters only (Left Δ), or differed in the right two letters only (Right Δ). An illustration of a sample intrablock word sequence (for Right Δ) is shown in Figure 1B. Words were displayed for 500 msec followed by 500-msec blank screen; this sequence repeated until the end of each 12-sec block. We counterbalanced for block order across four different runs. Each participant completed at least four runs.

OFA Localizer

We identified the OFA in eight of our participants using a conventional face selectivity localizer experiment. This experiment was performed independently (in a separate run) from the main experiment. Participants viewed 2-D grayscale (∼5° × 5°) photographs of faces, places, and common everyday objects, which alternated (in separate 15-sec blocks) with scrambled versions of the same images. At least one localizer scan was performed for each participant, with 19 blocks per scan. Fifteen images were presented in each block at 1-sec intervals.

RESULTS

Reduced fMRI Responses to Repeated Words

We performed whole-brain (coverage was complete for OT and occipital cortex) group level random effects analyses using data from all 18 participants. First, we identified a VWFA and a more posterior OWFA in the left hemisphere by comparing fMRI responses during the Different condition to those during the Same condition. The results of this contrast are shown as surface maps in Figure 2A (yellow/orange). Each of the left and right hemisphere pairs are reconstructions based on all participants using a cortex-based alignment procedure (Goebel, Esposito, & Formisano, 2006). The surface maps on the smaller hemispheres in the top of Figure 2 were obtained using false discovery rate (FDR; Genovese, Lazar, & Nichols, 2002) to control for the multiple corrections problem (Different > Same, q(FDR) < .05). The larger surface maps show results based on a more restrictive threshold (p < .01, Bonferroni-corrected).

Figure 2. 

Results of three analyses showing greater fMRI activation during the Different condition as compared with whole-word and half-word repetitions (Same, Left Δ and Right Δ). The yellow/orange surface maps displayed on the larger cortical hemispheres in (A) show Different > Same at a more liberal threshold (p < .01, Bonferroni-corrected) than those displayed on the smaller hemisphere pairs (p < .01, Bonferroni-corrected). This Different > Same effect was greatest in the VWFA and the OWFA in the left hemisphere. The blue surface maps in (B) and (C) show that half-word repetitions (Left Δ and Right Δ) also resulted in decreased fMRI responses relative the Different condition (both q(FDR) < .05), but to a lesser degree than Different > Same. The yellow outlines in (B) and (C) correspond to the surface maps in (A).

Figure 2. 

Results of three analyses showing greater fMRI activation during the Different condition as compared with whole-word and half-word repetitions (Same, Left Δ and Right Δ). The yellow/orange surface maps displayed on the larger cortical hemispheres in (A) show Different > Same at a more liberal threshold (p < .01, Bonferroni-corrected) than those displayed on the smaller hemisphere pairs (p < .01, Bonferroni-corrected). This Different > Same effect was greatest in the VWFA and the OWFA in the left hemisphere. The blue surface maps in (B) and (C) show that half-word repetitions (Left Δ and Right Δ) also resulted in decreased fMRI responses relative the Different condition (both q(FDR) < .05), but to a lesser degree than Different > Same. The yellow outlines in (B) and (C) correspond to the surface maps in (A).

In both hemisphere pairs in Figure 2A, maximal Different > Same differences were observed in OT, and a greater number of statistically significant (yellow) voxels were found in the left hemisphere than in the right hemisphere. No negative (Same > Different) voxels were observed, even at a liberal statistical threshold (p < .05, uncorrected).The most significant clusters (≥25 voxels) were observed in a Talairach location (x = −43, y = −53, z = −12) consistent with the location of the VWFA in other fMRI studies (Cohen et al., 2000, 2002). Additional highly significant voxels were observed in occipital cortex. The Talairach location (x = −38, y = −80, z = −10) of these voxels was symmetric to the known location of the OFA in the right hemisphere (Pitcher et al., 2011); we will refer to this as the OWFA.

Next, we determined the extent to which the Different > Same results could be accounted for by decreased fMRI responses to half-word repetitions during the Right Δ and Left Δ conditions. That is, if there are neurons that represent visual information corresponding to the contralateral half of a word—for instance, right hemisphere neurons that represent information in the left (contralateral) half of a word—then these neurons may show reduced fMRI responses to whole-word repetitions simply because the contralateral half of the word is always repeating. If this is the case, these brain areas should also show reduced fMRI responses during the half-word repetition conditions in which the other half of the word changed (Right Δ and Left Δ).

The blue surface maps in Figure 2B (we use blue instead of yellow/orange for contrasts in which Different is compared with the two half-change conditions) show the results of a Different > Left Δ contrast using all significant voxels from the previous Different > Same contrast at the q(FDR) < .05 threshold (i.e., voxels shown in the smaller surface maps in Figure 2A). All of the resulting significant (q(FDR) < .05) voxels were limited to left occipital cortex (Talairach: x = −35, y = −82, z = −8) and overlapped with the OWFA in Figure 2A. Therefore, at least some of the reduction of fMRI signal in left occipital cortex could have been due to the sensitivity of the OWFA to right visual field (RVF) (contralateral) half-word repetition. In contrast, no significant voxels were observed in the VWFA, even when we used an extremely liberal threshold (p < .05, uncorrected); no significant voxels were observed in medial visual cortex. The VWFA and the OWFA thus differ in terms of sensitivity to half-word repetitions and changes.

Voxels showing Different > Right Δ (q(FDR) < .05) were observed in bilateral occipital cortex (Figure 2C). The strongest effects were observed in right occipital cortex at a location (Talairach: x = 39, y = −80, z = −10) consistent with that of the OFA (Pitcher et al., 2011). At more liberal thresholds (p < .01, uncorrected), significant voxels were also observed in the left hemisphere and overlapped with the OWFA (x = −38, y = −82, z = −10), but not in the VWFA. Additionally, the OWFA—but not the VWFA—showed reduced fMRI responses to half-word repetitions (Right Δ and Left Δ) as compared with whole-word changes (Different), as shown in Figure 2B and C. This is interesting because it suggests that the VWFA encodes words as whole units, as reported previously (Glezer et al., 2009), whereas the OWFA represents information in both hemifield-specific portions of a word, but not as an integrated unit.

Additional Hemispheric Asymmetries

The previous analyses (Figure 2) identified voxels that showed decreases in fMRI responses to hemifield-specific repetition during half-word changes (Right Δ and Left Δ) as compared with whole-word changes (Different). Additional analyses were necessary to determine whether or not the Right Δ and Left Δ conditions differed from the Same condition. That is, by comparing the half-word change conditions (Right Δ and Left Δ) to the whole-word repeat (Same) condition, we were able to determine whether or not the VWFA and the OWFA both showed a similar “release” from reduced fMRI responses to whole-word repetition. Taken together with the results of the previous analyses (Figure 2), this would allow us to conclude that the VWFA and the OWFA differ in the way they represent whole words and hemifield-specific word halves, but that they are similar in that they both represent each half of a word irrespective of hemifield. In other words, the OWFA could represent each hemifield-specific half of a word split at fixation independent of the other half of the word and thus independent of the identity of the whole word, whereas the VWFA could represent the whole word, both halves bound into a meaningful whole.

Figure 3A shows the results of a Left Δ > Same comparison at q(FDR) < .05. The results of this comparison were similar to those of the Different > Same comparison. As in Figure 2A, a greater number of significant voxels were observed in left OT as compared with right OT. Thus, left visual field (LVF) half-word changes during the Left Δ condition produced elevated fMRI responses in both hemispheres but nevertheless resulted in left-lateralized VWFA activation.

Figure 3. 

Both the VWFA and the OWFA showed greater fMRI responses (q(FDR) < .05) during the half-word change conditions (Left Δ and Right Δ) as compared with the Same condition. In (A), the VWFA, OWFA, and right occipital cortex showed Left Δ > Same. Unlike the VWFA and the OWFA, however, right occipital cortex did not show Right Δ > Same (B), even at the same threshold used in (A).

Figure 3. 

Both the VWFA and the OWFA showed greater fMRI responses (q(FDR) < .05) during the half-word change conditions (Left Δ and Right Δ) as compared with the Same condition. In (A), the VWFA, OWFA, and right occipital cortex showed Left Δ > Same. Unlike the VWFA and the OWFA, however, right occipital cortex did not show Right Δ > Same (B), even at the same threshold used in (A).

Figure 3B shows the results of a Right Δ > Same contrast, again at q(FDR) < .05. The results of this comparison were considerably more left-lateralized than those of the previous comparison (Figure 3A)—no significant voxels were observed in right OT. This means that fMRI responses in the right hemisphere, but not the left, could be fully explained by LVF half-word changes. In summary, only the VWFA and the OWFA showed similarly elevated fMRI responses to all half-word and whole-word changes relative to whole-word repetitions. Thus, the left-lateralized fMRI responses in our experiment show that left OT (the VWFA and the OWFA in particular) differs from OT in terms of sensitivity to both whole-word and half-word changes.

Face Selectivity and the OWFA

To test our hypothesis that the OWFA would be symmetric to the OFA in right occipital cortex (Figure 4, top right) in terms of location, we identified the OFA using a face selectivity localizer experiment using an random effects analysis of Faces > Objects, Places, Scrambled. The smaller of the two right hemispheres in Figure 4 (top right) shows the results of this contrast, which revealed maximally significant (p < 10−3) voxels in posterior right occipital cortex (Talairach: x = 35, y = −80, z = −15) and also a more anterior right FFA (Kanwisher et al., 1997). No statistically significant voxels were observed in the left hemisphere at this threshold in either location. The location of the OFA and the absence of statistically significant voxels in the left hemisphere are consistent with the results of previous fMRI studies (Pitcher et al., 2011).

Figure 4. 

Relating word selectivity to face selectivity. The yellow outlines (left hemisphere) correspond to the surface maps in Figure 2A. The OWFA, but not the VWFA, overlapped with the greatest reduction in fMRI response during the half-word versus whole-word changes (Different > Right Δ and Different > Left Δ; only a subset of blue voxels from Figure 2B and C are shown). The red outline (right hemisphere) indicates the OFA, as identified using a Faces > Objects, Places, Scrambled contrast (both the OFA and the FFA are shown in the smaller right hemisphere). The OWFA and the OFA were anatomically symmetric and showed similar trends of Faces > Objects > Place > Scrambled (graphs), but only the OFA showed significantly greater fMRI responses to faces as compared with objects and places.

Figure 4. 

Relating word selectivity to face selectivity. The yellow outlines (left hemisphere) correspond to the surface maps in Figure 2A. The OWFA, but not the VWFA, overlapped with the greatest reduction in fMRI response during the half-word versus whole-word changes (Different > Right Δ and Different > Left Δ; only a subset of blue voxels from Figure 2B and C are shown). The red outline (right hemisphere) indicates the OFA, as identified using a Faces > Objects, Places, Scrambled contrast (both the OFA and the FFA are shown in the smaller right hemisphere). The OWFA and the OFA were anatomically symmetric and showed similar trends of Faces > Objects > Place > Scrambled (graphs), but only the OFA showed significantly greater fMRI responses to faces as compared with objects and places.

The graphs at the bottom of Figure 4 show clear face selectivity in the OFA, with greater fMRI responses to faces than all other stimulus types used in the localizer experiment (always p < .01, two-tailed t tests), and a trend for face selectivity in the OWFA. To achieve independence between the results in the graphs and the ROIs used to generate these graphs, we used ∼1000 of the most significant voxels from the results of the Different > Right Δ and Different > Left Δ analysis reported earlier (Figure 2B and C) as ROIs (shown in blue, in each hemisphere in Figure 4). The anatomical location of each cluster of (blue) voxels was highly symmetric, which means that voxels in the OWFA that showed Different > Right Δ and Different > Left Δ were symmetric in location to those of the OFA in the right hemisphere, which apparently represents only the contralateral half of hemifield-split words.

DISCUSSION

We used hemifield-split words to identify a VWFA in human OT, which we delineated from a more posterior OWFA. We found that these two areas in the left cortex represent both the left and right halves of a word split between the two hemifields. Both the VWFA and the OWFA (Figure 2A) showed a greater reduction of fMRI response to repeated words than any other brain areas included in our scans, and both areas showed a release from this reduced fMRI response when a word was changed in full or in part. The two areas differed, however, in that the VWFA showed equivalent fMRI responses to whole-word and half-word changes, whereas the OWFA showed greater fMRI responses to whole-word changes than half-word changes. This suggests that the OWFA represents hemifield-split word halves more independently than the VWFA. The cortical location of the OWFA was mirror-symmetric to that of the OFA, which has been shown to represent face parts more independently than other face-selective areas (Pitcher et al., 2011; Schiltz & Rossion, 2006; Haxby, Hoffman, & Gobbini, 2000). Our findings are in agreement with other findings of word-specific visual processing in the human brain (Dehaene & Cohen, 2011; Glezer et al., 2009; Dehaene, Cohen, Sigman, & Vinckier, 2005; McCandliss et al., 2003), the topic of considerable disagreement (Vogel et al., 2012; Price & Devlin, 2003, 2004, 2011; Seghier & Price, 2011).

Word “Parts” and “Wholes”

We used hemifield-split words that either changed in full, repeated in full, or repeated in part (i.e., two letters changed in one hemifield but not the other, thus changing the identity of the word but preserving the repetition of two hemifield-specific letters). This allowed us to identify brain areas that showed decreased fMRI responses to whole-word and half-word repetition. We found that the VWFA and the OWFA (Figure 2A) showed a greater reduction of fMRI signal as a result of whole-word repetition than any other part of the brain. Both areas showed a “release” from this reduced fMRI response when either half of a word was changed. This means that the VWFA and the OWFA represent both halves of a hemifield-split word. This result substantiates and extends that of a previous fMRI study VWFA (Glezer et al., 2009), which reported whole-word representation in the VWFA but did not explore the possibility that other brain areas—such as the OWFA in our study—represent both halves of a hemifield-split word.

Our results are important to understanding the convergence of the visual field split as it applies to reading. The neural basis of this convergence is not known, but influential accounts of rapid word recognition posit that interhemifield combination during normal reading occurs in the VWFA (McCandliss et al., 2003). VWFA damage is associated with letter-by-letter reading (Cohen et al., 2004), which suggests a critical role of the VWFA in binding individual letters into whole words. Therefore, by changing two “parts” (halves) of a word simultaneously, one would observe a greater difference from a whole-word representation than by changing only one half of the word, regardless of whether the representation under consideration is sublexical or lexical, for any cortical area sensitive to the amount of change in a visual stimulus from one presentation to another, even if the area is not purely selective for word stimuli. In terms of reduced fMRI responses to whole-word versus half-word repetitions, one would thus expect a greater “release” from reduced fMRI responses to repeated whole words with whole-word changes as compared with half-word changes—which we indeed observed in the OWFA, but not the VWFA (Figure 2). Our results therefore suggest that the OWFA represents individual letters corresponding to a whole word without binding these letters into coherent whole-word forms. This possibility is consistent with theories that propose a hierarchical organization of low-to-high orthographic selectivity that progresses anteriorly along the ventral visual processing stream (Glezer et al., 2015; Thesen et al., 2012; Vinckier et al., 2007), although our current method did not allow us to distinguish the VWFA and the OWFA in terms of lexical versus sublexical processing.

Our observation that the OWFA represents both halves of a hemifield-split word has implications for anatomical models of hemifield-independent word representation in visual cortex, which assert that the corpus callosum transmits (LVF) word form information from the right visual cortex to the VWFA in the left hemisphere (Molko et al., 2002). Like the VWFA, the OWFA may also receive LVF word form information via the corpus callosum. This would be consistent with claims that the convergence of RVF and LVF word form information in left visual cortex occurs in extrastriate areas beyond V4 (Hsiao, Shieh, & Cottrell, 2008). Our results also support the idea that visual word forms are represented as letters and bigrams before they are combined into whole-word representations (Vinckier et al., 2007) and that the former may occur in the OWFA and possibly other portions of OT (Thesen et al., 2012). In short, our findings highlight the possibility that a complete visual representation of hemifield-split visual word form information exists in the OWFA, prior to the orthographic representation of real whole words in the VWFA.

From Faces to Words in Visual Cortex

A widely recognized feature of the human brain is that functionally defined category-selective visual cortical areas come in pairs (Schwarzlose et al., 2008). This has been reported for many different categories of visual stimuli (faces, bodies, houses, objects, tools, etc.). For each pair of category-selective cortical areas, one of these regions tends to be found in occipital cortex, and the other in more anterior ventral (fusiform or OT) cortex. A possible explanation for this pairing is that more anterior areas exhibit a greater completeness of representation. For instance, faces and bodies appear to be represented more “holistically” in the more anterior of the two areas (Pitcher et al., 2011; Taylor, Wiggett, & Downing, 2007; Schiltz & Rossion, 2006). The generalization of this view to word form representation then predicts our identification of a part-based word OWFA, which can be differentiated from the VWFA in terms of holistic word representation, although there is intense ongoing debate as to whether or not category selectivity for words exists in the human visual system (Vogel et al., 2014; Dehaene & Cohen, 2011; Seghier & Price, 2011). The anatomical locations of the VWFA and the OWFA in our experiment are strikingly symmetric anatomically to the FFA and OFA (respectively) in the right hemisphere. Although it is possible that more anterior word and face areas show similar cortical symmetry of location, our fMRI coverage of traditional language regions (e.g., inferior frontal gyrus) and face processing in regions of the brain anterior to those of interest here were incomplete in some participants.

If it is the case that the VWFA is a product of “neuronal recycling” (Dehaene & Cohen, 2007), by which evolutionarily older neural mechanisms are modified to support recent cultural inventions such as reading, then it is possible that the OWFA emerges for the same reason. That is, faces and words may compete for representation during development, and this competition may occur in both anterior and posterior category-selective visual cortex. Moreover, it is possible that right-lateralization for the visual representation of faces in fusiform cortex (FFA) is driven by reading acquisition and the developmental emergence of the VWFA (Dundas, Plaut, & Behrmann, 2013; Dehaene et al., 2010). If so, the same account would explain the right lateralization of the OFA as the result of a developmentally emergent OWFA. This would be consistent with the role of each area in part-based representation of faces and words, respectively, and additional commonalities in the visual processing of faces and words (Behrmann & Plaut, 2013; Plaut & Behrmann, 2011).

Conclusion

The first goal of our study was to see if viewing repeated words as compared with words that changed, in part or in full, would produce left-lateralized decreases in fMRI responses and also to see whether or not this result would be limited to the VWFA. We found that, although decreased fMRI responses to repeated words were left lateralized, a second area—which we labeled the “occipital word form area” (OWFA)—shared this effect with the VWFA. On the basis of some differences in the fMRI results for each area, we also distinguished these two areas. Although the more VWFA was found to represent words as whole units, the more posterior OWFA—mirror-symmetric to face-selective occipital cortex (the OFA) in the right hemisphere—represented both halves of a word, but less holistically than the VWFA.

Although it is possible that some of our results could have been influenced by attention-related factors, it seems unlikely that visual attention accounts for the striking degree of left-lateralization and mirror-symmetric anatomical correspondence to known face areas. Another study (Harris et al., 2015), which used a method similar to ours, reported a right-lateralized reduction in fMRI response to repeated faces and left-lateralized fMRI responses to repeated names. Furthermore, it has been shown for both faces and words that attention merely changes fMRI signal strength in category-selective areas such as the FFA and the VWFA, rather than the involvement of these areas or the pattern of neural representations (Xu, Jiang, Ma, Yang, & Weng, 2012). Regardless, our results partially replicate, complement, and extend those of another fMRI study of whole-word representation (Glezer et al., 2009) and support the idea that the emergence of phylogenetically recent cognitive advances such as reading are accompanied by a competition for neural resources in which evolutionarily older mechanisms are used for new purposes (Dehaene et al., 2010; Dehaene & Cohen, 2007). The outcome of this competition and its relationship to commonalities in the visual processing demands for reading and face recognition is the mirror-symmetric lateralization of face-selective and word-selective areas in human visual cortex.

Acknowledgments

This work was supported by Canadian Institutes of Health Research grant 9335 to T. Vilis. We thank M. Behrmann, A. Nestor, and M. Joanisse for their helpful feedback during the course of our study.

Reprint requests should be sent to Lars Strother, Department of Psychology, University of Nevada, Reno, Mailstop 296, 1664 N. Virginia St., Reno, NV 89557, or via e-mail: lars@unr.edu.

REFERENCES

Behrmann
,
M.
, &
Plaut
,
D. C.
(
2013
).
Distributed circuits, not circumscribed centers, mediate visual recognition
.
Trends in Cognitive Sciences
,
17
,
210
219
.
Berlucchi
,
G.
(
2014
).
Visual interhemispheric communication and callosal connections of the occipital lobes
.
Cortex
,
56
,
1
13
.
Bouhali
,
F.
,
Thiebaut de Schotten
,
M.
,
Pinel
,
P.
,
Poupon
,
C.
,
Mangin
,
J. F.
,
Dehaene
,
S.
, et al
(
2014
).
Anatomical connections of the visual word form area
.
Journal of Neuroscience
,
34
,
15402
15414
.
Boynton
,
G. M.
,
Engel
,
S. A.
,
Glover
,
G. H.
, &
Heeger
,
D. J.
(
1996
).
Linear systems analysis of functional magnetic resonance imaging in human V1
.
Journal of Neuroscience
,
16
,
4207
4221
.
Brysbaert
,
M.
(
2004
).
The importance of interhemispheric transfer for foveal vision: A factor that has been overlooked in theories of visual word recognition and object perception
.
Brain and Language
,
88
,
259
267
.
Chee
,
M. W.
,
Venkatraman
,
V.
,
Westphal
,
C.
, &
Siong
,
S. C.
(
2003
).
Comparison of block and event-related fMRI designs in evaluating the word-frequency effect
.
Human Brain Mapping
,
18
,
186
193
.
Cohen
,
L.
, &
Dehaene
,
S.
(
2004
).
Specialization within the ventral stream: The case for the visual word form area
.
Neuroimage
,
22
,
466
476
.
Cohen
,
L.
,
Dehaene
,
S.
,
Naccache
,
L.
,
Lehericy
,
S.
,
Dehaene-Lambertz
,
G.
,
Henaff
,
M. A.
, et al
(
2000
).
The visual word form area: Spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients
.
Brain
,
123
,
291
307
.
Cohen
,
L.
,
Henry
,
C.
,
Dehaene
,
S.
,
Martinaud
,
O.
,
Lehericy
,
S.
,
Lemer
,
C.
, et al
(
2004
).
The pathophysiology of letter-by-letter reading
.
Neuropsychologia
,
42
,
1768
1780
.
Cohen
,
L.
,
Lehericy
,
S.
,
Chochon
,
F.
,
Lemer
,
C.
,
Rivaud
,
S.
, &
Dehaene
,
S.
(
2002
).
Language-specific tuning of visual cortex functional properties of the visual word form area
.
Brain
,
125
,
1054
1069
.
Cornelissen
,
P.
,
Tarkiainen
,
A.
,
Helenius
,
P.
, &
Salmelin
,
R.
(
2003
).
Cortical effects of shifting letter position in letter strings of varying length
.
Journal of Cognitive Neuroscience
,
15
,
731
746
.
Dehaene
,
S.
, &
Cohen
,
L.
(
2007
).
Cultural recycling of cortical maps
.
Neuron
,
56
,
384
398
.
Dehaene
,
S.
, &
Cohen
,
L.
(
2011
).
The unique role of the visual word form area in reading
.
Trends in Cognitive Sciences
,
15
,
254
262
.
Dehaene
,
S.
,
Cohen
,
L.
,
Sigman
,
M.
, &
Vinckier
,
F.
(
2005
).
The neural code for written words: A proposal
.
Trends in Cognitive Sciences
,
9
,
335
341
.
Dehaene
,
S.
,
Jobert
,
A.
,
Naccache
,
L.
,
Ciuciu
,
P.
,
Poline
,
J. B.
,
Le Bihan
,
D.
, et al
(
2004
).
Letter binding and invariant recognition of masked words: Behavioral and neuroimaging evidence
.
Psychological Science
,
15
,
307
313
.
Dehaene
,
S.
,
Pegado
,
F.
,
Braga
,
L. W.
,
Ventura
,
P.
,
Nunes Filho
,
G.
,
Jobert
,
A.
, et al
(
2010
).
How learning to read changes the cortical networks for vision and language
.
Science
,
330
,
1359
1364
.
Dundas
,
E. M.
,
Plaut
,
D. C.
, &
Behrmann
,
M.
(
2013
).
The joint development of hemispheric lateralization for words and faces
.
Journal of Experimental Psychology: General
,
142
,
348
358
.
Ellis
,
A. W.
, &
Brysbaert
,
M.
(
2010
).
Split fovea theory and the role of the two cerebral hemispheres in reading: A review of the evidence
.
Neuropsychologia
,
48
,
353
365
.
Gauthier
,
I.
,
Tarr
,
M. J.
,
Moylan
,
J.
,
Skudlarski
,
P.
,
Gore
,
J. C.
, &
Anderson
,
A. W.
(
2000
).
The fusiform “face area” is part of a network that processes faces at the individual level
.
Journal of Cognitive Neuroscience
,
12
,
495
504
.
Genovese
,
C. R.
,
Lazar
,
N. A.
, &
Nichols
,
T.
(
2002
).
Thresholding of statistical maps in functional neuroimaging using the false discovery rate
.
Neuroimage
,
15
,
870
878
.
Glezer
,
L. S.
,
Jiang
,
X.
, &
Riesenhuber
,
M.
(
2009
).
Evidence for highly selective neuronal tuning to whole words in the “visual word form area.”
Neuron
,
62
,
199
204
.
Glezer
,
L. S.
,
Kim
,
J.
,
Rule
,
J.
,
Jiang
,
X.
, &
Riesenhuber
,
M.
(
2015
).
Adding words to the brain's visual dictionary: Novel word learning selectively sharpens orthographic representations in the VWFA
.
Journal of Neuroscience
,
35
,
4965
4972
.
Goebel
,
R.
(
1996
).
Brainvoyager: A program for analyzing and visualizing functional and structural magnetic resonance data sets
.
Neuroimage
,
3
,
S604
.
Goebel
,
R.
,
Esposito
,
F.
, &
Formisano
,
E.
(
2006
).
Analysis of functional image analysis contest (FIAC) data with brainvoyager QX: From single-subject to cortically aligned group general linear model analysis and self-organizing group independent component analysis
.
Human Brain Mapping
,
27
,
392
401
.
Grill-Spector
,
K.
,
Henson
,
R.
, &
Martin
,
A.
(
2006
).
Repetition and the brain: Neural models of stimulus-specific effects
.
Trends in Cognitive Sciences
,
10
,
14
23
.
Harris
,
R. J.
,
Rice
,
G. E.
,
Young
,
A. W.
, &
Andrews
,
T. J.
(
2015
).
Distinct but overlapping patterns of response to words and faces in the fusiform gyrus
.
Cerebral Cortex
.
doi: 10.1093/cercor/bhv147
.
Haxby
,
J. V.
,
Hoffman
,
E. A.
, &
Gobbini
,
M. I.
(
2000
).
The distributed human neural system for face perception
.
Trends in Cognitive Sciences
,
4
,
223
233
.
Hsiao
,
J. H.
,
Shieh
,
D. X.
, &
Cottrell
,
G. W.
(
2008
).
Convergence of the visual field split: Hemispheric modeling of face and object recognition
.
Journal of Cognitive Neuroscience
,
20
,
2298
2307
.
Kanwisher
,
N.
,
McDermott
,
J.
, &
Chun
,
M. M.
(
1997
).
The fusiform face area: A module in human extrastriate cortex specialized for face perception
.
Journal of Neuroscience
,
17
,
4302
4311
.
Krekelberg
,
B.
,
Boynton
,
G. M.
, &
van Wezel
,
R. J.
(
2006
).
Adaptation: From single cells to BOLD signals
.
Trends in Neurosciences
,
29
,
250
256
.
Lavidor
,
M.
, &
Walsh
,
V.
(
2004
).
Opinion—The nature of foveal representation
.
Nature Reviews Neuroscience
,
5
,
729
735
.
Liu
,
J.
,
Harris
,
A.
, &
Kanwisher
,
N.
(
2010
).
Perception of face parts and face configurations: An fMRI study
.
Journal of Cognitive Neuroscience
,
22
,
203
211
.
Malach
,
R.
(
2012
).
Targeting the functional properties of cortical neurons using fMR-adaptation
.
Neuroimage
,
62
,
1163
1169
.
McCandliss
,
B. D.
,
Cohen
,
L.
, &
Dehaene
,
S.
(
2003
).
The visual word form area: Expertise for reading in the fusiform gyrus
.
Trends in Cognitive Sciences
,
7
,
293
299
.
Molko
,
N.
,
Cohen
,
L.
,
Mangin
,
J. F.
,
Chochon
,
F.
,
Lehericy
,
S.
,
Le Bihan
,
D.
, et al
(
2002
).
Visualizing the neural bases of a disconnection syndrome with diffusion tensor imaging
.
Journal of Cognitive Neuroscience
,
14
,
629
636
.
Muayqil
,
T.
,
Davies-Thompson
,
J.
, &
Barton
,
J. J.
(
2015
).
Representation of visual symbols in the visual word processing network
.
Neuropsychologia
,
69
,
232
241
.
Pitcher
,
D.
,
Walsh
,
V.
, &
Duchaine
,
B.
(
2011
).
The role of the occipital face area in the cortical face perception network
.
Experimental Brain Research
,
209
,
481
493
.
Pitcher
,
D.
,
Walsh
,
V.
,
Yovel
,
G.
, &
Duchaine
,
B.
(
2007
).
TMS evidence for the involvement of the right occipital face area in early face processing
.
Current Biology
,
17
,
1568
1573
.
Plaut
,
D. C.
, &
Behrmann
,
M.
(
2011
).
Complementary neural representations for faces and words: A computational exploration
.
Cognitive Neuropsychology
,
28
,
251
275
.
Price
,
C. J.
, &
Devlin
,
J. T.
(
2003
).
The myth of the visual word form area
.
Neuroimage
,
19
,
473
481
.
Price
,
C. J.
, &
Devlin
,
J. T.
(
2004
).
The pro and cons of labelling a left occipitotemporal region: “The visual word form area.”
Neuroimage
,
22
,
477
479
.
Price
,
C. J.
, &
Devlin
,
J. T.
(
2011
).
The interactive account of ventral occipitotemporal contributions to reading
.
Trends in Cognitive Sciences
,
15
,
246
253
.
Rayner
,
K.
(
1998
).
Eye movements in reading and information processing: 20 Years of research
.
Psychological Bulletin
,
124
,
372
422
.
Schiltz
,
C.
, &
Rossion
,
B.
(
2006
).
Faces are represented holistically in the human occipito-temporal cortex
.
Neuroimage
,
32
,
1385
1394
.
Schwarzlose
,
R. F.
,
Swisher
,
J. D.
,
Dang
,
S.
, &
Kanwisher
,
N.
(
2008
).
The distribution of category and location information across object-selective regions in human visual cortex
.
Proceedings of the National Academy of Sciences, U.S.A.
,
105
,
4447
4452
.
Seghier
,
M. L.
, &
Price
,
C. J.
(
2011
).
Explaining left lateralization for words in the ventral occipitotemporal cortex
.
Journal of Neuroscience
,
31
,
14745
14753
.
Shillcock
,
R.
,
Ellison
,
T. M.
, &
Monaghan
,
P.
(
2000
).
Eye-fixation behavior, lexical storage, and visual word recognition in a split processing model
.
Psychological Review
,
107
,
824
851
.
Shin
,
Y. W.
,
Kwon
,
J. S.
,
Kwon
,
K. W.
,
Gu
,
B. M.
,
Song
,
I. C.
,
Na
,
D. G.
, et al
(
2008
).
Objects and their icons in the brain: The neural correlates of visual concept formation
.
Neuroscience Letters
,
436
,
300
304
.
Talairach
,
J.
, &
Tournoux
,
P.
(
1988
).
Co-planar stereotaxic atlas of the human brain: 3-Dimensional proportional system: An approach to cerebral imaging
.
Stuttgart
:
Thieme u.a.
Tarkiainen
,
A.
,
Helenius
,
P.
,
Hansen
,
P. C.
,
Cornelissen
,
P. L.
, &
Salmelin
,
R.
(
1999
).
Dynamics of letter string perception in the human occipitotemporal cortex
.
Brain
,
122
,
2119
2132
.
Taylor
,
J. C.
, &
Downing
,
P. E.
(
2011
).
Division of labor between lateral and ventral extrastriate representations of faces, bodies, and objects
.
Journal of Cognitive Neuroscience
,
23
,
4122
4137
.
Taylor
,
J. C.
,
Wiggett
,
A. J.
, &
Downing
,
P. E.
(
2007
).
Functional MRI analysis of body and body part representations in the extrastriate and fusiform body areas
.
Journal of Neurophysiology
,
98
,
1626
1633
.
Thesen
,
T.
,
McDonald
,
C. R.
,
Carlson
,
C.
,
Doyle
,
W.
,
Cash
,
S.
,
Sherfey
,
J.
, et al
(
2012
).
Sequential then interactive processing of letters and words in the left fusiform gyrus
.
Nature Communications
,
3
,
1284
.
Ventura
,
P.
(
2014
).
Let's face it: Reading acquisition, face and word processing
.
Frontiers in Psychology
,
5
,
787
.
Vinckier
,
F.
,
Dehaene
,
S.
,
Jobert
,
A.
,
Dubus
,
J. P.
,
Sigman
,
M.
, &
Cohen
,
L.
(
2007
).
Hierarchical coding of letter strings in the ventral stream: Dissecting the inner organization of the visual word-form system
.
Neuron
,
55
,
143
156
.
Vogel
,
A. C.
,
Petersen
,
S. E.
, &
Schlaggar
,
B. L.
(
2012
).
The left occipitotemporal cortex does not show preferential activity for words
.
Cerebral Cortex
,
22
,
2715
2732
.
Vogel
,
A. C.
,
Petersen
,
S. E.
, &
Schlaggar
,
B. L.
(
2014
).
The VWFA: It's not just for words anymore
.
Frontiers in Human Neuroscience
,
8
,
88
.
Xu
,
G.
,
Jiang
,
Y.
,
Ma
,
L.
,
Yang
,
Z.
, &
Weng
,
X.
(
2012
).
Similar spatial patterns of neural coding of category selectivity in FFA and VWFA under different attention conditions
.
Neuropsychologia
,
50
,
862
868
.