Abstract

Reading relies on the rapid visual recognition of words viewed in a wide variety of fonts. We used fMRI to identify neural populations showing reduced fMRI responses to repeated words displayed in different fonts (“font-invariant” repetition suppression). We also identified neural populations showing greater fMRI responses to words repeated in a changing font as compared with words repeated in the same font (“font-sensitive” release from repetition suppression). We observed font-invariant repetition suppression in two anatomically distinct regions of the left occipitotemporal cortex (OT), a “visual word form area” in mid-fusiform cortex, and a more posterior region in the middle occipital gyrus. In contrast, bilateral shape-selective lateral occipital cortex and posterior fusiform showed considerable sensitivity to font changes during the viewing of repeated words. Although the visual word form area and the left middle occipital gyrus showed some evidence of font sensitivity, both regions showed a relatively greater degree of font invariance than font sensitivity. Our results show that the neural mechanisms in the left OT involved in font-invariant word recognition are anatomically distinct from those sensitive to font-related shape changes. We conclude that font-invariant representation of visual word form is instantiated at multiple levels by anatomically distinct neural mechanisms within the left OT.

INTRODUCTION

Our ability to recognize letters and words displayed in different fonts (e.g., , , , ) and other variable text styles or formats is an essential component of skilled reading. “Viewpoint-invariant” recognition of words and other kinds of visual objects refers to the ability to recognize words, letters, and other objects at different retinal locations, in different retinal sizes, and under different conditions of orientation and illumination. This ability implies the existence of “abstract” representations of visual objects in the human brain, the topic of numerous empirical studies of object recognition, and its neural basis. “Font-invariant” word recognition can be thought of as a specific case of abstract representation that is unique to words, alphanumeric characters, and other linguistic visual objects. Understanding the neural basis of font-invariant word recognition is therefore part of a larger endeavor to understand the neural basis of abstract representation of visual objects more generally.

Many models of word recognition posit that the process begins with the extraction of letter-level visual features, which then activate abstract letter detectors (Sanocki, 1988). These abstract letter detectors are typically assumed to be sensitive to the retinal location of letters but invariant to font and other format differences (Grainger, Dufau, & Ziegler, 2016). Some have argued that neural mechanisms in the left occipitotemporal cortex (OT), a “visual word form area” (VWFA) in particular, support abstract representation of linguistic visual objects by virtue of adapting viewpoint-invariant object recognition mechanisms to word recognition (Dehaene & Cohen, 2011). Perhaps the most compelling empirical evidence, both in favor of and against abstract word form representation in the VWFA, comes from fMRI studies of viewpoint-invariant word recognition. An early study by Cohen et al. (2002) reported equivalent word selectivity in the VWFA for words viewed in ipsilateral and contralateral visual field location, which the authors interpreted as evidence in favor of viewpoint-invariant word form representation. In contrast, a more recent study by Rauschecker, Bowen, Parvizi, and Wandell (2012) highlighted the sensitivity of the VWFA to retinal stimulus position, which potentially, though not necessarily, contradicts the earlier report of position-invariant word selectivity. Other potentially contradictory findings have also been reported with respect to the degree to which the VWFA represents a word irrespective of the format (e.g., letter case, font) in which it is viewed. Results of an early fMRI study by Dehaene et al. (2004) supported the view that binding of letters into whole words is achieved by increasingly invariant processing stages in the left OT, culminating in abstract word representation in the VWFA (Dehaene, Cohen, Sigman, & Vinckier, 2005). Results of subsequent fMRI studies directly challenge this view (Lochy et al., 2018; Wimmer, Ludersdorfer, Richlan, & Kronbichler, 2016; Burgund, Guo, & Aurbach, 2009), again highlighting the ongoing controversy surrounding abstract representation of letters and words in the VWFA. Similar controversy exists for the neural basis of viewpoint-invariant object recognition more generally (e.g., Tarr & Hayward, 2017; Andresen, Vinberg, & Grill-Spector, 2009; Chen, Kao, & Tyler, 2007; Vuilleumier, Henson, Driver, & Dolan, 2002; Grill-Spector et al., 1999).

Here, we sought to test whether font-invariant representation of words and constituent letters occurs in the VWFA, and if so, whether such representations are limited to the VWFA or not. Relatively few fMRI studies of font invariance per se have contributed to the ongoing debate concerning abstract representation of letters and words in the VWFA. An early fMRI study by Gauthier et al. (2000) reported bilateral OT (posterior, lateral, and slightly dorsal to the typical location of the VWFA) sensitivity to the font in which single letters were viewed, but their results did not clearly rule out the possibility of font-invariant letter representation in the VWFA. In contrast, several subsequent fMRI studies have reported results consistent with font-invariant representation of both individual letters and whole words in the VWFA (Rothlein & Rapp, 2014; Nestor, Behrmann, & Plaut, 2013; Qiao et al., 2010; Polk et al., 2002). However, two important questions concerning font-invariant word representation in OT remain. First, does the VWFA show sensitivity or invariance to font or both, and can the underlying neural populations be delineated? One possibility is that the VWFA contains distinct neural populations that show either extreme invariance or extreme sensitivity to font. Alternatively, the VWFA may contain a single neural population that shows some degree of both font invariance and sensitivity, which would highlight the possibility of understanding abstract word representation in terms of degree, with some neural populations showing a greater degree of font invariance than others and vice versa. A second question is, do neural populations in occipital cortex exhibit font invariance, and if so, how does this relate to abstract letter detection within a strictly posterior-to-anterior hierarchy of increasingly complex visual word form processing in the left OT (e.g., such as that proposed by Vinckier et al., 2007)? This possibility is foreshadowed by the emergence of coexistent retinotopic and abstract shape representation in extrastriate cortex (Vernon, Gouws, Lawrence, Wade, & Morland, 2016; Sayres & Grill-Spector, 2008; Larsson & Heeger, 2006).

Here, we used an fMRI word repetition paradigm to simultaneously measure sensitivity and invariance to repeating or changing words viewed in either a changing or unchanging font. Results from two previous studies used a similar method to show word-selective repetition suppression in the VWFA and also revealed a relatively posterior “occipital word form area” (OWFA), which was clearly delineable from the VWFA in the left OT (Strother, Zhou, Coros, & Vilis, 2017; Strother, Coros, & Vilis, 2016). These studies did not, however, investigate the degree to which repetition suppression in these brain regions was font-invariant. The current study tests the hypothesis that font invariance emerges in extrastriate cortex, a possible indication of abstract letter detection by neural mechanisms in the occipital lobe.

METHODS

Participants

Twelve right-handed volunteers (eight women; mean age = 24.0 years, range = 20–35 years) participated in this experiment. All participants were healthy college students with normal or corrected-to-normal vision and were literate native English speakers. Participants were recruited from Western University (London, Ontario, Canada). All consent forms and experimental procedures were approved by Western University's research ethics board. Based on an effect size (η2 = .89) from previously published results from highly similar experiments (Strother et al., 2016, 2017), a power analysis (G*Power) indicated that 12 participants would achieve power of (1 − β) = 95%.

fMRI Data Acquisition and Analysis

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). 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 = 2.98 msec, repetition time = 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: echo time = 30, repetition time = 2 sec (single shot), flip angle = 90°, field of view = 148 × 148 mm2, 2 × 2 × 2 mm3 voxel size. Whole-brain coverage (beyond OT) was not complete in all participants. Each run of the main experiment included 252 volumes.

Data were preprocessed and analyzed using BrainVoyager QX 2.1 (BVQX; Brain Innovation) and Neuroelf (neuroelf.net/) for cluster correction. 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, 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). Functional data were spatially smoothed using a Gaussian kernel of 8 mm (FWHM). Predictors were generated using rectangular wave functions (with a value of 1 for 1 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).

Stimuli and Procedure

We used a block design fMRI stimulus repetition paradigm, with five experimental conditions. Figure 1 shows each condition and the sequence of stimuli within each corresponding block. Word stimuli subtended a visual angle of ∼5° × 1.5° (viewed via mirror at 15-cm distance) and were split in half between the left (LVF) and right (RVF) visual hemifields between a centered fixation dot (∼0.05°). Each word was displayed in uppercase letters, monospaced, using 1 of the 12 fonts (all Microsoft typefaces: Arial, Times New Roman, Juice ITC, Segoe Print, Chiller, Bookman Old Style, Bradley Hand ITC, Lucida Handwriting, Engravers MT, Harrington, MV Boli, Curlz MT). The same 192 words, used in a previous study (Strother et al., 2016), were displayed in black against a white background.

Figure 1. 

The five experimental conditions and within-block stimulus sequence. In each 12-sec block, 12 words were presented at 1 Hz with 500 msec on and 500 msec off. Words were displayed either repeated in the same font (No Δ), repeated in a changing font (Font Δ), changed in the context of a changing font (Both Δ), or changed by two letters (LVF Δ or RVF Δ) in the context of a changing font.

Figure 1. 

The five experimental conditions and within-block stimulus sequence. In each 12-sec block, 12 words were presented at 1 Hz with 500 msec on and 500 msec off. Words were displayed either repeated in the same font (No Δ), repeated in a changing font (Font Δ), changed in the context of a changing font (Both Δ), or changed by two letters (LVF Δ or RVF Δ) in the context of a changing font.

Word frequency for those in the Both Δ, LVF Δ, and RVF Δ conditions were first log-transformed because word frequency is usually not normally distributed. The average values were 4.12 ± 2.99 for the Both Δ condition, 3.75 ± 2.72 for the LVF Δ condition, and 3.57 ± 2.83 for the RVF Δ condition. Paired-sample t tests showed no statistically significant differences between these three conditions (all ps > .1). We also calculated the phonemic overlap for the words within each block for the half-change condition (LVF Δ and RVF Δ) using the following formula: No. of overlapping phonemes / No. of phonemes total in the word (i.e., a four-phoneme word that shares two phonemes with the previous word would result an overlapping score of 0.5). The phonemic overlap between the LVF Δ (0.41 ± 0.05) words and the RVF Δ (0.39 ± 0.06) words was not significantly different, t(14) = 0.32, p = .75.

In each block, 12 words were presented at a rate of 1 Hz, with words displayed cyclically for 500 msec followed by 500-msec blank screen in a 12-sec block. The five experimental conditions were No Δ, Font Δ, Both Δ, LVF Δ, and RVF Δ. For the No Δ condition, each block contained the same four-letter word (all in the same font) repeated for 12 times. For the Font Δ condition, each block contained the same four-letter word repeated for 12 times, but all 12 words were different in their fonts. For the Both Δ condition, all 12 words were different in both identities and fonts, and the same letter did not appear in the same location for two consecutive words. The LVF Δ condition was the same as the Both Δ condition, except that the same letters always repeated in the RVF and vice versa for the RVF Δ condition.

Each run was composed of 42 blocks, 8 blocks per condition, as well as 2 fixation blocks (1 block in the beginning and 1 block at the end). Block order was counterbalanced across runs. Each participant completed four to six runs. All participants had considerable experience maintaining central fixation during fMRI experiments and were instructed to keep their gazes on the fixation dot throughout the entirety of scanning.

RESULTS

Group-level Analyses

All group-level analyses employed a random-effects general linear model to compute contrasts between predictors. We first identified voxels showing font-invariant repetition suppression using a [Both Δ + LVF Δ + RVF Δ] > Font Δ contrast. As shown in Figure 2A and Table 1, the resultant contrast map containing voxels surviving a threshold of t > 4.44, q(FDR) < 0.05 were limited to the left OT. In the left OT, peak activations occurred in the middle fusiform gyrus (mFus), at x = −42, y = −43, z = −17, the posterior fusiform gyrus (pFus), at x = −39, y = −58, z = −11, the inferior occipital gyrus (IOG), at x = −36, y = −82, z = −14, and the left middle occipital gyrus (MOG), at x = −30, y = −88, z = −4.

Figure 2. 

Contrast maps obtained in group-level analyses showing left-lateralized font-invariant repetition suppression and font sensitivity in both left and right OT. The first map (A) from a [Both Δ + LVF Δ + RVF Δ] > Font Δ contrast revealed font-invariant repetition suppression in the left middle and posterior fusiform gyrus (mFus and pFus) and the left inferior and middle occipital gyrus (IOG and MOG). Orange outlines in (B) and (C) indicate the font-invariant repetition suppression effect observed in A. The second map (B) from [Both Δ + LVF Δ + RVF Δ] > No Δ contrast showed that both left and right OT contains voxels of either font invariance, or font sensitivity, or both. Some of these voxels were also observed within the orange outlines in A. The third map (C) from a Font Δ > No Δ contrast revealed font-sensitive repetition suppression in both the left and right pFus and LO. These voxels did not overlap with the font-invariant voxels in the anterior mFus and posterior MOG shown in (A). The same statistical threshold is set at t > 4.44, p < .001, voxel-wise, and q(FDR) < 0.05, for all contrast maps. All color scales range between t = 4.44 and t = 8.

Figure 2. 

Contrast maps obtained in group-level analyses showing left-lateralized font-invariant repetition suppression and font sensitivity in both left and right OT. The first map (A) from a [Both Δ + LVF Δ + RVF Δ] > Font Δ contrast revealed font-invariant repetition suppression in the left middle and posterior fusiform gyrus (mFus and pFus) and the left inferior and middle occipital gyrus (IOG and MOG). Orange outlines in (B) and (C) indicate the font-invariant repetition suppression effect observed in A. The second map (B) from [Both Δ + LVF Δ + RVF Δ] > No Δ contrast showed that both left and right OT contains voxels of either font invariance, or font sensitivity, or both. Some of these voxels were also observed within the orange outlines in A. The third map (C) from a Font Δ > No Δ contrast revealed font-sensitive repetition suppression in both the left and right pFus and LO. These voxels did not overlap with the font-invariant voxels in the anterior mFus and posterior MOG shown in (A). The same statistical threshold is set at t > 4.44, p < .001, voxel-wise, and q(FDR) < 0.05, for all contrast maps. All color scales range between t = 4.44 and t = 8.

Table 1. 
Peak Activations Obtained Using Three Group-level Contrasts (Random-effects General Linear Model) Corresponding to Contrast Maps in Figure 2 
Brain RegionSideCluster Size (mm3)Peak Coordinatet
xyz
[Both Δ + LVF Δ + RVF Δ] > Font Δ 
mFus 1728 −42 −43 −17 7.49 
pFus 5049 −39 −58 −11 5.45 
IOG   −36 −82 −14 5.54 
MOG   −30 −88 −4 7.67 
  
[Both Δ + LVF Δ + RVF Δ] > No Δ 
mFus 15174 −42 −46 −14 7.28 
pFus   −33 −67 −14 7.81 
IOG   −36 −79 −14 8.58 
MOG   −33 −88 9.50 
pFus 4320 36 −56 −17 5.43 
LO 1026 36 −75 −8 7.20 
  
Font Δ > No Δ 
pFus 11070 −42 −55 −14 8.85 
LO   −39 −72 −9 8.81 
pFus 7506 36 −55 −19 7.62 
LO   42 −73 −14 8.56 
Brain RegionSideCluster Size (mm3)Peak Coordinatet
xyz
[Both Δ + LVF Δ + RVF Δ] > Font Δ 
mFus 1728 −42 −43 −17 7.49 
pFus 5049 −39 −58 −11 5.45 
IOG   −36 −82 −14 5.54 
MOG   −30 −88 −4 7.67 
  
[Both Δ + LVF Δ + RVF Δ] > No Δ 
mFus 15174 −42 −46 −14 7.28 
pFus   −33 −67 −14 7.81 
IOG   −36 −79 −14 8.58 
MOG   −33 −88 9.50 
pFus 4320 36 −56 −17 5.43 
LO 1026 36 −75 −8 7.20 
  
Font Δ > No Δ 
pFus 11070 −42 −55 −14 8.85 
LO   −39 −72 −9 8.81 
pFus 7506 36 −55 −19 7.62 
LO   42 −73 −14 8.56 

All contrast maps obtained at a statistical threshold of t > 4.44, q(FDR) < 0.05. L = left; R = right.

We performed a second contrast, [Both Δ + LVF Δ + RVF Δ] > No Δ, which allowed us to identify voxels showing release from suppression to words repeated in the same font (No Δ), either due to the changing identities of the words or constituent letters, or font changes, or both (Figure 2B and Table 1). The purpose of this contrast was to identify voxels that did not show font invariance in the first contrast (indicated by the orange outlines in Figure 2B, which correspond to the maps in Figure 2A). As expected, the contrast map in Figure 2B contained more voxels than that in Figure 2A, indicating that some voxels showing [Both Δ + LVF Δ + RVF Δ] > No Δ (t > 4.44, q(FDR) < 0.05) were sensitive to font changes. Peak activations in this map were consistent with those in Figure 2A but also included voxels in the right pFus and in the right lateral occipital cortex (LO).

The sensitivity of voxels to font changes alone (in each condition, a word was repeated either in the context of a changing font or not) was assessed directly in a third Font Δ > No Δ contrast. The resultant contrast map (Figure 2C) was not lateralized to the left hemisphere (unlike the maps in Figure 2A and B) and showed peak activations in bilateral pFus and LO (Table 1), confirming the sensitivity of these regions to font changes. Critically, the map shown in Figure 2C did not show complete overlap with that shown in Figure 2A (i.e., the orange outlines in Figure 2C do not contain blue voxels in the left mFus and the left MOG), which means that the left OT contains voxels that show greater font invariance than font sensitivity, the latter of which was either weak or nonexistent. In contrast, the blue voxels contained within the orange outline in the contrast maps shown in Figure 2C (i.e., left pFus and LO) showed both sensitivity and invariance to font changes. This explains the appearance of these voxels in all three maps shown in Figure 2 and distinguishes these voxels from the peak voxels in mFus and MOG in Figure 2A.

A previous study (Strother et al., 2016) found that the left OT showed release from suppression for repeated words when either the whole word or half of the word changed. A limitation of this study was that it could not distinguish generic shape changes (due to changing letters) from changes in the identity of words or constituent letters. We therefore performed three final contrasts to test whether or not the left OT was sensitive to letter changes in the left (ipsilateral) visual hemifield (in the LVF Δ condition) or merely to contralateral letter changes (in the RVF Δ and Both Δ conditions). Figure 3 and Table 2 report the results of these final group-level contrasts, all of which yielded font-invariant voxels in the left OT (but not in the right OT) at t > 4.44, p < .001 (the same t value used in Figure 2; voxel-wise because no voxels survived FDR correction). All three maps are similar to that shown in Figure 2A, which means that contralateral letter changes are not requisite for release from font-invariant repetition suppression, consistent with previous studies showing suppression for repeated words without concurrent font changes to infer font-invariant suppression (Strother et al., 2016; Glezer, Kim, Rule, Jiang, & Riesenhuber, 2015; Glezer, Jiang, & Riesenhuber, 2009).

Figure 3. 

Contrast maps obtained in group-level analyses showing consistently left-lateralized font-invariant repetition suppression in the Both Δ > Font Δ, LVF Δ> Font Δ, and RVF Δ > Font Δ contrast maps. All three contrast maps showed font-invariant repetition suppression in similar brain regions, including the left mFus, pFus, IOG, and MOG, as observed in Figure 2A. All contrast maps survived at t > 4.44, p < .001, voxel-wise. Color scale ranges between t = 4.44 and t = 8.

Figure 3. 

Contrast maps obtained in group-level analyses showing consistently left-lateralized font-invariant repetition suppression in the Both Δ > Font Δ, LVF Δ> Font Δ, and RVF Δ > Font Δ contrast maps. All three contrast maps showed font-invariant repetition suppression in similar brain regions, including the left mFus, pFus, IOG, and MOG, as observed in Figure 2A. All contrast maps survived at t > 4.44, p < .001, voxel-wise. Color scale ranges between t = 4.44 and t = 8.

Table 2. 
Peak Activations Obtained Using Three Group-level Contrasts (Random-effects General Linear Model) Corresponding to Contrast Maps in Figure 3 
Brain RegionSideCluster Size (mm3)Peak Coordinatet
xyz
Both Δ > Font Δ 
mFus 1458 −42 −43 −17 6.32 
pFus 2727 −39 −58 −11 4.89 
IOG   −34 −77 −14 4.94 
MOG   −30 −85 −2 6.67 
  
LVF Δ > Font Δ 
mFus 1404 −42 −43 −14 7.71 
pFus 4401 −33 −67 −14 7.59 
IOG   −34 −79 −14 4.71 
MOG   −27 −88 −6 7.07 
  
RVF Δ > Font Δ 
mFus 972 −40 −43 −17 6.46 
pFus 2781 −39 −58 −14 5.32 
IOG   −36 −82 −14 5.43 
MOG   −30 −88 −1 7.44 
Brain RegionSideCluster Size (mm3)Peak Coordinatet
xyz
Both Δ > Font Δ 
mFus 1458 −42 −43 −17 6.32 
pFus 2727 −39 −58 −11 4.89 
IOG   −34 −77 −14 4.94 
MOG   −30 −85 −2 6.67 
  
LVF Δ > Font Δ 
mFus 1404 −42 −43 −14 7.71 
pFus 4401 −33 −67 −14 7.59 
IOG   −34 −79 −14 4.71 
MOG   −27 −88 −6 7.07 
  
RVF Δ > Font Δ 
mFus 972 −40 −43 −17 6.46 
pFus 2781 −39 −58 −14 5.32 
IOG   −36 −82 −14 5.43 
MOG   −30 −88 −1 7.44 

All contrast maps were set at the same statistical threshold of t > 4.44, p < .001, voxel-wise. L = left; R = right.

Individual-level Analyses

In addition to our group-level analyses, we performed two additional types of analyses on individual-level fMRI data. First, we performed analyses of across-subject consistency of lateralization indices (LIs) corresponding to font invariance and font sensitivity. Second, we performed exploratory ROI analyses based on individually defined ROIs to assess font invariance and font sensitivity in the VWFA of the left OT and shape-selective LO.

Lateralization Analyses

For each participant, LIs were calculated using the LI toolbox (Wilke & Lidzba, 2007; Wilke & Schmithorst, 2006), for contrast maps obtained using the same contrasts reported in our group-level analyses. Individual contrast maps were masked to include a large region of the OT within a box ranging between −60 and 60 of the x axis, −90 and −20 of the y axis, and −25 to 10 of the z axis (Talairach coordinates) and excluding voxels ± 5 mm from midline. LIs were defined using the formula:
LI=VoxelsleftVoxelsrightVoxelsleft+Voxelsright
The LI toolbox used a bootstrap algorithm to compute 10,000 LIs at a given threshold, for 20 different thresholds varying equally between 0 and the maximum t value. For each individual-level contrast, LIs were calculated by averaging the middle 50% of the bootstrapped LIs (excluding the upper and lower 25% LI values). This LI is robust to outliers and serves as threshold-free measurement of cerebral laterality. The LI ranges between −1 and 1, and the laterality is indicated by LI > 0.2 (left lateralized), LI < −0.2 (right lateralized), or LI between −0.2 and 0.2 as no lateralization (Jansen et al., 2006).

Figure 4 shows the LIs of all contrast maps for individual participants. On average, the LIs of [Both Δ + LVF Δ + RVF Δ] > Font Δ contrast map (M = 0.51, SEM = 0.06; Figure 4A) showed left lateralization, and 11 of 12 participants were greater than 0.2 (the participant who did not show left lateralization of font-invariant repetition suppression also showed the least significant difference in the corresponding contrast). The left-lateralized font-invariant suppression effect was confirmed using a one-sample t test (against 0.2), t(11) = 5.42, p < .001. This means that the left lateralization apparent in the group-level contrast map in Figure 2A occurred consistently at the individual participant level. In contrast, this was not the case for the two contrasts used to assess font sensitivity. LIs for the [Both Δ + LVF Δ + RVF Δ] > No Δ contrast map (M = 0.17, SEM = 0.09; Figure 4B) were not lateralized (between −0.2 and 0.2) but exhibit a trend of left biased. The font-sensitive LIs (M = −0.17, SEM = 0.10; Figure 4C), revealed by the Font Δ > No Δ contrast map, ranged between −0.58 and 0.41 (i.e., were not consistently lateralized but showed a trend toward right lateralization, indicated by the negative mean), and showed a greater variability than the font-invariant LIs (which were all positive).

Figure 4. 

Individual LIs for all contrast maps. Strong left lateralization for the [Both Δ + LVF Δ + RVF Δ] > Font Δ contrast was observed across 11 of 12 participants (A). No significant lateralization was observed for the [Both Δ + LVF ΔSEM.

Figure 4. 

Individual LIs for all contrast maps. Strong left lateralization for the [Both Δ + LVF Δ + RVF Δ] > Font Δ contrast was observed across 11 of 12 participants (A). No significant lateralization was observed for the [Both Δ + LVF ΔSEM.

Finally, the LIs of the Both Δ > Font Δ (M = 0.47, SEM = 0.06; Figure 4D), LVF Δ > Font Δ (M = 0.37, SEM = 0.07; Figure 4E), and RVF Δ > Font Δ (M = 0.53, SEM = 0.07; Figure 4F) contrasts were all left lateralized (all ps < .05). The LVF Δ > Font Δ contrast showed the weakest left lateralization as compared with either Both Δ > Font Δ or RVF Δ > Font Δ contrast, F(2, 22) = 11.34, p < .005 (post hoc paired comparisons, both ps < .05). Taken together, these results show that font-invariant repetition suppression is more strongly associated with the left OT as compared with font sensitivity.

ROI Analyses

We performed ROI analyses for each individual to assess font invariance and font sensitivity in word-selective and shape-selective regions of OT. These analyses were somewhat exploratory in that we defined ROIs based on previously published results with the intention of defining shape-selective regions of OT, namely LO, without a planned independent localizer. Of the 12 participants in the current study, eight had VWFA and LO ROIs defined in previously published experiments (Strother et al., 2016, 2017), which directly compared repetition suppression effects for repeated words and objects to results from standard word-selective (words > nonwords; Zhang, He, & Weng, 2018) and shape-selective localizers (objects > scrambled objects; Strother, Aldcroft, Lavell, & Vilis, 2010). For these eight participants, ROIs were defined as a cube (∼700 mm3 for VWFA and ∼1200 mm3 for bilateral LO) centered on the peak word-selective (VWFA) and object-selective (LO) voxels identified previously. For the remaining four, ROI cubes were centered on the average Talairach coordinates for each ROI type, obtained from the eight participants who had individually defined ROIs. For the VWFA, the average peak coordinate was located at x = −44.3, y = −53.8, z = 15.3; for the left LO, the average peak coordinate was located at x = −43.1, y = −73.8, z = −8.3; and for the right LO, the average peak coordinate was located at x = 42.8, y = 71.4, z = −9.9. Figure 5 shows mean beta weight values extracted for all five conditions from three ROIs (the VWFA and the left and right LO).

Figure 5. 

ROI analyses showed font invariance in the VWFA and font sensitivity in bilateral LO. The VWFA showed both font-sensitive and font-invariant effect, but the font-invariant effect was relatively stronger than the font-sensitive effect (A). No difference was observed between Both Δ, LVF Δ, and RVF Δ conditions. The left LO also showed both font-sensitive and font-invariant effect (B); however, the font-sensitive effect was relatively stronger than the font-invariant effect. No difference was observed between Both Δ, LVF Δ, and RVF Δ conditions. (C) The right LO only showed font-sensitive effect. No difference was observed between the Font Δ, Both Δ, LVF Δ, and RVF Δ conditions. NS refers to nonsignificant. p < .1, **p < .01, ***p < .001, Bonferroni-corrected. Error bars represent 1 SEM.

Figure 5. 

ROI analyses showed font invariance in the VWFA and font sensitivity in bilateral LO. The VWFA showed both font-sensitive and font-invariant effect, but the font-invariant effect was relatively stronger than the font-sensitive effect (A). No difference was observed between Both Δ, LVF Δ, and RVF Δ conditions. The left LO also showed both font-sensitive and font-invariant effect (B); however, the font-sensitive effect was relatively stronger than the font-invariant effect. No difference was observed between Both Δ, LVF Δ, and RVF Δ conditions. (C) The right LO only showed font-sensitive effect. No difference was observed between the Font Δ, Both Δ, LVF Δ, and RVF Δ conditions. NS refers to nonsignificant. p < .1, **p < .01, ***p < .001, Bonferroni-corrected. Error bars represent 1 SEM.

A one-way repeated-measures ANOVA was conducted on individuals' beta weights for each of the five conditions. Bonferroni correction for multiple comparisons was applied for the following paired comparisons. For the VWFA (Figure 5A), the ANOVA results showed a main effect of Condition, F(4, 44) = 45.14, p < .001. Post hoc paired comparisons showed that the Both Δ beta weights were larger than the Font Δ beta weights (p < .001) and the Font Δ beta weights were also larger than the No Δ beta weights (p < .001). Additionally, paired comparisons showed no significant difference among the Both Δ, LVF Δ, and RVF Δ beta weights (all ps > .1). This means that, at the individual level, the VWFA contains both font-invariant and font-sensitive effects and that both whole-word and half-word changes yield similar beta weights. Because both font invariance and font sensitivity were observed in the VWFA, we further compared the effects of the corresponding conditions by subtracting the Font Δ beta weights from the Both Δ beta weights, and the No Δ beta weights from the Font Δ beta weights, respectively. The difference values between the Both Δ and Font Δ beta weights were 0.38 (SEM = 0.05), and the difference between Font Δ and No Δ beta weights were 0.20 (SEM = 0.03). A paired sample t test on these differences suggested that the font-invariant effect (Both Δ − Font Δ) was greater than the font-sensitive effect (Font Δ − No Δ), t(11) = 4.85, p < .001. Additional analysis on individually defined VWFA of the eight participants showed a similar result as observed in all the 12 participants, that the font-invariant effect was greater than the font-sensitive effect, t(7) = 5.55, p < .001. This means that, although the VWFA showed both font invariance and font sensitivity, it showed greater font invariance.

Figure 5B and C shows beta weights for the five conditions from the left and right LO. A one-way repeated-measures ANOVA was conducted before paired comparison. For the left LO, the result revealed a main effect of Condition, F(4, 44) = 39.40, p < .001. Post hoc paired comparisons indicated that the Both Δ beta weights were marginally larger than the Font Δ beta weights, p = .073 (Bonferroni-corrected p), and Font Δ beta weights were significantly larger than the No Δ beta weights, p < .001. In addition, paired comparisons showed no significant difference among the Both Δ, LVF Δ, and RVF Δ beta weights (all ps > .1). We again tested the font-sensitive and font-invariant effects using the beta differences between the Both Δ and Font Δ, and Font Δ and No Δ conditions. Unlike the VWFA, the left LO showed a reverse effect that the font-sensitive effect (M = 0.38, SEM = 0.06) was greater than the font-invariant effect (M = 0.14, SEM = 0.05), t(11) = 7.30, p < .001. For the right LO, the result showed a main effect of Condition, F(4, 44) = 36.49, p < .001. Post hoc paired comparisons indicated that the Both Δ beta weights were not different from the Font Δ beta weights, p = 1.00; but the Font Δ beta weights were significantly larger than the No Δ beta weights, p < .001. Again, paired comparisons showed no significant difference among the Font Δ, Both Δ, LVF Δ, and RVF Δ beta weights (all ps > .1). Additional analyses of individually defined left and right LOs of the 8 participants showed a similar result as observed in all the 12 participants. In the left LO, the font-sensitive effect was greater than the font-invariant effect, t(7) = 6.47, p < .001, and in the right LO, there was no significant difference between the Both Δ and Font Δ conditions, t(7) = 1.37, p = .21. Taken together, these findings suggest that the left OT contains both font-sensitive and font-invariant effects, whereas the right OT only contains font-sensitive effect. Importantly, even though the left LO showed font invariance, it showed greater font sensitivity, which clearly delineates the left LO from the VWFA.

DISCUSSION

We measured fMRI repetition suppression in OT for words repeated in different fonts as compared with different words in different fonts, which we interpreted as an indicator of font-invariant representation. We also measured suppression of fMRI responses for words repeated in the same font as compared with those repeated in different fonts. This allowed us to infer sensitivity to font-related shape changes and to compare font sensitivity to font-invariant repetition suppression directly. We observed the greatest degree of font-invariant suppression in the left OT, especially in the vicinity of the VWFA and in the left MOG. Font-invariant suppression was weak or nonexistent in the right OT. In contrast, font sensitivity was greatest in bilateral shape-selective cortex (LO). Although the VWFA and left MOG both showed some sensitivity to font changes, font invariance was greatest where font sensitivity was weakest (VWFA and left MOG), and font sensitivity was greatest where font invariance was weakest or absent (LO). Our findings are consistent with font-invariant representation letters and words by neural mechanisms in both the VWFA and more posterior brain regions in the left occipital lobe.

Font Invariance in the Left Hemisphere

The results shown in Figures 2 and 3 clearly indicate font-invariant repetition suppression in the left OT. These results complement and extend those obtained by Strother et al. (2016), who showed repetition suppression in the VWFA and in the left occipital lobe, but not in the context of varying font. The release from suppression reported here results from changes in word or letter identities rather than identity-irrelevant shape changes, because both the baseline (Font Δ) and releasing conditions (Both Δ, LVF Δ, and RVF Δ) involved constant font changes in successively viewed words. Our finding of font-invariant release from repetition suppression in the VWFA is consistent with the view that this region of the left OT contains abstract representations of visual word form (Dehaene et al., 2004; Rothlein & Rapp, 2014). Surprisingly, our finding of font invariance in the left OT was not limited to the VWFA and was also observed in a substantially posterior brain region in the left (but not the right) occipital lobe. Our finding of font-invariant repetition suppression in the left MOG and the left IOG is consistent with other findings implicating neural mechanisms in the left occipital lobe in visual word recognition (e.g., Zhang et al., 2018; Boros et al., 2016; Strother et al., 2016; Yu, Jiang, Legge, & He, 2015; Rothlein & Rapp, 2014).

Neurally plausible models of word recognition generally posit that neural mechanisms in the occipital lobe extract basic visual features from which increasingly complex representations of letters and words arise during subsequent stages of visual processing in more anterior regions of OT. This progression is synonymous with a graded posterior-to-anterior hierarchy, within which the VWFA resides at the anterior-most portion of the left fusiform cortex (Vinckier et al., 2007). Our results both support and complicate this posterior-to-anterior account. Although our results are consistent with increasingly complex and invariant word representation in the VWFA, our finding of equivalent font invariance in the left occipital lobe (MOG and, to a lesser degree, the left IOG) supports the possibility that abstract representation of letters and words begins at the level of visual feature processing by neural mechanisms posterior to the VWFA (Hauk, Davis, Ford, Pulvermüller, & Marslen-Wilson, 2006; Flowers et al., 2004; Tarkiainen, Helenius, Hansen, Cornelissen, & Salmelin, 1999), possibly in conjunction with letter-level processing within or near the VWFA (Thesen et al., 2012; James, James, Jobard, Wong, & Gauthier, 2005).

It is worth noting that another fMRI study reported repetition suppression for cross-case single letters in the left VWFA, which they interpreted as evidence of a similarity-based explanation (Burgund & Edwards, 2008), based on the visual similarity of letter pairs (e.g., suppression for j → J but less for a → A). Although our results might at first appear to show some consistency with a visual similarity explanation applied to whole words, our ROI results, in particular at least two findings, run counter to this type of explanation. First, if visual similarity were to explain our results, then we would expect fMRI responses in the LVF Δ and RVF Δ conditions (half-word changes) to be intermediate to those in the Both Δ (whole-word changes) and Font Δ (font change only) conditions. Second, our laterality analyses showed substantial differences in font sensitivity between the two hemispheres, consistent with the emergence of font invariance in the left hemisphere to a greater degree than in the right. Ideally, future experiments should explicitly manipulate visual similarity as a function of letter identity changes versus font changes and address metric versus qualitative shape changes to clarify the relative contributions of each type of shape change.

Font Sensitivity versus Invariance

In contrast to the font-invariant repetition suppression in the left hemisphere, font-sensitive release from repetition suppression (i.e., greater fMRI responses to the repeated words viewed in a changing font as compared with repeated word viewed in an unchanging font) occurred bilaterally. We observed the greatest degree of font sensitivity in bilateral shape-selective LO (Figure 2), which is associated with shape-based object recognition (Kourtzi & Kanwisher, 2001; Malach et al., 1995). Font sensitivity was weaker in regions of OT both anterior and posterior to LO. In the left OT, font sensitivity occurred to a lesser degree than font invariance, especially in the VWFA and MOG, which showed relatively little if any sensitivity to font-specific changes in letter shape. The right OT showed somewhat equivalent sensitivity to font changes but no indication of font-invariant repetition suppression. The posterior-to-anterior transition from font-invariant left MOG to font-sensitive left LO, and then to the font-invariant VWFA in the left OT, disfavors the possibility of a simple posterior-to-anterior gradient of abstract word representation in the left OT with respect to font invariance. Also, it is possible that the font-sensitive regions of bilateral OT identified here are related to the bilateral font-sensitive “letter-specific area” reported by Gauthier et al. (2000), which may not have been sensitive to font per se but, rather, font-related differences in the shape of a specific letter.

The observed transition from font invariance to font sensitivity in the left occipital lobe and then back to font invariance in the VWFA raises the possibility that these regions (and other regions showing either font invariance, font sensitivity, or both; see Tables 1 and 2) work together to segregate shape information specifying word form from that corresponding to shape (i.e., font) variation unrelated to changing letters or words. Alternatively, the sensitivity to font changes in LO (and in the left pFus, which showed greater font invariance than left LO) may simply reflect the sensitivity to generic shape changes in this region, with only the VWFA having the ability to discriminate font changes from changes in letter identity and doing so possibly without any involvement of generic shape-processing mechanisms in LO. Regardless, our results show that font invariance occurs in regions of the left occipital cortex associated with the early processing of visual feature information during word recognition (Levy et al., 2008). Furthermore, our results show clear laterality of font invariance in the occipital lobe (the MOG and IOG in particular; see Figure 3 and Table 2), consistent with other findings of cerebral laterality in these regions during visual word recognition (Zhang et al., 2018; Strother et al., 2016; Stigliani, Weiner, & Grill-Spector, 2015; Yu et al., 2015; Levy et al., 2009; Gold & Rastle, 2007). The possible involvement of the left occipital cortex in abstract letter detection or other prelexical visual word recognition processes again highlight the possibility that font invariance occurs at different stages of visual word form processing accomplished by neural mechanisms at disparate anatomical locations within the left OT.

Left-lateralized Visual Word Form Processing in Occipital Cortex

Our results showed consistent left lateralization of font-invariant repetition suppression, which was not the case for font-sensitive release from suppression (Figure 4). The left MOG and IOG showed font-invariant repetition suppression, the magnitude of which was roughly equivalent to that observed in the VWFA, in both group-level and individual-level results (Figures 2 and 5). We will refer to these regions of the left occipital lobe collectively as an OWFA. Like the VWFA, both regions of the OWFA show unique responses to words as compared with nonlinguistic stimuli (Zhang et al., 2018; Strother et al., 2017; Yu et al., 2015). Strother et al. (2016) defined the IOG portion of the OWFA (which in the current study showed less font invariance than the MOG portion) as a possible cortical site of hemifield integration for foveally split words. Here, we propose that an adjacent region in the MOG may represent font-invariant letter information (Rothlein & Rapp, 2014), in both hemifields (Barca et al., 2011), in addition to its previously reported roles in orthography-related coding of letter order and grapheme parsing (Boros et al., 2016; Levy et al., 2009).

In conclusion, by measuring fMRI responses to repeated or changing words in the context of a repeated or changing font, we delineated font-invariant neural populations in the left OT from bilateral shape-selective LO. We also partially replicated a previous finding by Strother et al. (2016) that used a similar repetition suppression method to show the neural basis of hemifield-split word integration in the left occipital lobe. Our findings complicate a simple posterior-to-anterior gradient of visual word form processing (also see Lochy et al., 2018), in which abstract representation of letters and words is limited to relatively anterior regions of the left OT and highlights the potential role of neural mechanism in the left occipital lobe in visual word recognition. The results reported here complement and add to those of previous fMRI studies of word recognition in the left OT by revealing equivalently abstract (font-invariant) representation of word form in the VWFA and OWFA, either or both of which may contain neural mechanisms involved in abstract letter detection.

Acknowledgments

This research was funded by Canadian Institutes of Health grant 9335 to T. V. and the National Institute of General Medical Sciences of the National Institutes of Health under grant Number P20 GM103650. We thank M. Joanisse for assistance with the word stimuli.

Reprint requests should be sent to Lars Strother, Department of Psychology, University of Nevada, Reno, MS0296, 1664 N. Virginia Street, Reno, NV 89557, or via e-mail: lars@unr.edu.

REFERENCES

REFERENCES
Andresen
,
D. R.
,
Vinberg
,
J.
, &
Grill-Spector
,
K.
(
2009
).
The representation of object viewpoint in human visual cortex
.
Neuroimage
,
45
,
522
536
.
Barca
,
L.
,
Cornelissen
,
P.
,
Simpson
,
M.
,
Urooj
,
U.
,
Woods
,
W.
, &
Ellis
,
A. W.
(
2011
).
The neural basis of the right visual field advantage in reading: An MEG analysis using virtual electrodes
.
Brain and Language
,
118
,
53
71
.
Boros
,
M.
,
Anton
,
J.-L.
,
Pech-Georgel
,
C.
,
Grainger
,
J.
,
Szwed
,
M.
, &
Ziegler
,
J. C.
(
2016
).
Orthographic processing deficits in developmental dyslexia: Beyond the ventral visual stream
.
Neuroimage
,
128
,
316
327
.
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
.
Burgund
,
E. D.
, &
Edwards
,
J. K.
(
2008
).
Identity versus similarity priming for letters in left mid-fusiform cortex
.
NeuroReport
,
19
,
761
764
.
Burgund
,
E. D.
,
Guo
,
Y.
, &
Aurbach
,
E. L.
(
2009
).
Priming for letters and pseudoletters in mid-fusiform cortex: Examining letter selectivity and case invariance
.
Experimental Brain Research
,
193
,
591
601
.
Chen
,
C.-C.
,
Kao
,
K.-L. C.
, &
Tyler
,
C. W.
(
2007
).
Face configuration processing in the human brain: The role of symmetry
.
Cerebral Cortex
,
17
,
1423
1432
.
Cohen
,
L.
,
Lehéricy
,
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
.
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
.
Flowers
,
D. L.
,
Jones
,
K.
,
Noble
,
K.
,
VanMeter
,
J.
,
Zeffiro
,
T. A.
,
Wood
,
F. B.
, et al
(
2004
).
Attention to single letters activates left extrastriate cortex
.
Neuroimage
,
21
,
829
839
.
Gauthier
,
I.
,
Tarr
,
M. J.
,
Moylan
,
J. A.
,
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
.
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
.
Gold
,
B. T.
, &
Rastle
,
K.
(
2007
).
Neural correlates of morphological decomposition during visual word recognition
.
Journal of Cognitive Neuroscience
,
19
,
1983
1993
.
Grainger
,
J.
,
Dufau
,
S.
, &
Ziegler
,
J. C.
(
2016
).
A vision of reading
.
Trends in Cognitive Sciences
,
20
,
171
179
.
Grill-Spector
,
K.
,
Kushnir
,
T.
,
Edelman
,
S.
,
Avidan
,
G.
,
Itzchak
,
Y.
, &
Malach
,
R.
(
1999
).
Differential processing of objects under various viewing conditions in the human lateral occipital complex
.
Neuron
,
24
,
187
203
.
Hauk
,
O.
,
Davis
,
M. H.
,
Ford
,
M.
,
Pulvermüller
,
F.
, &
Marslen-Wilson
,
W. D.
(
2006
).
The time course of visual word recognition as revealed by linear regression analysis of ERP data
.
Neuroimage
,
30
,
1383
1400
.
James
,
K. H.
,
James
,
T. W.
,
Jobard
,
G.
,
Wong
,
A. C. N.
, &
Gauthier
,
I.
(
2005
).
Letter processing in the visual system: Different activation patterns for single letters and strings
.
Cognitive, Affective, & Behavioral Neuroscience
,
5
,
452
466
.
Jansen
,
A.
,
Menke
,
R.
,
Sommer
,
J.
,
Förster
,
A. F.
,
Bruchmann
,
S.
,
Hempleman
,
J.
, et al
(
2006
).
The assessment of hemispheric lateralization in functional MRI—Robustness and reproducibility
.
Neuroimage
,
33
,
204
217
.
Kourtzi
,
Z.
, &
Kanwisher
,
N.
(
2001
).
Representation of perceived object shape by the human lateral occipital complex
.
Science
,
293
,
1506
1509
.
Larsson
,
J.
, &
Heeger
,
D. J.
(
2006
).
Two retinotopic visual areas in human lateral occipital cortex
.
Journal of Neuroscience
,
26
,
13128
13142
.
Levy
,
J.
,
Pernet
,
C.
,
Treserras
,
S.
,
Boulanouar
,
K.
,
Aubry
,
F.
,
Démonet
,
J.-F.
, et al
(
2009
).
Testing for the dual-route cascade reading model in the brain: An fMRI effective connectivity account of an efficient reading style
.
PLoS One
,
4
,
e6675
.
Levy
,
J.
,
Pernet
,
C.
,
Treserras
,
S.
,
Boulanouar
,
K.
,
Berry
,
I.
,
Aubry
,
F.
, et al
(
2008
).
Piecemeal recruitment of left-lateralized brain areas during reading: A spatio-functional account
.
Neuroimage
,
43
,
581
591
.
Lochy
,
A.
,
Jacques
,
C.
,
Maillard
,
L.
,
Colnat-Coulbois
,
S.
,
Rossion
,
B.
, &
Jonas
,
J.
(
2018
).
Selective visual representation of letters and words in the left ventral occipito-temporal cortex with intracerebral recordings
.
Proceedings of the National Academy of Sciences, U.S.A.
,
115
,
E7595
E7604
.
Malach
,
R.
,
Reppas
,
J. B.
,
Benson
,
R. R.
,
Kwong
,
K. K.
,
Jiang
,
H.
,
Kennedy
,
W. A.
, et al
(
1995
).
Object-related activity revealed by functional magnetic resonance imaging in human occipital cortex
.
Proceedings of the National Academy of Sciences, U.S.A.
,
92
,
8135
8139
.
Nestor
,
A.
,
Behrmann
,
M.
, &
Plaut
,
D. C.
(
2013
).
The neural basis of visual word form processing: A multivariate investigation
.
Cerebral Cortex
,
23
,
1673
1684
.
Polk
,
T. A.
,
Stallcup
,
M.
,
Aguirre
,
G. K.
,
Alsop
,
D. C.
,
D'Esposito
,
M.
,
Detre
,
J. A.
, et al
(
2002
).
Neural specialization for letter recognition
.
Journal of Cognitive Neuroscience
,
14
,
145
159
.
Qiao
,
E.
,
Vinckier
,
F.
,
Szwed
,
M.
,
Naccache
,
L.
,
Valabrègue
,
R.
,
Dehaene
,
S.
, et al
(
2010
).
Unconsciously deciphering handwriting: Subliminal invariance for handwritten words in the visual word form area
.
Neuroimage
,
49
,
1786
1799
.
Rauschecker
,
A. M.
,
Bowen
,
R. F.
,
Parvizi
,
J.
, &
Wandell
,
B. A.
(
2012
).
Position sensitivity in the visual word form area
.
Proceedings of the National Academy of Sciences, U.S.A.
,
109
,
E1568
E1577
.
Rothlein
,
D.
, &
Rapp
,
B.
(
2014
).
The similarity structure of distributed neural responses reveals the multiple representations of letters
.
Neuroimage
,
89
,
331
344
.
Sanocki
,
T.
(
1988
).
Font regularity constraints on the process of letter recognition
.
Journal of Experimental Psychology: Human Perception and Performance
,
14
,
472
480
.
Sayres
,
R.
, &
Grill-Spector
,
K.
(
2008
).
Relating retinotopic and object-selective responses in human lateral occipital cortex
.
Journal of Neurophysiology
,
100
,
249
267
.
Stigliani
,
A.
,
Weiner
,
K. S.
, &
Grill-Spector
,
K.
(
2015
).
Temporal processing capacity in high-level visual cortex is domain specific
.
Journal of Neuroscience
,
35
,
12412
12424
.
Strother
,
L.
,
Aldcroft
,
A.
,
Lavell
,
C.
, &
Vilis
,
T.
(
2010
).
Equal degrees of object selectivity for upper and lower visual field stimuli
.
Journal of Neurophysiology
,
104
,
2075
2081
.
Strother
,
L.
,
Coros
,
A. M.
, &
Vilis
,
T.
(
2016
).
Visual cortical representation of whole words and hemifield-split word parts
.
Journal of Cognitive Neuroscience
,
28
,
252
260
.
Strother
,
L.
,
Zhou
,
Z.
,
Coros
,
A. K.
, &
Vilis
,
T.
(
2017
).
An fMRI study of visual hemifield integration and cerebral lateralization
.
Neuropsychologia
,
100
,
35
43
.
Talairach
,
J.
, &
Tournoux
,
P.
(
1988
).
Co-planar stereotaxic atlas of the human brain: 3-dimensional proportional system: An approach to cerebral imaging
.
Stuttgart
:
Thieme
.
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
.
Tarr
,
M. J.
, &
Hayward
,
W. G.
(
2017
).
The concurrent encoding of viewpoint-invariant and viewpoint-dependent information in visual object recognition
.
Visual Cognition
,
25
,
100
121
.
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
.
Vernon
,
R. J. W.
,
Gouws
,
A. D.
,
Lawrence
,
S. J. D.
,
Wade
,
A. R.
, &
Morland
,
A. B.
(
2016
).
Multivariate patterns in the human object-processing pathway reveal a shift from retinotopic to shape curvature representations in lateral occipital areas, LO-1 and LO-2
.
Journal of Neuroscience
,
36
,
5763
5774
.
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
.
Vuilleumier
,
P.
,
Henson
,
R. N.
,
Driver
,
J.
, &
Dolan
,
R. J.
(
2002
).
Multiple levels of visual object constancy revealed by event-related fMRI of repetition priming
.
Nature Neuroscience
,
5
,
491
499
.
Wilke
,
M.
, &
Lidzba
,
K.
(
2007
).
LI-tool: A new toolbox to assess lateralization in functional MR-data
.
Journal of Neuroscience Methods
,
163
,
128
136
.
Wilke
,
M.
, &
Schmithorst
,
V. J.
(
2006
).
A combined bootstrap/histogram analysis approach for computing a lateralization index from neuroimaging data
.
Neuroimage
,
33
,
522
530
.
Wimmer
,
H.
,
Ludersdorfer
,
P.
,
Richlan
,
F.
, &
Kronbichler
,
M.
(
2016
).
Visual experience shapes orthographic representations in the visual word form area
.
Psychological Science
,
27
,
1240
1248
.
Yu
,
D.
,
Jiang
,
Y.
,
Legge
,
G. E.
, &
He
,
S.
(
2015
).
Locating the cortical bottleneck for slow reading in peripheral vision
.
Journal of Vision
,
15
,
3
.
Zhang
,
B.
,
He
,
S.
, &
Weng
,
X.
(
2018
).
Localization and functional characterization of an occipital visual word form sensitive area
.
Scientific Reports
,
8
,
6723
.