(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Similar object shape representation encoded in the inferolateral occipitotemporal cortex of sighted and early blind people [1]

['Yangwen Xu', 'Center For Mind Brain Sciences', 'Cimec', 'University Of Trento', 'Trento', 'Lorenzo Vignali', 'International School For Advanced Studies', 'Sissa', 'Trieste', 'Federica Sigismondi']

Date: 2023-08

We can sense an object’s shape by vision or touch. Previous studies suggested that the inferolateral occipitotemporal cortex (ILOTC) implements supramodal shape representations as it responds more to seeing or touching objects than shapeless textures. However, such activation in the anterior portion of the ventral visual pathway could be due to the conceptual representation of an object or visual imagery triggered by touching an object. We addressed these possibilities by directly comparing shape and conceptual representations of objects in early blind (who lack visual experience/imagery) and sighted participants. We found that bilateral ILOTC in both groups showed stronger activation during a shape verification task than during a conceptual verification task made on the names of the same manmade objects. Moreover, the distributed activity in the ILOTC encoded shape similarity but not conceptual association among objects. Besides the ILOTC, we also found shape representation in both groups’ bilateral ventral premotor cortices and intraparietal sulcus (IPS), a frontoparietal circuit relating to object grasping and haptic processing. In contrast, the conceptual verification task activated both groups’ left perisylvian brain network relating to language processing and, interestingly, the cuneus in early blind participants only. The ILOTC had stronger functional connectivity to the frontoparietal circuit than to the left perisylvian network, forming a modular structure specialized in shape representation. Our results conclusively support that the ILOTC selectively implements shape representation independently of visual experience, and this unique functionality likely comes from its privileged connection to the frontoparietal haptic circuit.

Funding: This work was supported by the Belgian Excellence of Science (EOS) program (Project No. 30991544) attributed to O.C., the Research Projects of National Interest (PRIN) grant from the Italian Ministry of Education, University and Research (MIUR) attributed to D.C. and O.C. (Project No. 2015PCNJ5F_001), the Flag-ERA HBP PINT-MULTI (R.8008.19) attributed to O.C., and a mandate d’impulsion scientifique from the Fonds National de la Recherche Scientifique de Belgique (FRS-FNRS) attributed to O.C. O.C. is a research associate at the Fonds National de la Recherche Scientifique de Belgique (FRS-FNRS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

If the ILOTC implements supramodal shape representation, we should find the ILOTC showing greater activation in the shape task than in the conceptual task in both sighted and early blind participants, and the activity pattern in the ILOTC should encode objects’ shape but not conceptual properties. The ILOTC is expected to have stronger connections to the frontoparietal haptic network than the perisylvian language network. If the ILOTC represents objects’ conceptual knowledge instead, we should observe greater activation in the conceptual task than in the shape task in both sighted and early blind participants, and the activity pattern in the ILOTC should encode objects’ conceptual properties. Alternatively, if the activation in the ILOTC depends on visual experience, we should observe the ILOTC’s involvement in shape processing/representation only in the sighted but not in the early blind participants. If such “visual” ILOTC repurposes its function to conceptual representations in the early blind, we should find the ILOTC’s involvement in conceptual representation only in the early blind but not in the sighted participants.

To address these unsolved questions comprehensively in a single study, we used functional magnetic resonance imaging (fMRI) to characterize the brain activity of sighted and early blind participants when they were performing both shape and conceptual verification tasks on the same set of auditory words referring to manmade objects. Univariate contrast between shape and conceptual tasks was performed to localize brain areas specific for shape or conceptual processing. We chose words instead of haptic objects as stimuli because words are arbitrary symbols bearing no obvious resemblance to the objects signified. That means the words, per se, do not carry object information and can elicit shape and conceptual representations without bias. In contrast, haptic objects carry shape information. The participants would have to process the shape information to recognize the objects in both shape and conceptual tasks, and we could no longer isolate the shape representation by contrasting the shape task with the conceptual task. Besides task manipulation, we also orthogonalized the pairwise shape similarity and the pairwise conceptual association among the objects we selected (e.g., a “plate” is perceptually similar to a “coin” in shape but is conceptually associated with a “fork” in function). Representational similarity analyses (RSA, [ 35 ]), therefore, can be conducted to disentangle the regions implementing shape and conceptual representations. Furthermore, we used resting-state functional connectivity (RSFC) to detect the possible synchronizations between the ILOTC and the frontoparietal haptic network or the perisylvian language network.

Third, the ILOTC might engage in (visual) shape representation in the sighted but conceptual representation in the early blind. The pluripotent neuroplasticity hypothesis predicts that the “visual” cortex in the early blind, due to a lack of visual input since birth, could repurpose its function for cognitive faculties that are distant from its native computation in vision, like language or mathematics (see review [ 32 ]). This neurofunctional reorganization process usually accompanies enhanced connectivity between the “visual” cortex in the early blind and high-order brain systems [ 32 ]. In line with this hypothesis, it has been reported that the “visual” cortex in the early blind is more sensitive to lexical semantics than the sighted participants (e.g., [ 33 , 34 ]). Moreover, the activity in the lateral occipital cortex in the early blind is more synchronized to the areas in the perisylvian language network than in the sighted participants [ 34 ]. It is thus possible that the ILOTC in the early blind implements conceptual instead of shape representation due to functional reorganization.

Second, the ILOTC might engage in conceptual representation of objects. An object does not only have a shape, it carries meaning and serves a function. Whenever in the above contrasts between objects and textures (e.g., [ 12 , 13 ]) or between the haptic condition with objects and the hand movement condition without objects [ 22 ], the isolated cognitive component could be conceptual, not perceptual. Previous studies have indeed suggested that the ventral visual pathway might encode semantic relatedness among objects (e.g., [ 24 ]). This is even more likely for the ILOTC. Regions overlapping or slightly superior to the ILOTC show category preference for manmade objects that persists in the early blind participants (e.g., [ 25 – 28 ]), and patients with lesions in the left lateral occipitotemporal cortex are slower to make conceptual associations among manmade objects (e.g., hammer-nail) [ 29 ]. However, this hypothesis was challenged by a recent study showing that the activity pattern in the ILOTC can encode object shapes when stimuli are meaningless novel shape models [ 30 ]. Nevertheless, these findings cannot rule out ILOTC’s involvement in conceptual representation; the ILOTC might support an integrative coding of both visual and conceptual knowledge, as already shown in some other regions in the ventral visual stream [ 31 ].

First, the LOTC might engage in haptic tasks simply due to visual imagery. This hypothesis is supported by studies showing that experiences of visual imagery during haptic shape perception are common, and ratings of the vividness of visual imagery strongly predict the amount of haptic shape-selective activity in the right LOC [ 21 ]. To test whether visual imagery is a prerequisite for ILOTC’s involvement during nonvisual tasks, two studies have tested early blind participants who lack visual imagery. These two studies, however, do not allow to settle the debate. One study found ILOTC’s activation when contrasting a haptic object recognition task and a task imitating the grasping and exploration of objects [ 22 ]. Since this study did not match the two contrasted conditions on task demand and object semantics (see next paragraph), the isolated cognitive components might not be specific to shape processing. The other study, instead, did not find that the ILOTC-encoded object shape in the early blind participants and localized shape representation in other occipitotemporal regions [ 23 ]. Participants in this study performed a shape-irrelevant task (i.e., size judgment task), which might have dampened the brain activation relating to shape representation in the ILOTC.

In addition to the frontoparietal circuit, haptic shape perception intriguingly involved the anterolateral part of the LOC, a region located in the inferolateral occipitotemporal cortex (ILOTC); this region shows stronger activation when participants both see or touch objects in comparison to shapeless textures (e.g., [ 12 – 15 ]). Based on this unique multisensory property, researchers termed the ILOTC region the lateral occipital tactile-visual complex (LOtv, [ 13 ]) and suggested it implements supramodal shape representation [ 3 ]. However, the nature of ILOTC remains debated, as current findings could also support alternative hypotheses.

Studies on haptic shape representation highlighted the neural circuit in the ventral frontoparietal cortex. Researchers found that the intraparietal sulcus (IPS; e.g., [ 12 – 15 ]) and the ventral part of the premotor cortex (vPMC, e.g., [ 14 , 15 ]) show greater activation when participants touch objects than textures. Lesions in the superior parietal cortex and the adjacent IPS induce contralateral tactile agnosia characterized by somatosensory discrimination deficits in the macrogeometrical domain (i.e., detecting differences in length of cuboids) but not in the microgeometrical domain (i.e., detecting subtle differences in grating profiles), whereas lesions in the postcentral gyrus cause the opposite syndrome [ 16 ]. Lesions in the anterior IPS (aIPS) and vPMC can also impair contralateral object exploration—patients cannot recognize objects haptically due to the disturbance of finely tuned finger movements, specifically when interacting with objects [ 17 , 18 ]. In the macaque brain, the homologous regions of both the aIPS (i.e., the AIP) and the vPMC (i.e., the F5) host the neurons that fire when monkeys configure their hands to grasp objects in particular shapes (e.g., [ 19 , 20 ]).

Cognitive neuroscientists usually investigate object shape representation along separate visual and haptic brain pathways. Studies on visual shape representation mostly focus on the ventral visual pathway in the occipitotemporal cortex. Researchers found that the lateral occipital cortex and the posterior fusiform gyrus (i.e., the lateral occipital complex, LOC) show greater activation to object images than texture images (see review [ 5 ]). By contrast, the medial part of the visual cortex is more sensitive to visual texture than visual shape (e.g., [ 6 – 8 ]). Lesions in the LOC induce visual form agnosia manifested as impaired shape discrimination but preserved texture discrimination performance [ 9 , 10 ], whereas lesions in the medial part of the visual cortex cause the opposite syndrome (e.g., [ 11 ]).

Object properties can be accessed through multiple sensory channels. For example, knowledge of an object’s shape can be acquired both by vision and touch. This brings up a critical question about the cerebral architecture of object representation: Are shape representations derived from different senses segregated from each other in the human brain, or, alternatively, could the brain implement a shared representation of object shape that is abstracted from the senses (e.g., [ 1 – 4 ])?

Fig 6C illustrates the mean RSFC matrix across participants in EB and SC among the regions showing stronger activation in the shape tasks or in the conceptual task. It shows that the brain areas sensitive to the shape task and those sensitive to the conceptual task belonged to separate network modules in both EB and SC. Fig 6D further compares the mean RSFC across all the pairs among the shape-sensitive regions, among the conceptual-sensitive regions, and between the shape- and the conceptual-sensitive regions. In both EB and SC, the mean RSFC within the shape module (EB: paired t(15) = 10.650, p < 0.001; SC: paired t(15) = 9.563, p < 0.001) and within the conceptual module (EB: paired t(15) = 10.024, p < 0.001; SC: paired t(15) = 8.014, p < 0.001) were significantly stronger than the mean RSFC between the two network modules.

The left panel shows the RSFC results in the EB, and the right panel shows the RSFC results in the SC. (A, B) The significant seed-based RSFC results in the left ILOTC (A) and the right ILOTC (B) (vertex-wise FWE corrected p < 0.005, cluster size > 400 mm 2 ). Dots in colors denote the location of LOtv in three representative studies. (C) The mean RSFC matrix across participants in EB and SC among the shape- and conceptual-relevant brain areas. (D) Comparison among the mean RSFC among the shape-relevant regions (“Within Shape”), among the conceptual-relevant regions (“Within Conceptual”), and between the shape- and the conceptual-relevant regions (“Between Modules”). ***: p < 0.001. The underlying data for this figure can be found in S1 Data . EB, early blind; FWE, family-wise error; ILOTC, inferolateral occipitotemporal cortex; LOtv, lateral occipital tactile-visual; RSFC, resting-state functional connectivity; SC, sighted control.

We last used the seed-based RSFC to trace the regions having the neural activity synchronized with bilateral ILOTC (left ILOTC: Fig 6A ; right ILOTC: Fig 6B ; vertex-wise FWE corrected p < 0.005, cluster size > 400 mm 2 ). The ILOTC had strong RSFC to the other bilateral regions sensitive to the shape task—the aIPS, the pIPS, and the vPMC in both EB and SC. The left ILOTC in EB had stronger connectivity to the “visual” cortex than in SC ( S10 Fig ; vertex-wise p < 0.001, cluster-level FWE corrected p < 0.05).

S9 Fig illustrates the whole-brain searchlight results of conceptual association in the conceptual tasks across all participants (N = 48) (vertex-wise p < 0.001, cluster-level FWE corrected p < 0.05). The effects were mainly on bilateral dorsal AG, the left pIPS, the left precuneus, and the left dorsal medial prefrontal cortex. Given that some of the regions could also be spotted in the shape effect in the shape task ( S8A Fig ), they were likely to be driven by the task context [ 71 ].

We also used the RSA to investigate whether the brain areas sensitive to the conceptual task in the univariate analyses represented multivariate conceptual association ( Fig 5 ). Interestingly, although all these regions showed significantly stronger univariate activation in the conceptual task than in the shape task, only the left AG represented the conceptual association in the conceptual task across all the participants (orbital IFG: t(47) = 2.395, p = 0.021; aLTC: t(47) = −1.268, p = 0.211; pSTG: t(47) = −0.621, p = 0.537; AG: t(47) = 3.337, p = 0.002, SMG: t(47) = 1.174, p = 0.246; only the AG survived from multiple comparison correction, as Bonferroni corrected p < 0.05 for five ROIs is p < 0.01). The conceptual representation in the left AG was more evident in the conceptual task than in the shape task (paired t(47) = 2.163, p = 0.036), and no group differences were found between EB and SC (F(1, 30) = 0.192, p = 0.664).

We also investigated the multivariate object representation in other regions showing an enhanced univariate response to the shape task than the conceptual task. S4 to S6 Tables show the three-way mixed ANOVA results between groups (EB versus SC), tasks (shape versus conceptual task), and representations (shape similarity versus conceptual association) in bilateral aIPS, bilateral pIPS, and bilateral vPMC, respectively. They all had the same pattern, with a significant effect in representations and a significant interaction between tasks and representations. S7 Fig shows that all these regions represented shape similarity in the shape tasks (t(47): 5.531–10.074, ps < 0.001). Bilateral aIPS and pIPS also represented shape similarity in the conceptual tasks (t(47): 2.216–2.902, ps: 0.032–0.006), whereas shape representation in bilateral vPMC was not evident in the conceptual task (left: t(47) = 1.875, p = 0.067; right: (47) = 1.677, p = 0.100). Shape representation was more apparent in the shape task than in the conceptual task in all these regions (right vPMC: paired t(47) = 2.602, p = 0.012; other regions: paired t(47): 0.408–5.055, ps < = 0.001).

We averaged the neural RDMs of bilateral ILOTC across all participants (N = 48) and provided a planar visualization of the representational pattern using multidimensional scaling ( Fig 4D ). The color of words denoted the three clusters in the model RDM of shape similarity, mainly corresponding to elongated, round, and square objects. Representations of the three shape categories were separated in bilateral ILOTC.

We also investigated whether bilateral ILOTC in EB and SC share a matched shape representation ( Fig 4C ). By doing so, we measured the within-group coherence—the correlation between each participant’s neural RDM and the mean neural RDM of the other participants within the same group (i.e., EB-EB and SC-SC) and the between-group coherence—the correlation between each participant’s neural RDM and the mean neural RDM of all the other participants in the other group (i.e., EB-SC). A two-way ANOVA was performed between tasks (shape versus conceptual tasks) and group pairs (EB-EB versus SC-SC versus EB-SC). No significant interaction was found between tasks and group pairs (left ILOTC: F(2,90) = 1.366, p = 0.260; right ILOTC: F(2, 90) = 1.446, p = 0.241). There is a significant difference between tasks (left ILOTC: F(1,90) = 90.743, p < 0.001; right ILOTC: F(1, 90) = 75.809, p < 0.001), suggesting the shape task induced more coherent representations in bilateral ILOTC across participants. A weak effect in group pairs was also spotted in the left ILOTC (F(2,90) = 4.746, p = 0.011) but not in the right one (F(2, 90) = 1.065, p = 0.349). The post hoc comparison found that the mean value across levels of tasks was significantly different between SC-SC and EB-EB in the left ILOTC (Tukey’s test: p = 0.011), suggesting that the neural representation in the left ILOTC was more homogeneous in the SC group than in the EB group. However, there was no significant difference between EB-EB and EB-SC (Tukey’s test: p = 0.742) or between SC-SC and EB-SC (Tukey’s test: p = 0.073), suggesting no significant evidence showing a boundary effect between the neural representations across groups.

Fig 4B highlighted that the population without visual experience (i.e., the EB) showed a largely similar pattern. Bilateral ILOTC represented shape similarity in the shape task (left ILOTC: paired t(15) = 4.568, p < 0.001; right ILOTC: paired t(15) = 3.610, p = 0.003), whereas their shape representation in the conceptual task was less evident (left ILOTC: paired t(15) = 1.220, p = 0.241; right ILOTC: paired t(15) = 1.852, p = 0.084). The paired t test revealed a significant difference between the two tasks in the left ILOTC (paired t(15) = 3.361, p = 0.004) but not in the right ILOTC (paired t(15) = 1.466, p = 0.163). No evidence supported bilateral ILOTC represented conceptual association in either shape or conceptual tasks (t(15) < 1.282, ps > 0.219).

Fig 4A illustrates the RSA results in bilateral ILOTC across all participants (N = 48). Bilateral ILOTC represented shape similarity in both the shape task (left ILOTC: t(47) = 10.367, p < 0.001; right ILOTC: t(47) = 7.705, p < 0.001) and the conceptual task (left ILOTC: t(47 = 4.066), p < 0.001; right ILOTC: t(47) = 3.209, p = 0.002). The shape representation was stronger in the shape task than in the conceptual task (left ILOTC: paired t(47) = 5.183, p < 0.001; right ILOTC: paired t(47) = 3.776, p < 0.001). We found no clear evidence that bilateral ILOTC represented the conceptual association in either the shape or the conceptual tasks—only the conceptual effect in the left ILOTC in the conceptual task was marginally significant (t(47) = 2.123, p = 0.039). No significant difference was found in conceptual representation between shape and conceptual tasks (left ILOTC: paired t(47) = 0.558, p = 0.580; right ILOTC: paired t(47) = 0.395, p = 0.695).

The left column showed the neural representation in the left ILOTC. The right column showed the neural representation of the right ILOTC. (A) The RSA results across all participants (N = 48). (B) The RSA results in the EB (N = 16). (C) Inter-subject correlation between brain RDMs within and between the EB and the SC. (D) The MDS visualization of the mean brain RDM of the ILOTC across all participants (N = 48). ns: not significant, *: p < 0.05, **: p < 0.01, ***: p < 0.001. The underlying data for this figure can be found in S1 Data . EB, early blind; ILOTC, inferolateral occipitotemporal cortex; RDM, representational dissimilarity matrix; RSA, representational similarity analysis; SC, sighted control.

We then used RSA to investigate whether the ILOTC identified implemented shape representation ( Fig 4 , left and right panels corresponding to left and right ILOTC). A three-way mixed ANOVA was first performed between groups (EB versus SC), tasks (shape versus conceptual tasks), and representations (shape similarity versus conceptual association). The groups factor was between-subject, whereas tasks and representations were within-subject factors. In bilateral ILOTC, we only found a significant effect in representations and a significant interaction between tasks and representations ( Table 2 ).

Fig 3B and 3C shows the brain areas sensitive to the other object properties. The object size was mainly localized to the three scene-selective regions—the transverse occipital sulcus, the parahippocampal place area, and the retrosplenial cortex ( Fig 3B ). It has already been reported that these areas also prefer large nonmanipulable objects (e.g., [ 42 , 67 ]) and objects with a strong contextual association (e.g., [ 44 , 68 ]). Since the object size component here had a moderate correlation with the rating scores on contextual association ( Fig 1F ), we cannot distinguish between these two factors in this study. Moreover, we found a region in the left ventral and medial temporal cortex (mainly in the BA 20), of which the level of activity negatively correlated to touch experience ( Fig 3C ), suggesting this region was sensitive to the novelty of objects. We did not find any brain areas significantly modulated by toolness, which might result from the lack of typical tools (e.g., hammers or scissors) in the stimuli. Directly comparing the effects of all these parametric modulators between EB and SC also failed to reveal any significant brain regions.

When potential confounding factors were modeled, the difference between the two task types was still preserved ( Fig 3A ): The shape task activated bilateral brain areas, including the ILOTC, the aIPS, the pIPS, and the vPMC. The conceptual tasks mainly activated brain areas in the left hemisphere, including the orbital part of the inferior frontal gyrus (IFG) (i.e., BA 47), the aLTC, the posterior part of the STG (pSTG), the SMG, and the AG. These regions neatly matched the language network [ 48 , 49 ] with the absence of the triangular part of the IFG and the 55b region in the premotor cortex [ 63 ] (see the overlap in S5 Fig ), in line with previous studies suggesting these two dorsal regions play a non-semantic role in language processing (e.g., [ 64 – 66 ]). Since the brain clusters in Fig 3A were more discrete than those reported in the univariate contrast reported in Fig 2 (with no control for alternative object properties), we used the significant regions in Fig 3A to define the ROIs in the following analyses. No regions showed significant differences between EB and SC. We also found the same region in the cuneus when directly comparing EB and SC under a lower threshold (vertex-wise p < 0.001, uncorrected).

Dots in colors denote the location of LOtv in three representative studies. (A) Neural correlates of task types (the shape task coded as 1 and the conceptual task coded as −1). (B) Neural correlates of object size. Activations in the significant brain areas positively correlated with object size, i.e., larger objects induced higher activation. (C) Neural correlates of touch experience. Activations in the significant brain area negatively correlated with touch experience, i.e., less-touched objects induced higher activation. The underlying data for this figure can be found in S1 Data . FWE, family-wise error; LOtv, lateral occipital tactile-visual.

To investigate whether the other object properties modulated brain activity in the ILOTC, we conducted a parametric modulation analysis. The set of the parametric modulators included the task type (i.e., the shape task coded as 1 and the conceptual task coded as −1), the z-scores of the RT across all the trials in each run, the RCs corresponding to object size, toolness, touch experience, word duration, and word frequency. Fig 3 presents the significant brain areas encoding these parametric modulators (N = 48; vertex-wise p < 0.001, cluster-level FWE corrected p < 0.05).

S4A Fig illustrates the RT effect across the two tasks (N = 48; vertex-wise p < 0.001, cluster-level FWE corrected p < 0.05). As expected, it involved both frontoparietal and cingulo-opercular networks underlying top-down control [ 61 ]. It also involved regions in the default mode network, which could be because both the shape and the conceptual tasks require mental simulation [ 62 ]. Intriguingly, contrasting the RT effect between EB and SC revealed the lateral and ventral parts of the occipital cortex ( S4B Fig ; vertex-wise p < 0.001, cluster-level FWE corrected p < 0.05). These regions substantially overlapped with the LOC involved in visual shape perception in the sighted population, suggesting a functional reorganization of these regions in EB. Note that these regions did not overlap with the ILOTC.

Next, we directly contrasted the neural activity between EB and SC. As a sanity check, we first compared the brain activity level in shape and conceptual tasks to the resting state between EB and SC. As both tasks included auditory input, the occipital cortex in EB should show enhanced activation due to cross-modal neuroplasticity (e.g., [ 56 – 59 ]), and the results showed up as expected ( S3 Fig ; vertex-wise p < 0.001, cluster-level FWE corrected p < 0.05). We then compared the activity level between shape and conceptual tasks between EB and SC. We found only one significant region in the left cuneus near the parieto-occipital sulcus ( Fig 2D ; vertex-wise p < 0.001, cluster-level FWE corrected p < 0.05). ROI analysis showed that this region in EB had greater activation in the conceptual task than in the shape task (t(15) = −3.447, p = 0.004), whereas, in SC, it showed an opposite pattern (t(15) = 3.213; p = 0.006). This finding suggests that the earlier “visual” cortex in EB (i.e., the left cuneus) might repurpose itself to a similar role as what the language network played in the conceptual task (see also a recent meta-analysis [ 60 ]).

Both EB and SC activated the language network in the conceptual task. However, EB exhibited reduced left lateralization than SC. To measure the extent of lateralization, we extracted the T scores of the top 5% percentage of vertices showing the strongest activation in the contrast between the conceptual task and the shape task within the language network, which was anatomically defined in each participant’s native space by combing bilateral STG, bilateral inferior parietal cortices (i.e., the AG), and bilateral SMG in the DKT atlas [ 54 ]. The left lateralization was measured as (L − R)/(L + R), where L and R were the sums of T scores in the left and right hemispheres. While the SC had clear left lateralization (M = 0.381, SD = 0.344, t(15) = 4.440, p < 0.001), the EB’s lateralization was not evident (M = 0.143, SD = 3.397, t(15) = 1.442, p = 0.397). The paired t test showed a significant difference between the SC and the EB (paired t(15) = 2.452, p = 0.027), while no significant difference was found in handedness scores (SC: M = 76.875, SD = 20.238; EB: M = 73.750, SD = 16.279; paired t(15) = 0.543, p = 0.595). The reduced left lateralization for language processing in EB has been reported in a recent study and is still open to interpretation [ 55 ].

We then looked at the brain activation in EB and SC separately ( Fig 2B and 2C ). Both EB and SC had ILOTC activation in the shape task compared to the conceptual task. To confirm that the regions in the ILOTC found in the two groups were the same, we calculated the overlap coefficient, i.e., the area of the intersection region divided by the smaller area of the two regions. The overlap coefficient of the left ILOTC was 100%, i.e., EB’s ILOTC fell within SC’s ILOTC. The overlap coefficient of the right ILOTC was 81.5%. Consistent with the results pooling all participants (N = 48), SC also had significant activation in bilateral aIPS, pIPS, and vPMC in the contrast between shape and conceptual tasks ( Fig 2C ). Although these regions did not survive the multiple comparison correction at the whole-brain level in EB ( Fig 2B ), analyses using the significant areas in SC as regions of interest (ROIs) showed bilateral aIPS, bilateral pIPS, and the left vPMC in the EB also showed greater activation in the shape task than in the conceptual task ( S2 Fig ; left aIPS: t(15) = 3.486, p = 0.003; right aIPS: t(15) = 2.487, p = 0.025; left pIPS: t(15) = 2.478, p = 0.026; right pIPS: t(15) = 3.357, p = 0.004; left vPMC: t(15) = 2.632, p = 0.019; right vPMC: t(15) = 1.861, p = 0.083).

Fig 2A shows the contrast between the shape task and the conceptual task using all participants (N = 48). The shape task and the conceptual task involved dissociable brain networks. The shape task activated bilateral brain areas, including the ILOTC (i.e., the lateral part of the Broadman area (BA) 37), the aIPS, the posterior IPS (pIPS), the vPMC, and the inferior frontal sulcus. To verify whether the ILOTC activated in the shape task was the same region as the LOtv reported in previous literature, we projected the peak coordinates of the LOtv from three representative studies (i.e., [ 12 , 13 , 47 ]) to the brain surface and found that these coordinates largely fell over the geometric gravity center of the ILOTC region. In Fig 2A , we can identify two activity epicenters in the IPS—one was anterior and the other was posterior and joined to the intraoccipital sulcus.

Dots in colors denote the location of LOtv in three representative studies. (A) Shape versus conceptual tasks across all the participants. (B) Shape versus conceptual tasks in the EB. (C) Shape versus conceptual tasks in the SC. (D) Interaction between groups (EB vs. SC) and tasks (shape vs. conceptual). The error bars indicate the standard error. **: p < 0.01. The underlying data for this figure can be found in S1 Data . EB, early blind; FWE, family-wise error; LOtv, lateral occipital tactile-visual; SC, sighted control.

We first contrasted the neural activity level between the shape and conceptual tasks. To remove the domain-general RT effect, we modeled the trial-by-trial RT variability across the two tasks in the first-level general linear model (GLM) using both the variable epoch approach and the variable impulse approach [ 46 ]. Fig 2 illustrates results while the domain-general RT effect was controlled (vertex-wise p < 0.001, cluster-level family-wise error (FWE) corrected p < 0.05).

Table 1 shows the accuracy and reaction time (RT) across participants within each group in shape and conceptual verification tasks. All groups of participants had near-ceiling accuracy on both tasks. The shape verification task took about 130 to 200 ms longer than the conceptual verification task. We built a linear mixed model to predict the RT in the correct trials with groups of participants (EB versus SC) and types of tasks (shape versus conceptual tasks) as fixed effects variables and each participant as random effects grouping factors. The analysis revealed a significant task effect (F(1,30) = 73.055; p < 0.001), whereas the group effect was found insignificant (F(1,30) = 0.732, p = 0.399), and the interaction effect between groups and tasks had only a slight trend toward significance (F(1,30) = 2.552, p = 0.123). The significant difference between shape and conceptual tasks aligns with the evidence suggesting that retrieving specific semantic features (e.g., shape knowledge) requires more time than general semantic knowledge (i.e., function knowledge [ 45 ]). The interaction effect showed a weak trend that the shape task was slightly more difficult than the conceptual task for the EB than the SC ([(EB > SC) × (shape > conceptual tasks)]; z = 1.588, p = 0.112), which might be due to a lack of visual experience.

During the scanning, participants performed two tasks on the same set of auditorily presented words. In the shape verification task, participants thought carefully about the typical shape of each object and judged whether it was elongated, angular, hollow, circular, and disc-shaped. In the conceptual verification task, participants thought carefully about the primary function of each object and judged whether it was for eating, writing, sleeping, lighting, and purchasing (see S1 Fig and Procedures in Materials and methods for details).

We then averaged the rating scores across all participants (N = 48) to get a mean rating score vector for each rating item. Fig 1E illustrates the Z-scores of all the ratings across objects. To orthogonalize these unidimensional variables, we conducted the principal component analysis and applied varimax rotation to improve the interpretability of the resulting principal components. Five components had eigenvalues greater than 1. Fig 1F shows the correlation of these five rotated components (RCs) with each rating item. RC1 to RC5 corresponded to object size, toolness, touch experience, word frequency, and word duration, respectively (r(19): 0.915–0.981). The RCs corresponding to object size and toolness also had moderate correlations with the contextual association (r(19): 0.656 and 0.584). These RC scores were used in the subsequent parametric modulation analysis.

We averaged the above rating scores within each group of participants and evaluated the reliability of the mean rating score across participant groups. Fig 1B shows that the rating scores across three groups of participants were reliable (r(19) on objects size: 0.973–0.998; on contextual association: 0.732–0.940; on toolness: 0.883–0.933; on touch experience: 0.935–0.974). From this figure, we can also spot a moderate positive correlation between object size and contextual association (r(19): 0.363–0.529) and between toolness and contextual association (r(19): 0.264–0.622), which means the bigger the size, or the more likely an object is a tool, the more likely this object is bound to a specific context. Moreover, we also added two linguistic measures—word frequency (i.e., the Zipf value of the word occurrence in film and television subtitles; http://crr.ugent.be/subtlex-it/ ) and word duration. There was a moderate positive correlation between word frequency and touch experience (r(19): 0.419–0.446) and a moderate negative correlation between word frequency and word duration (r(19) = −0.577).

Besides the three object properties, all participants rated on a 7-point Likert scale about the degree to which they knew each object’s typical shape and primary function. Since most stimuli selected were everyday objects, both shape and conceptual rating scores hit the ceiling and varied only slightly across objects (averaged shape familiarity score across objects: M = 6.744, SD = 0.285; averaged conceptual familiarity score across objects: M = 6.944, SD = 0.066). Participants also rated how frequently they touched each object (1: have never touched it before; 7: touch it every day), which can be considered a sensitive and common index reflecting object familiarity across sighted and early blind groups. The inter-rater reliability on touch experience within each group of participants reached “excellent” (ICC(C,k): 0.965–0.975; S3 Table ).

Potential confounding factors were also considered. It has been reported that other properties of manmade objects can also modulate brain activity, like object size (big versus small; e.g., [ 42 ]), toolness (tools versus non-tool manmade objects; e.g., [ 43 ]), and contextual association (strong versus weak contextual association objects; e.g., [ 44 ]). These three variables were rated on a 7-point Likert scale (see Procedures in Material and methods about the rating procedure). S3 Table shows the inter-rater reliability within each group of participants. The inter-rater reliability reached “excellent” on object size (ICC(C,k): 0.979–0.992) and varied from “good” to “excellent” on toolness (ICC(C,k): 0.893–0.928). The inter-rater reliability on contextual association differed between sighted and early blind groups. While sighted groups had a “good” to “excellent” inter-rater reliability (SC: ICC(C,k) = 0.856; SI: ICC(C,k) = 0.919), the early blind group only had a “moderate” one (EB: ICC(C,k) = 0.613). Such heterogeneity in the early blind might result from a lack of instantaneous and global information about the environment from the visual input.

We then averaged the pairwise rating scores of all the participants (N = 48) and calculated the representational dissimilarity matrix (RDM) of shape similarity and conceptual association (i.e., 7 minus the mean rating score). The resulting two model RDMs had comparable variance across pairs of objects (shape similarity: variance = 2.163; conceptual association: variance = 2.498) and therefore offered equated discovery possibilities when correlated with brain RDMs in the subsequent RSA. Fig 1C and 1D show the organizational structure of the two RDMs, where 21 items were grouped according to the clusters generated by the k-means clustering algorithm [ 38 , 39 ], with the silhouette criterion used to decide the optimal number of clusters [ 40 ]. The shape similarity RDM fell into three clusters, corresponding to square, round, and elongated objects ( Fig 1C ). The conceptual association RDM fell into seven smaller clusters, corresponding to different occasions in which objects were used ( Fig 1D ). For example, the two biggest clusters were related to eating and writing. The conceptual rating results accorded closely with the teleological perspective, which suggests the essence of a manmade object lies in its function, not its physical properties (e.g., [ 41 ]).

Shape similarity and conceptual association were rated on a 7-point Likert scale in a pairwise manner (see Procedures in Material and methods about the rating procedure). We assessed the inter-rater reliability within each group of participants using the intraclass correlation based on a mean-rating, consistency, two-way random model (i.e., ICC(C,k)) [ 36 ]. Both shape rating (ICC(C,k): 0.953–0.973) and conceptual rating (ICC(C,k): 0.984–0.985) showed “excellent” inter-rater reliability [ 37 ] ( S3 Table ). We averaged the rating scores within each group and compared them across groups. Fig 1A illustrates that the rating scores on both object properties were highly reliable across three groups (r(208) on shape similarity: 0.957–0.983; on conceptual association: 0.982–0.984), and the pairwise shape similarity was orthogonal to the pairwise conceptual association (r(208): 0.103–0.132).

(A) Correlation between ratings on pairwise shape similarity and pairwise conceptual association across three participant groups (EB: early blind, SC: sighted control, IS: independent sighted). (B) Correlation among linguistic variables and ratings on other object properties across three participant groups. (C) Pairwise ratings on shape similarity (i.e., the shape similarity RDM). (D) Pairwise ratings on conceptual association (the conceptual association RDM). (E) Linguistic variables and ratings on other object properties. (F) Correlations between the first five RCs and linguistic variables and ratings on other object properties. The underlying data for this figure can be found in S1 Data . RC, rotated component; RDM, representational dissimilarity matrix.

In this study, we selected 21 Italian words, which referred to 21 manmade objects, as our stimuli. The selection was mostly based on behavior ratings of object properties from an independent group of sighted participants who did not take part in the fMRI experiment (N = 19; see Stimuli in Materials and methods about the stimulus selection procedure and criteria). To validate the rating results from the stimulus selection stage and to verify whether the early blind population had a similar shape and conceptual knowledge as the sighted control, all participants who took part in the fMRI experiments (N = 48) also rated the object properties of the stimuli selected ( Fig 1 ). These participants consisted of three groups: 16 early blind (EB) participants, 16 gender- and age-matched sighted control (SC) participants, and 16 independent sighted (IS) participants (see Participants in Materials and methods for details).

Discussion

Our study investigated where and how shape representations are stored in the brain and distinguished from the conceptual representation of the same manmade objects. By testing early blind participants, we assessed whether occipital regions implement shape representation independently of visual experience/imagery (e.g., [1–3]) or, alternatively, whether the “visual” cortex would repurpose its function for conceptual representation due to early visual deprivation [32]. We found that bilateral ILOTC, a region that overlaps with the LOtv [12,13,47], together with bilateral aIPS, pIPS, and vPMC, showed greater activation when people processed shape rather than conceptual attributes of the same objects, and their activity pattern encoded shape similarity but not conceptual association among objects. In contrast, regions in the left perisylvian area, including the orbital IFG, the aLTC, the pSTG, the AG, and the SMG, showed greater activation in the conceptual task than in the shape task. RSFC analysis further demonstrated that shape- and conceptual-relevant regions formed distinct brain networks. Interestingly, in all the above results, visual experience had little influence—EB and SC had similar activity profiles and connectivity patterns.

Our results thus favor the hypothesis suggesting the ILOTC implements supramodal shape representation and argue against the alternative hypotheses that such activation depends on visual imagery or conceptual associations based on functional relevance. These results echoed various perspectives suggesting object representation in the brain is organized according to properties, not modalities (e.g., [3,72,73]).

In contrast to the view that ILOTC implements supramodal shape representation, one could argue that this region might represent visual shapes in the sighted and haptic shapes in the early blind. Testing this possibility using fMRI is challenging as it is difficult to distinguish supramodal representation and visual representation derived from visual imagery triggered by touch in the sighted participants. One option would be to examine whether sighted patients with bilateral lesions in the ILOTC have both visual and tactile shape agnosia or only visual shape agnosia. Unfortunately, the two existing cases of bilateral ILOTC lesions cannot convincingly answer this question. One case is patient D.F., who had bilateral lesions in the LOC [10] and had both visual and tactile agnosia [74]. However, D.F. also had bilateral lesions to the parieto-occipital cortex [75] and her tactile agnosia might result from parietal damage. The other case is patient M.C., who had bilateral lesions in the LOC, including the LOtv [15]. Unlike D.F., M.C. only had visual agnosia, and her tactile recognition ability was fast and accurate. However, although the haptic shape task did not activate the ILOTC of M.C. due to lesions in this region, it activated a nearby region in the posterior middle temporal gyrus. Such activation might reflect post-lesion reorganization, compensating for the shape representation that should be implemented in the ILOTC [15]. Besides resorting to rare patient cases, another seemingly plausible option would be selective transcranial magnetic stimulation (TMS) over bilateral ILOTC to evaluate whether it interferes with both haptic and visual shape tasks or only visual shape tasks. However, the pitfall is that even if TMS over bilateral ILOTC does disrupt haptic shape tasks (e.g., longer RT), such disruption might be mediated by the disruption in visual imagery, a strategy sighted participants would adopt to facilitate haptic shape tasks (e.g., [21,76]).

While conclusive proof is still warranted, there is other evidence supporting the role of ILOTC in supramodal shape representation in the sighted population. On the one hand, the ILOTC (mainly in BA 37) is anterior to the lateral occipital cortex (LO, mainly in BA 18), a visual shape perception region representing shape features like curvatures and medial axes, in contrast to the earlier visual cortex implementing retinotopic representation (e.g., silhouettes) (e.g., [77,78]). According to embodied semantic theories [79] and the “anterior shift” phenomenon [80], the associative cortex anterior to each sensorimotor area can gradually capture the regularities of the activity patterns in its nearby sensorimotor cortices induced by different exemplars of the same concept (e.g., different exemplars of an apple) and generate a schema-like representation as the sensorimotor knowledge of that concept (e.g., the typical color, shape, and action related to an apple). In line with this hypothesis, previous studies have shown that the region representing objects’ color knowledge is localized to the fusiform gyrus anterior to the color perception area in V4 [81,82], and language-induced category-specific activations are aligned with but anterior to the visual-induced activations of the same semantic category [83]. The ILOTC, which is anterior to the LO, thus possibly represents objects’ shape knowledge—the schematic or prototypic shape of an object concept—derived from various concrete shape exemplars represented in the LO of the sighted people.

On the other hand, the ILOTC was strongly connected to the IPS and the vPMC (Fig 6), a frontoparietal circuit that has long been proposed to be involved in hand configuration to grasp objects in particular shapes in light of single-neuron recording evidence (see reviews [84,85]). Neuropsychological evidence confirms that lesions in the aIPS can induce both tactile shape agnosia [16] and tactile apraxia [17], and lesions in the vPMC can lead to syndromes resembling tactile apraxia [18]. Our study found that the IPS-vPMC circuit implemented shape representation even in the early blind population with no visual experience (S2 Fig, S4–S6 Tables), further demonstrating that haptic sources alone can form the shape representation in these regions.

Converging the two groups of evidence described above—the position in the ventral visual pathway and the connection to the frontoparietal haptic circuit, it appears parsimonious to postulate that the ILOTC act as an operator bridging visual and haptic shape representations. Given this supramodal nature, the ILOTC might not only schematize the visual shape representation from the LO but also integrate the haptic shape representation from the IPS-vPMC circuit by amplifying the “affordance” shape features utilized for object grasping. This hypothesis is supported by the evidence that the ILOTC is more sensitive to pictures of graspable tools over non-graspable manmade objects (e.g., [86]), and its activity pattern better reflects the shape of objects’ handles than their functional parts [87]. Nevertheless, the shape representation in the ILOTC is essentially sensorimotor-derived and would still be in the analogical format, in contrast to the amodal symbolic format usually supported by the language system (e.g., the symbol of a “ring” associated with the symbol of “round”; similarly as the way the early blind represent color knowledge [53,88–90]).

As for the neural representation of functional knowledge, contrasting the conceptual task with the shape task revealed the left perisylvian regions related to linguistic processing (Figs 2 and 3), implying that function knowledge is supported by the language system. This result is supported by a recent massive study with 136 acute left hemisphere stroke patients [91]. They found that the deficit in tool selection (e.g., choosing the nail for the hammer) was specifically related to lesions in the left perisylvian regions, mainly including the whole length of the lateral temporal lobe and the anterior IFG. The language system might provide a symbolic format of representations, which can better capture the abstract “associations” among holistic concepts. It contrasts with the analogical format of representation grounded in the sensorimotor system (as discussed for the shape representation in ILOTC), which can better reflect the “similarity” in one particular semantic feature. Such findings suggest that function is not an explicit object property, which can be directly derived from sensorimotor experience—we cannot reduce an object’s function to what it looks like and how it is manipulated; it must therefore rely on some sort of abstract/linguistic coding.

The differences between these two neural coding mechanisms may explain the discrepancy in the RSA results between shape similarity and conceptual association. RSA assumes that the representational content can be inferred from the distributed activity pattern across cortical surfaces. The most definitive evidence supporting this assumption comes from the primary sensorimotor system following a topographic organization (e.g., retinotopy), where the input and output information is transparently projected to the cortical surface. Since the shape representation (e.g., curvatures and medial axes) is transited and abstracted from the topographic representation [77,78], the activity pattern across the cortical surface in the shape-relevant regions would still be informative. However, in the linguistic system, the representation is presumed to be coded in the format of “arbitrary” symbols, where the linguistic sign (e.g., word forms) bears no obvious resemblance to the content signified. The content represented in the language system thus is not directly transited or abstracted from the word form representations in the sensorimotor cortex and might not be transparently reflected on the activity pattern across the cortical surface. Our results confirmed this hypothesis. Whereas all the shape-relevant regions defined by the univariate contrast encoded the shape similarity among objects (Figs 4 and S7), among the conceptual-relevant regions defined by the univariate contrast (i.e., aLTC, pSTG, SMG, AG, and orbital IFG), only the activity pattern in the AG was correlated to the conceptual association RDM (Fig 5).

The exception for the AG indicates that conceptual associations can be represented in a format other than linguistic. Previous studies have shown that the AG is not a purely linguistic region but also part of the default mode network engaged in memory-based simulation (e.g., [51,52,62,92]). Compared to the other linguistic regions, the AG is less responsive to word forms (e.g., [93]) but more sensitive to the retrieval of multimodal episodic memories (e.g., see review [94]). It is thus possible that the AG codes thematic relations based on the spatiotemporal continuity in our sensorimotor experience (e.g., hammers and nails often co-occur; e.g., [95]), which is apt to reflect on activity patterns [96], in contrast to the coding based on linguistic associations in the other language areas. In line with this idea, we found other brain regions in the default mode network also representing conceptual associations in the whole-brain searchlight RSA analysis, including the left precuneus and the left dorsal medial prefrontal cortex (S9 Fig).

It is worth noting that this study only focused on one type of conceptual knowledge: thematic relations or conceptual associations based on functional knowledge (“plate” and “fork” used for dining versus “pillow” used for sleeping). The conclusion, therefore, cannot extend to other knowledge types like taxonomic categories (“plate” and “pillow” as manmade objects versus “dog” as animals). In the “Stimuli” section, we elaborated on why we narrowed down the stimuli to one taxonomic category (i.e., the manmade objects) and only focused on thematic relations. One of the main reasons is that it is challenging for a neuroimaging study to distinguish whether a brain area represents taxonomic knowledge per se or simply shows a preference for the features of specific categories. Take the ILOTC as an example. Previous studies have shown that regions overlapping or superior to the ILOTC are more sensitive to manmade objects than the other categories, even in early blind participants (e.g., [25–28]). However, this does not necessarily mean the ILOTC represents taxonomic knowledge at the conceptual level. Instead, our results suggest that the ILOTC represents shape knowledge derived from both visual and haptic modalities, and its preference for manmade objects is likely perceptual. Either because people have more haptic experiences with manmade objects or the manmade objects have more affordance shape features for grasping, the ILOTC can receive additional shape information from haptic modalities and therefore become more sensitive to the shapes of manmade objects (see previous discussions).

Our study also reveals crucial neuroplastic principles about how the “visual” cortex reorganizes its function after vision loss. In the high-order visual cortex, where brain areas receive not only visual input but also information from other sensorimotor systems, brain functions are likely to be resilient to vision loss through compensation. The most well-documented example is the region hMT+/V5, a highly specialized area for visual motion processing. This region also has a direct white matter connection to the planum temporale specialized in auditory motion processing [97] and preferentially responds to moving auditory and tactile stimuli in the early blind (e.g., [98–102]). Our results reveal the ILOTC has a similar nature—it had strong connections to the frontoparietal regions involved in haptic processing and preserved its functionality despite the lack of visual input (Figs 2 and 4).

In contrast, in the more primary visual cortex, where visual input is dominant, vision loss will leave a functional vacancy that would be difficult for another sense to fill in. Higher-order brain systems might have the opportunity to take over, pushing for a more radical functional repurposing in those early visual regions. This hypothesis is supported by neuroimaging studies showing that part of the “visual” cortex of the early blind is sensitive to linguistic components (semantics and syntax; e.g., [34,60,103,104]) and mathematical difficulty [105]. In line with these findings, we found that the left cuneus in EB showed greater activation to the conceptual task than the shape task, whereas the same conceptual preference can only be observed in the left perisylvian language areas in SC (Fig 2). Similarly, the lateral occipital cortex and the posterior fusiform gyrus in EB—two “earlier” regions along the visual processing stream than the ILOTC—showed a domain-general RT effect, which is typically observed in the frontoparietal and cingulo-opercular areas in SC (S3 Fig).

To conclude, our study identified dissociable brain networks representing objects’ shape and conceptual knowledge. The bilateral ILOTC-IPS-vPMC circuit represented shape knowledge, and the left perisylvian circuit related to language processing represented conceptual knowledge. Relying on data collected in EB, we highlighted that the ILOTC represented shape knowledge independently of visual experience. We argue that the ILOTC implements a supramodal shape representation by virtue of its position in the ventral visual pathway and its strong connections to the IPS-vPMC circuit involved in haptic processing, and such sensorimotor-derived representation differs from the disembodied representation supported by the language system in their representational formats.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001930

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.