TY - JOUR
AU1 - Machizawa, Maro, G
AU2 - Driver,, Jon
AU3 - Watanabe,, Takeo
AB - Abstract Visual working memory (VWM) refers to our ability to selectively maintain visual information in a mental representation. While cognitive limits of VWM greatly influence a variety of mental operations, it remains controversial whether the quantity or quality of representations in mind constrains VWM. Here, we examined behavior-to-brain anatomical relations as well as brain activity to brain anatomy associations with a “neural” marker specific to the retention interval of VWM. Our results consistently indicated that individuals who maintained a larger number of items in VWM tended to have a larger gray matter (GM) volume in their left lateral occipital region. In contrast, individuals with a superior ability to retain with high precision tended to have a larger GM volume in their right parietal lobe. These results indicate that individual differences in quantity and quality of VWM may be associated with regional GM volumes in a dissociable manner, indicating willful integration of information in VWM may recruit separable cortical subsystems. attention, event-related potential (ERP), neural correlates, visual short-term memory, voxel-based morphometry (VBM) Introduction It has been widely believed that the capacity for retaining items in mental representation is a crucial factor that constrains visual working memory (VWM) (Cowan 2001; de Fockert et al. 2001; Engle 2002; Vogel et al. 2005). Several studies have agreeably revealed that the number of items that can be successfully retained in VWM reaches an asymptote while the set size is increased, with characteristically different maxima being achieved by different individuals. Individuals’ VWM capacity is approximately three items on average, but this capacity substantially fluctuates across participants, from just one or two items for low-capacity individuals up to approximately six or seven items for high-capacity individuals (Luck and Vogel 1997; Cowan 2001; Vogel and Machizawa 2004; Xu and Chun 2006; Vogel and Awh 2008; Luck and Vogel 2013). Such evidence supports a concept that each person has a unique VWM capacity limited by a certain number of discrete memory slots, which would primarily favor a “quantitative” account of VWM capacity. However, other studies (Wilken and Ma 2004; Tombu and Jolicœur 2005; Bays and Husain 2008; Ma et al. 2014) have proposed an alternative hypothesis that VWM is limited by flexible memory resources rather than a fixed number of memory resources. These studies have instead suggested that VWM constitutes a limited resource that can be flexibly allocated to a large number of items with less precision or a smaller number of items with more precision. That is, VWM may be limited by the precision with which items are retained, rather than by its capacity, which would primarily favor a qualitative account. Thus, there has been considerable controversy in whether VWM is constrained by the capacity or precision with which items are retained (Bays and Husain 2008; Zhang and Luck 2008; Fukuda et al. 2010a; Mance and Vogel 2013; Ma et al. 2014). To date, researchers have proposed various “neural” indices for VWM maintenance, including electrophysiological markers using sustained components (Vogel and Machizawa 2004; McCollough et al. 2007) or oscillatory activities (Sauseng et al. 2009; Fukuda et al. 2015) observed on electroencephalography (EEG) or magnetic encephalography (Moran et al. 2010; Robitaille et al. 2010). These neural markers covaried with a set size of targets in a visual change-detection WM task and asymptoted in a manner that correlated with individual differences in behavioral WM capacity. In independent yet potentially related “functional” magnetic resonance imaging (fMRI) work, the blood-oxygen-level-dependent signal in the parietal cortex also showed a similar activation pattern (Todd and Marois 2004, 2005; Xu and Chun 2006). These studies highlighted “functionally” relevant key brain regions within the posterior parietal cortex and lateral occipital cortical brain areas with activations in these regions patterns correlated with individual differences in VWM that favored a quantitative account. Other researchers have claimed that these neural responses may represent putatively intermixed or qualitative accounts (Zhang and Luck 2008; Murray et al. 2011; Machizawa et al. 2012; Ma et al. 2014) and have suggested that each functionally relevant brain region may have a distinct role, depending on the visuospatial feature of interest (Xu and Chun 2006, 2009; Matsuyoshi et al. 2012). These studies with fMRI have found that among the many cortical brain regions associated with the maintenance of visual information in mental representation (Pessoa et al. 2002; Xu 2017), an activation pattern of the posterior parietal and lateral occipital brain regions played key roles in individual differences in VWM capacity. For example, activations within the intraparietal sulcus (IPS) and lateral occipital complex (LOC) tracked the number of objects being maintained in memory (Todd and Marois 2005; Xu and Chun 2006), whereas the dorsal occipital (alternatively, the inferior part of the IPS) seemed to track a fixed number of objects in memory load (Xu and Chun 2006, 2009; Matsuyoshi et al. 2012). In the last decade, another trend in neuroscience research has emerged relating one’s behavioral abilities with neuroanatomical features such as cortical or subcortical volume (Kanai and Rees 2011) using the so-called voxel-based morphometry (VBM) (Ashburner and Friston 2000). VBM is a measure of the overall volume of each subregion (consisting of a group of voxels) reflecting regional volumetric differences between populations or across individuals (Ashburner and Friston 2001). While little evidence supports the notion that the volume of a single brain region predicts one’s VWM capacity, one study reported that the gray matter (GM) volume of the insula was “positively” correlated with a behavioral measure of VWM capacity for object identity, whereas the volume of the parietal cortex was positively associated with individual variability in the spatial aspects of VWM (Konstantinou et al. 2017). Based on the uncorrelated nature of behavioral performance in conditions with a focus on item precision or number of representations, previous studies have proposed that the quantitative and qualitative aspects of VWM may be dissociable (Fukuda et al. 2010b; Machizawa and Driver 2011). In addition, a study showed that a neural marker for VWM maintenance equally responded in both conditions in which only a few items were retained with fine precision and conditions in which a larger number of items were maintained with coarse precision (Machizawa et al. 2012). To date, functional neural markers that reflect transient neural resource consumption dedicated to VWM, or “states” of mental representation, have been well studied. Nevertheless, relatively little is known about the neuroanatomical basis for one’s “trait” of VWM abilities that may reflect individual differences in VWM, particularly with a focus on the precision or number of representations that are being held. In this article, we report our attempt to establish neuroanatomical correlates of capacity and precision of VWM, as well as to examine whether the cortical anatomy of the same brain region may relate to individual differences in both aspects of VWM representations. First, we examined behavior-to-brain relations in Experiment 1 using different tasks, each of which focused on either the number or precision of items being maintained. In Experiment 2, by controlling for the visual feature and particular task, we tested a capacity-focused and precision-focused way to retain items within the same task while monitoring the direct neural activities of the participants with EEG. Across the two sets of experiments, both the behavioral and neural indices of VWM were correlated with the individual’s cortical anatomy to examine the behavior-to-brain anatomy relations as well as neural activity to brain anatomy relations. In this manuscript, we define “capacity” as the retention of items with a quantitative focus by trying to remember a large number of items and “precision” as the retention of items with a qualitative focus by trying to remember the details of items. It has been hypothesized that given a resource for VWM solely constrains one’s ability regardless of the intent, the GM volume of the same brain region would correlate with the performance for both aspects of VWM. In contrast, given putatively dissociable functions of the qualitative and quantitative information integration processes, an alternative hypothesis was that individual differences in brain structures of distinct brain regions might independently correlate with the two aspects of VWM. Materials and Methods for Experiment 1 Experiment 1 examined the capacity and precision of VWM abilities in two different tasks, both of which were previously shown to assess VWM with different intent. We examined whether there was any cortical anatomy, particularly regional GM volume, associated with the capacity or precision of VWM. Specifically, we regressed individuals’ behavioral performances regarding the VWM capacity or precision to regional GM volumes across individuals. The procedures of the behavioral tasks in Experiment 1 were identical to a subset of those used in a previous study (Machizawa and Driver 2011). Participants Forty-nine participants (31 females, aged between 19 and 35 years old with a mean age of 25.9 years) were recruited from a local community in London, United Kingdom. All participants gave written informed consent that was approved by the UCL Research Ethics Committee in accordance with the Declaration of Helsinki. The same participants performed the two tasks, with the order counterbalanced across participants. Capacity Task The capacity of VWM was tested by a color change–detection task that was primarily based on a widely used task (Vogel and Machizawa 2004; see Fig. 1a). Each trial started with the presentation of two arrows. The participants were informed that these arrows indicated the side of the visual field in which task-relevant items would be presented later in the trial (cued side). After an interval of 300–500 ms following the offset of the arrows, a sample array was presented. The array consisted of 2, 4, 6, or 8 squares in different colors in each visual hemifield. Participants were asked to remember as many colors of the squares on the cued side as possible. After a 900-ms interval following the end of the sample presentation, a probe was presented. The task required participants to indicate whether there were any color differences between the sample array and the probe in the cued side. For both capacity and precision tasks, there was a total of 64 trials (for both left and right hemifield conditions). See Experiment 1 of Machizawa and Driver (2011) for further details. Figure 1 Open in new tabDownload slide Schematics of the behavioral task procedures for assessing the VWM capacity and precision and the results from Experiment 1. (a) VWM quantity task in Experiment 1. The number of presented items was varied from two to eight in each visual hemifield in both the sample and probe arrays. Participants were asked to indicate whether there was a discrete change in the color of one item between the sample and probe arrays only within a cued hemifield. This color change–detection task required the retention of multiple colors with only coarse precision. (b) VWM quality task in Experiment 1. The number of items was always two in each visual hemifield of the sample display. Participants were asked to indicate whether the item in the probe array was rotated clockwise or counterclockwise (varied from ±15° to ±60°) from the item in the same location of the sample array. This orientation change–discrimination task required retention of only two items but necessitated discrimination of the fine precision of orientation. (c) The GM volume was significantly correlated with VWM quantity (red blob) and VWM quality (blue blobs). Highlighted areas were thresholded at Puncorrected = 0.005 for display purposes. (d) Two scatterplots of GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of the behavioral index of VWM for the quantity task (K). Gray lines are fitted trend lines. The GM volume in the left LOC was positively correlated with the behavioral measure of VWM quantity but not with quantity (n.s.). (e) Two scatterplots of GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of the behavioral index of VWM for the quality task (K). The GM volume in the right IPS was positively correlated with the behavioral measure of VWM quality (right panel) but not with quantity. Figure 1 Open in new tabDownload slide Schematics of the behavioral task procedures for assessing the VWM capacity and precision and the results from Experiment 1. (a) VWM quantity task in Experiment 1. The number of presented items was varied from two to eight in each visual hemifield in both the sample and probe arrays. Participants were asked to indicate whether there was a discrete change in the color of one item between the sample and probe arrays only within a cued hemifield. This color change–detection task required the retention of multiple colors with only coarse precision. (b) VWM quality task in Experiment 1. The number of items was always two in each visual hemifield of the sample display. Participants were asked to indicate whether the item in the probe array was rotated clockwise or counterclockwise (varied from ±15° to ±60°) from the item in the same location of the sample array. This orientation change–discrimination task required retention of only two items but necessitated discrimination of the fine precision of orientation. (c) The GM volume was significantly correlated with VWM quantity (red blob) and VWM quality (blue blobs). Highlighted areas were thresholded at Puncorrected = 0.005 for display purposes. (d) Two scatterplots of GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of the behavioral index of VWM for the quantity task (K). Gray lines are fitted trend lines. The GM volume in the left LOC was positively correlated with the behavioral measure of VWM quantity but not with quantity (n.s.). (e) Two scatterplots of GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of the behavioral index of VWM for the quality task (K). The GM volume in the right IPS was positively correlated with the behavioral measure of VWM quality (right panel) but not with quantity. As a behavioral index of VWM capacity in the capacity task, Pashler’s K estimates (Pashler 1988; Rouder et al. 2011) at set sizes of 6 and 8 were averaged. The estimated VWM capacity (K) was defined as N*(H−F)/(1−F), where H and F correspond to hit and false alarm rates, respectively, and N is the number of target items. The K estimate can potentially range between 0 and 7, indicating an estimated “number” of objects one can maintain in VWM. Precision Task The precision of VWM was examined using an orientation discrimination task that was primarily based on a widely used task (Bays and Husain 2008; see Fig. 1b). Each trial started with the presentation of two arrows, which were followed by a 300–500-ms interval, as in the abovementioned color change–detection task. Then, two oriented bars were presented on each side of the visual field. The orientations of these bars were randomly chosen from a total of 36 orientations (in 5° steps, while avoiding perpendicular orientations). After a 900-ms interval following the offset of the sample array, one bar was presented at the same location as one of the previously presented bars but only on the cued side from the sample array. The degree was varied randomly, one out of 8 alternative cases (+15, −15, +30, −30, +45, −45, +45 or −45), but not 8-degree steps. The participants were asked to indicate whether the bar was rotated clockwise or counterclockwise, compared with the one at the same location in the sample array. As a behavioral index of VWM precision, the proportion correct was taken only from the results of the most challenging trials (±15° difference between the bars in the sample array and probe). The order of the two task conditions was counterbalanced across participants. See Experiment 2 of Machizawa and Driver (2011) for details. The proportion correct in the most difficult (15° change) condition was multiplied by 2 (the number of target items) and used as an outcome in the precision task. The precision estimate could have ranged between 0 and 2. As an index for one’s precision of VWM, it is common to obtain the reciprocal error of a fitted Gaussian curve across degree-change conditions (Bays and Husain 2008). We calculated both proportion-based and reciprocal error-based measures. Because we aimed to regress the outcome against cortical GM volume in a parametric manner, it was essential to run the analysis on normally distributed scores. As it turned out, the distribution of the error-based precision measure was not normally distributed (Shapiro–Francia W = 0.84, P < 0.001), whereas the scores based on the proportion correct were normally distributed across participants (Shapiro–Wilk W = 0.97, P = 0.36, n.s., not significant). Therefore, we selected the results from proportion correct as a covariate of interest instead of the scores based on reciprocal error. Notably, scores obtained by the proportion correct and reciprocal error were strongly correlated with each other (Spearman ρ2 = 0.72, P < 0.001), confirming the scores based on proportion correct as an appropriate alternative index of precision of VWM for the sake of examining individual variability here. This reasoning was further supported because obtaining proportion correct measures could be transferred to the measures in Experiment 2 (see below for details). VBM Analysis The observed behavioral indices of VWM abilities were regressed to individuals’ GM volume measures. To assess GM volumes, all participants underwent an extra session for an MRI scan. Structural MRI scans were performed in a 1.5 T whole-body MRI scanner (Siemens Sonata) up to 2 months prior to the behavioral tasks. A 3D modified driven equilibrium Fourier transform sequence (TR = 12.24 ms; TE = 3.56 ms; field of view = 256 × 256 mm; and voxel size = 1 × 1 × 1 mm) was used to acquire MR images in sagittal sections with 1-mm isotropic resolution (176 sections; 256 × 240 mm2). Structural images were processed by an optimized VBM method (Wright et al. 1995; Ashburner and Friston 2000; Ashburner 2007). We conducted VBM analyses (Wright et al. 1995; Ashburner and Friston 2000, 2001) including optimized normalization with Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra (Ashburner 2007) with SPM 8 (Wellcome Trust Centre for Neuroimaging: www.fil.ion.ucl.ac.uk/spm/). In the analysis, the voxel size was 1.5-mm cubic. The initial height threshold was set at Puncorrected < 0.005, and the extent threshold was set at 26 voxels, roughly corresponds to 4.5-mm cubic. Based on previous fMRI studies of VWM, bilateral superior intraparietal sulci and bilateral LOC were selected as regions of interest (ROIs), as these areas have been implicated in both WM storage capacity and possibly WM precision (Todd and Marois 2004; Xu and Chun 2006). It was assumed that the cortical volume of the brain regions that correlated with individual differences in VWM scores should also fall within one of the functionally relevant brain regions. Therefore, particular ROIs were selected based on previous fMRI studies on VWM maintenance. Some studies have reported that VWM contents may be decoded in early visual areas, such as V1 (Kamitani and Tong 2005; Emrich et al. 2013), and found a structural association with the precision of VWM (Bergmann et al. 2016). However, the decoding of VWM contents may not necessarily be a source of VWM resources but rather reflect reactivation guided by feedback signals from higher-order brain regions that predominantly play a key role in the maintenance of VWM (Xu 2017). Our primary focus was the higher-order brain regions rather than the early visual cortex. We selectively restricted the analysis to ROIs having a particular function for the maintenance of WM contents over the delay period. Likely, prefrontal brain regions could have been a target as a putative origin of top-down signals during the execution of commands; however, the frontal ROIs were not considered our primary aim regarding the maintenance of VWM. The bilateral superior intraparietal sulci (MNI coordinates, left/right, −21/23, −70/−56, 42/46) and bilateral LOC (left/right, −44/42, −71/−69, 5/0) were selected as was particularly implicated in both the capacity and precision of VWM (cf., Xu and Chun 2006). Functional activations in the parietal and occipital regions were measured while participants were performing a change-detection task with stimuli presented at fixation or in the periphery. Activations in the superior IPS and LOC were correlated with individual differences in VWM capacity (Xu and Chun 2006). In the report, coordinates for the “off-center” and the “centered (at fixation)” presentation conditions slightly differed. Because all target stimuli were presented in the periphery in our tasks, our ROIs were selected from the off-center conditions in Xu and Chun (2006). To supplement our approach of identifying independent behavior-to-brain associations (separately regressing both measures rather than combining the two), we also performed a conjunction analysis to seek mutually related brain regions (see Supplementary Results). Results for Experiment 1 Behavioral Results There were no significant mean differences in behavioral performance between the two hemifields (left and right visual fields) either for capacity (t48 = 0.996; P = 0.324, n.s.) or for precision (t48 = 0.235; P = 0.815, n.s.). Behavioral accuracies (K-value) of the set sizes 2, 4, 6, and 8 conditions of the capacity task were 1.85 ± 0.16, 2.68 ± 0.68, 2.51 ± 1.17, and 2.91 ± 1.59. Proportion correct measures of the precision task at 15°, 30°, 45°, and 60° were 0.72 ± 0.10, 0.80 ± 0.10, 0.78 ± 0.10, and 0.70 ± 0.10. As for the precision task, precision (deg−1) was also calculated for each participant, as in Bays and Husain (2008). The mean and standard deviation (SD) of precision was 0.03 ± 0.02, within a good agreement with the previous report at set size 2 for the orientation task as reported in Bays and Husain (2008), cf., their Figure 2b right panel. Since we found significant correlations between behavioral performance in the two hemifields both for capacity (r49 = 0.608; P < 0.001) and for precision (r49 = 0.819; P < 0.001), the averaged behavioral scores for the two hemifields were regressed (see hemisphere-specific results in Supplementary Results and Supplementary Figure 1 that shows the left LOC equally related to the behavioral scores in both visual fields). Figure 2 Open in new tabDownload slide Task procedures and results from Experiment 2 relating behavioral performance to precision and capacity of VWM. (a) VWM quantity condition. Only the target item in the probe display did not change color, while the rest of the items turned black. Participants were asked to indicate whether the orientation of the target item in the probe array was rotated clockwise or counterclockwise relative to the item at the same location in the sample array. The orientation difference between the target item was ±45°. The set size in both the sample and probe arrays was 4 items in each visual hemifield in both the sample and probe arrays. The condition thus required a low load on discrimination (coarse precision) but a high load on retention of multiple items (capacity). (b) VWM quality condition. The same paradigm as in the quantity condition was used, except that the set size was 2 items in each visual hemifield and the orientation difference between the target item in the probe array and the item at the same location in the sample array was ±15°. Therefore, this condition required a high load regarding the discrimination but a low load on the retention of multiple items. Brain regions in which GM volume positively correlated with the behavioral measures for capacity (red) and precision (blue) in Experiment 2 (left). All voxels shown are thresholded at Puncorrected < 0.005 for display purposes. Behavior-to-brain relations in Experiment 2. The GM volumes in the left LOC and right IPS significantly correlated with behavioral measures of VWM capacity (left scatterplot) and VWM precision (right scatterplot), respectively. Notably, no voxels around the right IPS survived our threshold for the behavioral measure for quantity. Figure 2 Open in new tabDownload slide Task procedures and results from Experiment 2 relating behavioral performance to precision and capacity of VWM. (a) VWM quantity condition. Only the target item in the probe display did not change color, while the rest of the items turned black. Participants were asked to indicate whether the orientation of the target item in the probe array was rotated clockwise or counterclockwise relative to the item at the same location in the sample array. The orientation difference between the target item was ±45°. The set size in both the sample and probe arrays was 4 items in each visual hemifield in both the sample and probe arrays. The condition thus required a low load on discrimination (coarse precision) but a high load on retention of multiple items (capacity). (b) VWM quality condition. The same paradigm as in the quantity condition was used, except that the set size was 2 items in each visual hemifield and the orientation difference between the target item in the probe array and the item at the same location in the sample array was ±15°. Therefore, this condition required a high load regarding the discrimination but a low load on the retention of multiple items. Brain regions in which GM volume positively correlated with the behavioral measures for capacity (red) and precision (blue) in Experiment 2 (left). All voxels shown are thresholded at Puncorrected < 0.005 for display purposes. Behavior-to-brain relations in Experiment 2. The GM volumes in the left LOC and right IPS significantly correlated with behavioral measures of VWM capacity (left scatterplot) and VWM precision (right scatterplot), respectively. Notably, no voxels around the right IPS survived our threshold for the behavioral measure for quantity. Behavior-to-Brain Anatomy Relations The VBM analyses indicated that the behavioral measure of VWM capacity (the ability to retain the number of items with coarse precision) was significantly correlated with GM volume in the left LOC (T = 3.67, Psvc < 0.05; peak voxel MNI coordinates (X, Y, Z) = [−38, −60, 9]), whereas that of precision (the ability to retain a few items with fine precision) remarkably correlated with those in the right superior IPS (T = 3.10, Psvc < 0.05; [18, −62, 46]) as well as in the left superior IPS (T = 3.17, Psvc < 0.05; [−18, −70, 44]) across individuals (see Fig. 1d). No significant correlation was observed between VWM capacity and GM volume in the right LOC, between VWM capacity and bilateral IPS, or between VWM precision and those in the bilateral LOC (all did not survive our initial threshold criteria; psuncorrected > 0.005, n.s.). An exploratory whole-brain analysis revealed a cluster in the left medial orbito-prefrontal cortex [−16, 57, 15] that was significantly correlated with VWM capacity (T = 6.10, PFWE-corrected = 0.014) and a cluster in the left precentral gyrus that was significantly correlated with VWM precision (T = 6.10, PFWE-corrected = 0.014). Supplementary Table 1 and Supplementary Figure 2 depict cortical regions positively correlated with both aspects of VWM, following the conjunction analysis, revealing “positive” associations between both the behavioral scores and GM volume in the insula and precentral gyrus. Materials and Methods for Experiment 2 As the results above have shown, we found a uniquely dissociable behavior-to-brain relationship in Experiment 1. However, the aspects of VWM for capacity and precision were assessed by using different visual features (color for capacity and bar orientation for precision), and the task paradigm also differed with a change-detection task for capacity and change-discrimination task for precision. As the tasks were different, these two tasks were tested in a blocked manner. Although these two tasks might have been widely accepted means to examine specific aspects of VWM, these discrepancies needed to be ruled out. In Experiment 2, the same visual feature, bar orientation, was used to test the two aspects of VWM, and the paradigm was set to be a change-discrimination task for both. Furthermore, these two conditions were intermixed in the same task and tested in a “randomized” manner. Participants were asked to report whether a probed bar was rotated clockwise or counterclockwise relative to the orientation of the bar at the same location in the sample array. In the capacity condition, participants were asked to perform coarse (less demanding) orientation discriminations (±45°) while being required to retain as many items as they could (more demanding). In contrast, in the precision condition, participants were required to perform fine (more demanding) discriminations (±15°), while being asked to retain only two items (less demanding). The instructions for the two tests were identical to Experiment 1. However, in Experiment 2, the capacity and precision conditions alternated from trial to trial, and participants were required to switch strategies (whether focusing on capacity or precision). In this manner, the potential confounding factors were controlled such as the feature difference between the capacity test (colored squares) and the precision test (oriented bars), the task difference between the capacity test (change detection) and the precision test (discrimination), and the order differences between blocked in Experiment 1 and randomized in Experiment 2. We examined whether neural activities specific to the retention interval separately obtained for the VWM capacity and precision assessments would also relate to GM volumes. For this purpose, the amplitude of contralateral delay activity (CDA) was measured during the retention interval between the presentation of a sample array and probe. CDA refers to an interhemispheric difference in a sustained event-related potential component during the delay period. The amplitude of CDA reportedly correlated with the amount of information retained during the retention interval (Vogel and Machizawa 2004; McCollough et al. 2007; Machizawa et al. 2012; Fukuda et al. 2015) and increases with the number of items successfully retained in mental representation (Vogel and Machizawa 2004; McCollough et al. 2007). Also, CDA responds to the required precision with which items are retained (Machizawa et al. 2012). Thus, CDA would be a useful neural marker related to VWM retention. Participants Twenty participants (aged between 19 and 35 years old with a mean age of 25.9 years) were recruited from a local community in London, United Kingdom. All participants gave written informed consent that was approved by the UCL Research Ethics Committee in accordance with the Declaration of Helsinki. Behavioral Task The procedures of the behavioral task in Experiment 2 were identical to those used in a previous study (Machizawa et al. 2012) except that the capacity- and precision-focused conditions were randomized in this study, whereas the capacity-focused (“coarse-precision”) and precision-focused (“fine-precision”) conditions were blocked in the previous study. In short, the participants needed to discriminate between a clockwise and counterclockwise orientation change in a probe target (one out of two or four samples remained the same color as in the sample array, while all task-irrelevant bars that did not change their orientations were black). In the probe array, a fine-orientation change occurred (15° change) for the precision-focused condition, whereas a coarse-orientation change (45° change) took place for the capacity-focused condition. Notably, an equally challenging orientation change (30° change) was “covertly” tested with observers. The participants were instructed to retain the orientation of bars with “high” precision for the precision-focused condition because a change would be subtle (15°), while it was discouraged for the capacity-focused condition because the change would be large (45°). Conversely, the number of items to retain was stressed for the capacity condition. There were 192 trials per condition collapsing across the two hemifield conditions. Behavioral Indices for the Precision and Capacity of VWM The behavioral index of VWM precision was defined by the difference in the proportion correct between “2 fine-precision” and “2 coarse-precision” conditions of intermediate difficulty trials (“2-fine”–“2-coarse”), in which the participants anticipated precision differently (15° change for expect-fine trials and 45° change for expect-coarse trials), while the same degree of precision (30° change) was required at test. Likewise, the behavioral index for VWM capacity was defined by the differences in proportion correct between the “4-coarse” and 2-coarse conditions of congruent-difficulty trials (2-coarse–4-coarse). EEG Recording Procedures The recording and analysis procedure was identical to a previously reported study (Machizawa et al. 2012; please note that the participants and behavioral procedures reported in “this” manuscript are “different” from the previous study). Electroencephalographs (EEGs) were collected from 64 active electrodes of the ActiveTwo system at 512 Hz (Biosemi, Amsterdam, the Netherlands. https://www.biosemi.com/) in accord with the international 10–20 system. Of those, two active electrodes were placed on the bilateral mastoids for subsequent re-referencing. A vertical and a horizontal electrooculogram (EOG) was placed beside the eyes (cf., McCollough et al. 2007). EEG Analysis Procedures The collected data were band-pass filtered offline (with 8th-order Butterworth) at 0.05–30 Hz, re-referenced to the mastoid channels, epoched to −200 to 1400 ms relative to the onset of the sample array, and baseline corrected (−200 to 0 ms relative to the onset of sample array). The trials contaminated by artifacts were rejected and excluded from further analyses. As in previous studies measuring CDA (Vogel and Machizawa 2004; McCollough et al. 2007), artifacts such as large eye blinks (>50 μV) detected on a vertical EOG channel under the left eye as well as losses of fixation detected on a horizontal EOG channel (>25 μV, i.e., roughly equivalent to > 2° of visual angle against 4–10° of peripheral targets) were rejected. The data were then averaged across conditions and participants for subsequent analyses. CDA was calculated from the posterior parietal and lateral occipital channels (viz, P5, P7, PO3, PO7, and O1 for the left hemisphere and P6, P8, PO4, PO8 and O2 for the right hemisphere). CDA amplitudes were averaged across the 400–1400 ms after the onset of the sample array during the delay period. With regard to the total number of EEG trials after the artifact rejection, out of 192 trials per condition pooled over the left and right hemifield conditions (i.e., 2-coarse, 4-coarse, and 2-fine conditions each), repeated measures ANOVA confirmed that there was no main effect difference between the capacity-focused or precision-focused conditions (F1,19 = 0.30, P = 0.59) or between set sizes 2 and 4 (F1,19 = 0.65, P = 0.43). Notably, the proportion of retained trials seems relatively lower than typically reported (i.e., Vogel and Machizawa 2004). However, since the nature of the task made it highly challenging for the participants to discriminate for fine changes, we speculate that in our task, participants tended to drift or make saccades toward the target hemifield more often than for a simpler task. As the number of retained trials was considered to be sufficient for obtaining a CDA amplitude (the minimum number of retained trials for a condition was 82 trials for one participant), all 20 participants were included in the subsequent analyses. Neural Indices of VWM Precision and Capacity As a neural index of VWM capacity, we calculated the CDA amplitude difference between 2 coarse-precision conditions in which the number of items on the cued side was two and 4 coarse-precision conditions in which the number was four, a conventionally applied method to estimate the size of VWM capacity (Vogel and Machizawa 2004; Vogel et al. 2005). For a neural index of VWM precision, we calculated the CDA amplitude difference between 2 fine-precision and 2 coarse-precision conditions in both of which the number of items on the cued side was two (cf., Machizawa et al. 2012). Notably, a previous study using a similar paradigm showed that the CDA amplitudes for the 2 fine-precision and the 4 coarse-precision conditions were equally larger than those for the 2 coarse-precision conditions (Machizawa et al. 2012). It has been hypothesized that if the CDA increases for precision and capacity do not correlate across participants, then distinct regions would emerge as the result of our multiple regression analyses. In turn, if the two neural indices strongly correlate with each other, the same brain regions would emerge. Results for Experiment 2 Behavioral Results Table 1 below summarizes the accuracy of each condition. As in convention, we found significantly lower performance for set size 4 conditions (mean ± 1SE: 0.66 ± 0.02) compared with set size 2 conditions (0.77 ± 0.03; F1,19 = 83.39; P < 0.001; partial η2 = 0.81). Accuracy for the fine-precision conditions (0.70 ± 0.02) was significantly lower than that of coarse-precision conditions (0.73 ± 0.03) as was expected (F1,19 = 9.28; P = 0.007; partial η2 = 0.33) on average. These had led to inter-individual variation for the behavioral indices for capacity, the difference between 2-coarse and 4-coarse conditions (0.11 ± 0.06, ranged 0.02 to 0.22) and for precision, the differences between 2-coarse and 2-fine conditions (0.001 ± 0.07, ranged −0.14 to 0.13). Table 1 Mean and SD of accuracy (proportion correct) for Exp. 2 . Set size 2 . Set size 4 . Fine . Coarse . Fine . Coarse . Congruent (15°/45°) 0.72 ± 0.11 0.77 ± 0.13 0.62 ± 0.11 0.66 ± 0.11 Intermediate (30°) 0.80 ± 0.11 0.79 ± 0.13 0.67 ± 0.09 0.68 ± 0.10 . Set size 2 . Set size 4 . Fine . Coarse . Fine . Coarse . Congruent (15°/45°) 0.72 ± 0.11 0.77 ± 0.13 0.62 ± 0.11 0.66 ± 0.11 Intermediate (30°) 0.80 ± 0.11 0.79 ± 0.13 0.67 ± 0.09 0.68 ± 0.10 Open in new tab Table 1 Mean and SD of accuracy (proportion correct) for Exp. 2 . Set size 2 . Set size 4 . Fine . Coarse . Fine . Coarse . Congruent (15°/45°) 0.72 ± 0.11 0.77 ± 0.13 0.62 ± 0.11 0.66 ± 0.11 Intermediate (30°) 0.80 ± 0.11 0.79 ± 0.13 0.67 ± 0.09 0.68 ± 0.10 . Set size 2 . Set size 4 . Fine . Coarse . Fine . Coarse . Congruent (15°/45°) 0.72 ± 0.11 0.77 ± 0.13 0.62 ± 0.11 0.66 ± 0.11 Intermediate (30°) 0.80 ± 0.11 0.79 ± 0.13 0.67 ± 0.09 0.68 ± 0.10 Open in new tab Behavior-to-Brain Anatomy Relations In Experiment 2, we found the same tendency in the results with the VBM analyses as in the behavior-to-brain relation assessments. The GM volume in the left LOC (T = 3.74; peak MNI coordinates [−29, −61, 9]) positively correlated with the capacity, whereas volumes in the right IPS (T = 3.74; [20, −63, 46]) and right LOC (T = 6.34; [51, −66, 4]) positively correlated with the precision indices (all PSVC < 0.05). As shown in Figure 2, one part of the left occipital cortex also survived the initial threshold but was outside of our a priori ROI for the left LOC that fell within ROI (shown as a red blob). As shown in Figure 2, the peak volume in the left LOC correlated with capacity was not significantly correlated with precision (Spearman’s ρ2 = 0.15, P = 0.09, n.s.). Likewise, the peak volume in the right IPS was not significantly correlated with capacity (Spearman’s ρ2 = 0.15, P = 0.09, n.s.). In addition to the ROI analyses, an exploratory whole-brain exploratory analysis revealed a positive correlation between the behavioral index for VWM capacity and a cluster in the right motor cortex (T = 5.51; PFWE-corrected = 0.047; [6, 2, 64]). No clusters survived the whole-brain analysis for VWM precision. Brain Activity to Brain Anatomy Relation CDA amplitudes (mean ± SD) for 2-fine, 2-coarse, 4-fine, and 4-coarse conditions were −1.21 ± 0.95, −1.25 ± 0.86, −1.49 ± 1.06, and −1.29 ± 0.97. The number of retained trials (mean ± SD) for each conditions after artifact rejection for 2-fine, 2-coarse, 4-fine, and 4-coarse conditions was 136.0 ± 22.6, 133.6 ± 24.2, 135.8 ± 19.9, and 136.7 ± 21.3, respectively. CDA responses for capacity (4-coarse minus 2-coarse conditions) and those for precision (2-fine minus 2-coarse conditions) that were significantly yet weakly correlated (r220 = 0.28; P = 0.02; see Supplementary Fig. 3). Because a putative dissociation of these two CDA responses was expected for the VBM analysis due to the weak relationship between these two, we further examined the neuroanatomical correlates of the CDA for capacity and precision. Figure 3 shows that a CDA amplitude increase for the capacity condition was significantly correlated with the GM volume in the left LOC (but not in the right hemisphere), whereas a CDA amplitude increase for the precision significantly correlated with the GM volume in the right IPS (but not in the left hemisphere). The results indicated that CDA for capacity was significantly associated with GM volume in the left LOC (T = 3.66, Psvc < 0.05; [−42, −66, 2]), whereas that for precision was significantly associated with GM volume in the right superior IPS (T = 4.70, Psvc < 0.01; [29, −60, 44], see Fig. 2d). The GM volumes of other ROIs (viz., the right LOC and bilateral IPS for CDA capacity and bilateral LOC for “CDA precision”) did not even survive our initial threshold criteria (Puncorrected > 0.005, n.s.). While it was outside of our primary ROIs, insular cortex and cingulate cortex also survived the initial threshold but did not survive correction at the whole-brain level (see Supplementary Fig. 4). Figure 3 Open in new tabDownload slide Results of Experiment 2 relating a neural marker of VWM (CDA) to the precision and capacity conditions. (a) GM volume correlates of neural indices that were significantly correlated with the neural measures of VWM quantity (red blob) and that with VWM quality (blue blobs). (b, c) Scatterplots of GM volumes as a function of neural indices of VWM quantity (b, “CDA quantity”) and quality (c, “CDA quality”) with fitted trend lines (gray lines). (b) GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of CDA changes between 2-coarse and 4-coarse items (CDA quantity). (c) GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of CDA changes between 2-coarse and 2-fine items (“CDA precision”). The GM volume in the left LOC was positively correlated with CDA quantity (left panel in b), and GM volume in the right IPS was positively correlated with CDA quality (right panel in c). Figure 3 Open in new tabDownload slide Results of Experiment 2 relating a neural marker of VWM (CDA) to the precision and capacity conditions. (a) GM volume correlates of neural indices that were significantly correlated with the neural measures of VWM quantity (red blob) and that with VWM quality (blue blobs). (b, c) Scatterplots of GM volumes as a function of neural indices of VWM quantity (b, “CDA quantity”) and quality (c, “CDA quality”) with fitted trend lines (gray lines). (b) GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of CDA changes between 2-coarse and 4-coarse items (CDA quantity). (c) GM volumes in the left LOC (left panel) and the right IPS (right panel) as a function of CDA changes between 2-coarse and 2-fine items (“CDA precision”). The GM volume in the left LOC was positively correlated with CDA quantity (left panel in b), and GM volume in the right IPS was positively correlated with CDA quality (right panel in c). Provided the same variance is expressed in both the behavior–brain and CDA–brain relations, we should observe a significant influence on the correlation by controlling for the EEG measurements. To validate the behavior–brain and CDA–brain correlations, we additionally performed a partial correlation analysis. For the observed correlation between the behavioral capacity and the volume in the left LOC, the EEG values (CDA capacity effect) were covaried out. At the commonly observed cluster in the left LOC (MNI coordinate at [−37, −61, 9]), for the capacity effect, the Pearson correlation was drastically reduced to r220 = 0.29 (P = 0.02). Likewise, after partialling out the EEG values (CDA precision effect) from the behavior–brain relation at the right IPS ([21, −60, 45]), the Pearson correlation also dropped to r220 = 0.25 (P = 0.03). As added reassurance, the correlations for the left LOC and for the right IPS were not significantly different (Z = 0.12; P = 0.91, two-tails), suggesting the same variance was expressed for both the behavior–brain and CDA–brain relations. In addition to seeking putatively positive relationships between volume and VWM performance, as many of previous structural MRI studies have found, we also examined the negative correlation between individual differences in the CDA amplitude responses and GM volume for completeness. As it turned out, no negative correlations were significant for CDA differences for the precision effect (viz., differences between 2-fine and 2-coarse conditions). Interestingly, a cluster in the left dorsolateral prefrontal cortex (T = 5.07; non-stationarity corrected Pcorrected = 0.031; 679 resels, [−28, 30, 39]) was significantly and “negatively” correlated with the capacity measure of CDA, which was the difference between 4-coarse and 2-coarse conditions (see Supplementary Fig. 5). Confirmatory Analysis for the Correlations in the Left and Right ROIs The results also indicate a dissociation between the left LOC and the right IPS in terms of correlations. For both Experiments 1 and 2, we further examined whether the behavior–brain correlation for the left LOC differed from that for the right IPS. The correlations between the left LOC and the right IPS for the behavior–brain relations for capacity (Z49 = 0.98; P = 0.33, n.s.) in Experiment 1 were not significantly different but were weakly dissociated for precision (Z49 = −1.63; P = 0.05; one-tailed). In Experiment 2, the correlations of the two ROIs were significantly different for the behavior–brain relations for both capacity (Z20 = 1.77; P = 0.04; one-tailed) and precision (Z20 = 2.76; P = 0.01; two-tailed), as were those of the CDA–brain relations for both CDA capacity (Z20 = 3.17; P < 0.01; two-tailed) and CDA precision (Z20 = −4.80; P < 0.01; two-tailed). To summarize, the correlations for the behavioral measures showed relatively weak dissociations; however, relations between the neural activity and the brain structure robustly showed dissociation in these two regions. Discussion The results from these correlational analyses suggested that a few shared regions may constrain both capacity-focused and precision-focused representations of VWM, but there is some tendency that individual differences in each aspect independently correlated with individual variabilities in different cortical regions. Throughout the three sets of VBM results, we consistently found that both capacity and precision of VWM, each of which was predominantly correlated with different structures, were related to different cortical areas within previously reported VWM networks (see Supplementary Table 1, summarizing results of all ROIs considered). Multiple regions were found to be related, at least surviving for our initial threshold; however, we observed intriguingly dissociable, as well as some shared, links between the qualitative and quantitative aspects of VWM and GM volume in these cortical regions. Neuroanatomical Correlates Specific to VWM Capacity With regard to VWM capacity, the results of our two experiments consistently indicated that the structure of the left lateral occipital region positively correlated with one’s ability to maintain a varying number of items in VWM in Experiments 1 and 2, and these correlational outcomes only with the behavioral scores were further supported by relating their neural activity during the delay period to the brain anatomy. In the behavioral experiment, we found that individuals who had a larger GM volume in the LOC only in the left hemisphere tended to be able to maintain a large number of items in VWM compared with those with a smaller volume in this region. In another experiment using EEG, individual differences in both behavioral performance and the neural activity responsible for the maintenance of many items, as assessed by the CDA, were significantly correlated with the GM volume in the left LOC. Neuroanatomical Correlates Specific to VWM Precision As for the “quality” of VWM, the GM volume in the right parietal region across individuals positively correlated with the behavioral and neural measures of retaining a few items with fine precision. As was reported previously, superior regions of IPS are associated with maintenance of object identity in detail (Xu and Chun 2006; Xu 2007). It would be feasible to state that IPS region may be associated with the maintenance of visual information with fine precision. Although this area is well known to be responsible for a variety of visual attention and working memory operations (Todd and Marois 2004; Xu and Chun 2006; Ruff et al. 2008; Malhotra et al. 2009; Shulman et al. 2010), it was somewhat surprising not to find a correlation of IPS regions with the capacity of VWM. Neuroanatomical Dissociation between Quantity and Quality of VWM With a particular focus of the relationship between visual attention and VWM, previous studies of visual attention have proposed that the separable aspects of attention may be independently linked to putatively dissociable aspects of VWM (Fan et al. 2005; Machizawa and Driver 2011). In an old study examining three attentional networks, Fan et al. (2005) reported that functional activation in the thalamus and bilateral LOC, but not in the IPS, were responsible for the “alerting” of attention that requires vigilance to the entire visual field. In contrast, functional activations in primarily the right IPS, but not in the left parietal regions, were responsible for the “spatial orienting” of attention. This orienting component of attention reflects a process to shift their focus of attention to a particular spot in space. Additionally, a previous study linking such putatively separable components of visual attention and VWM reported underlying relations of alerting attention to VWM capacity and spatial orienting attention to VWM precision (Machizawa and Driver 2011). Taken together, the current finding on the structural basis of VWM may suggest that the separable aspects of attentional subprocesses may also play a role in dissociating “quantity” from quality in VWM. Common Regions Related to Both Quantity and Quality of VWM As for the whole-brain analysis that sought a conjunction of the two aspects in Experiment 1, the cortical volume of midprefrontal region positively affiliated with both aspects of VWM. Notably, this region survived the initial threshold for the behavioral measure of precision but not for that of capacity. It has been reported that functional activations in this region correlated with the attentional selection of task-irrelevant items in VWM (McNab and Klingberg 2008). Contribution of this middle frontal region may be closely associated with situations in which sophisticated allocation of attentional resources with high precision is necessary. Nevertheless, this region emerged to be significant for the conjunction of capacity and precision tasks, suggesting a putative role of selective attention to ignore task-irrelevant items presented in the other hemifield that was shared by both tasks. Another area also emerged in the conjunction analysis. The bilateral insula seemed to link with both behavioral measures of VWM. In addition, the cortical volume of bilateral insula was weakly related to CDA capacity effect (Supplementary Fig. 4). As the central region of the saliency network, the insula is associated with a variety of cognitive functions, but recent theories suggest its primary role in predictive coding of interoceptive awareness associated with emotional processing (Nguyen et al. 2016; Salomon et al. 2016; Seth and Friston 2016). It was intriguing to find, however, that a previous structural MRI study also reported a positive association of the insula with one’s ability to retain objects in short-term memory (Konstantinou et al. 2017). In our paradigm, participants also primarily retained object identity in memory; in the meantime, the retention process possesses a prediction coding mechanism to prepare for an upcoming probe array. Alternatively, the saliency network was straightforwardly relevant to the task for the identification of salient objects. Given the predictive nature of the insula and the role of the middle frontal cortex in selective attention, the sophisticated interplay among these remotely interconnected brain regions may support the underlying processes of VWM. Hemispheric Laterality Differences Observed between the Quantity and Quality of VWM In this study, to-be-remembered items, as well as to-be-ignored items, were displayed in each visual hemifield. A question arises: how are the maintained items within one visual hemifield mentally restricted in VWM? In accord with the above-mentioned conventional model, it was thought that visual information held as a mental representation was bilaterally processed with the capacity for each visual hemifield predominantly associated with a different hemisphere contralateral to the to-be-remembered hemifield (Gratton 1998; Vogel and Machizawa 2004; Umemoto et al. 2010). Overall, such tread may be rooted; however, this was not consistent with our results such that individual differences in the capacity of VWM were primarily related to GM volume in the left LOC region. As one line of speculation, unilaterally encoded information may be transferred to the region for an integration of multiple objects in mental representation. Alternatively, these findings were solely structural correlates; that is, only individuals with high capacity had a larger volume. It is plausible that such large volumes may reflect a large number of neurons and synapses in the region (Kanai and Rees 2011) that support the encoding of multiple objects at a time. Studies on “verbal” working memory report a critical function of the dorsolateral prefrontal area (DLPFC) or Broca’s area. Coincidentally, we found the left LOC was involved in the color task in Experiment 1. It may suggest a question that this “left” lateralization may be due to potential verbalization in the color task for quantity but not for orientation task for the quality. One can imagine that the color task was susceptible to verbalization, leaving putative association to the left hemisphere. However, because there was little time to verbalize identities of all items (i.e., at the most challenging condition, naming eight colors and their locations within 900 ms), even if they tried, people might fail to achieve high performance. While the color was the target feature in Experiment 1, the canonical orientations (horizontal, vertical, or diagonal angles) were avoided in Experiment 2 as well as the precision task in Experiment 1 to restrict the use of verbalization. Therefore, the finding of the left LOC may not be related to verbalization. To further support, previous research indicates that extraneous loads in verbal working memory should not influence the capacity of VWM (Luck and Vogel 1997), indicating purely visual processing could take place even for the color task. Henceforth, we may be able to put aside the potential counterargument on the role of the left LOC in verbalization for the color task in Experiment 1. Relating to the above section about verbal working memory, in the supplementary analysis seeking for negative relation to the CDA in Experiment 2, the GM volume in left mid frontal gyrus (in a close proximity to DLPFC) negatively correlated with the CDA increase responsible for the capacity (Supplementary Fig. 5). This observation may instead propose an involvement of verbalization during the task and its impact on the CDA—a neural marker responsible for the maintenance of visual information. While exact angles of the bar orientation were not canonical, some might still verbalized (i.e., “left tilt,” “right tilt,” “a bit left,” “a bit right,” etc.) for up to four items. Individuals with their preference or ability to verbalize might counteract against the putatively visual neural activities of working memory, as measured by the CDA. In this discussion, however, the limitations in this experimental design restrict our speculation because an explicit suppression of verbalization, just like Luck and Vogel (1997), was not implemented in this study (see Supplementary Fig. 5 for further discussion). Limitation Factors in This Study There are several limitations. First, while we observed dissociable behavioral correlations for the capacity and precision, these two aspects were tested by giving different instructions encouraging the use of different aspect of visual cognition. In addition, separable correlations were also found for the neural correlates of VWM for capacity and precision contrasts (2-coarse vs. 4-coarse conditions for capacity-focused condition); however, the capacity contrast was within task contrast, but the precision contrast was between task contrast (2-coarse for capacity-focused condition vs. 2-fine for precision-focused condition). One could elaborate the paradigm within the same task-set to fully validate the capacity and precision dissociation within the same framework in the future. Furthermore, given the regional specificity of the brain, our findings may only generalize to the examined visual features, namely, color patches or bar orientations. Indeed, our report still lacks conclusive evidence for causality as well as its statistical power with a relatively small sample size, particularly with the brain-to-brain relation accompanied by neural measures. Future studies would need to fill the gap we have left in this study. Conclusion To summarize, it has been debated whether our VWM is solely constrained by capacity, also known as the discrete slot model (Luck and Vogel 1997; Zhang and Luck 2008; Luck and Vogel 2013), or by the precision with which items are retained, also known as the flexible resource model (Bays and Husain 2008; Ma et al. 2014). Given the assumption that our VWM representation is solely constrained by the shared resources, in which VWM resources are shared by a unitary mechanism, one may expect the same brain regions to be responsible for both capacity and precision. While we observed some common brain regions that were correlated with both aspects of VWM, we primarily found dissociations between capacity and precision. The resolution-only account does not predict the observed dissociation. Instead, our current finding supports the hybrid account of VWM in which, under limited conditions, individual behavioral and neural differences are dissociably correlated with separable aspects of VWM operation. To date, neuroimaging studies have revealed potential functional sources for VWM storage in the brain (Todd and Marois 2004; Xu and Chun 2006), yet to date, the neuroanatomical underpinnings have not been clear in association with attention and VWM. In the current study, both behavioral and neural measures of the quantity of VWM were positively correlated with individual differences in GM volume of the left LOC, whereas neural measures of the quality of VWM were predominantly correlated with individual differences in GM volume in the right IPS as well as a few other regions. To date, several lines of research have suggested potential dissociations of the two hemispheres based on an independence in VWM resources (Alvarez and Cavanagh 2005; Umemoto et al. 2010; Buschman et al. 2011), dissociable global versus local processing (Husain and Rorden 2003; Malhotra et al. 2005), or saliency processing (Mevorach et al. 2006, 2009). As such, the varieties of the putatively different aspects of these functions seem to interact in VWM. Even with some limitations in this study, our results provide empirical evidence that both capacity and precision constrain VWM with a uniquely distinctive cortical anatomy supporting the separable aspects of attention and working memory. Thus, these results may provide a rapprochement of the controversy as to whether VWM function is restricted solely by the quantity or quality of mental representations. In other words, both discrete representations and unified representation may be dissociably associated with individual differences in remotely interconnected cortical regions, proposing a new line of neuroanatomical correlates of VWM. Funding Wellcome Trust (a programme grant to J.D.); Royal Society (Royal Society 2010 Anniversary Professorship to J.D.); UCL Graduate School award to M.G.M.; Japan Science and Technology Agency Center of Innovation Program (JPMJCE1311 to M.G.M), National Institutes of Health (R01EY019466, R01EY027841 to T.W.); United States-Israel Binational Science Foundation (BSF 2016058 to T.W.). Notes In memory of J.D., who sadly deceased prior to the completion of this manuscript. J.D. held a Royal Society 2010 Anniversary Research Professorship. We thank Sophie B Gordon-Smith and Crystal CW Goh for supporting data collection for Exp. 1 and 2, respectively, as well as Prof. John Ashburner for providing his guidance on the VBM approaches. Conflict of Interest: The authors declare no conflict of interest. Author Contributions M.G.M. and J.D. designed and discussed Exp. 1 and 2. M.G.M. and T.W. discussed and wrote this manuscript. M.G.M. performed experiments and statistical analyses. References Alvarez GA , Cavanagh P. 2005 . Independent resources for attentional tracking in the left and right visual hemifields . Psychol Sci. 16 : 637 – 643 . Google Scholar Crossref Search ADS PubMed WorldCat Ashburner J . 2007 . A fast diffeomorphic image registration algorithm . NeuroImage. 38 : 95 – 113 . Google Scholar Crossref Search ADS PubMed WorldCat Ashburner J , Friston KJ. 2000 . Voxel-based morphometry—the methods . NeuroImage. 11 : 805 – 821 . Google Scholar Crossref Search ADS PubMed WorldCat Ashburner J , Friston KJ. 2001 . Why voxel-based morphometry should be used . NeuroImage. 14 : 1238 – 1243 . Google Scholar Crossref Search ADS PubMed WorldCat Bays PM , Husain M. 2008 . Dynamic shifts of limited working memory resources in human vision . Science. 321 : 851 – 854 . Google Scholar Crossref Search ADS PubMed WorldCat Bergmann J , Genç E, Kohler A, Singer W, Pearson J. 2016 . Neural anatomy of primary visual cortex limits visual working memory . Cereb Cortex. 26 : 43 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat Buschman TJ , Siegel M, Roy JE, Miller EK. 2011 . Neural substrates of cognitive capacity limitations . PNAS. 108 : 11252 – 11255 . Google Scholar Crossref Search ADS WorldCat Cowan N . 2001 . The magical number 4 in short-term memory: a reconsideration of mental storage capacity . Behav Brain Sci . 24 : 87 – 114 discussion114–discussion185 . Google Scholar Crossref Search ADS PubMed WorldCat de Fockert JW , Rees G, Frith CD, Lavie N. 2001 . The role of working memory in visual selective attention . Science. 291 : 1803 – 1806 . Google Scholar Crossref Search ADS PubMed WorldCat Emrich SM , Riggall AC, LaRocque JJ, Postle BR. 2013 . Distributed patterns of activity in sensory cortex reflect the precision of multiple items maintained in visual short-term memory . J Neurosci. 33 : 6516 – 6523 . Google Scholar Crossref Search ADS PubMed WorldCat Engle RW . 2002 . Working memory capacity as executive attention . Curr Dir Psychol Sci. 11 : 19 – 23 . Google Scholar Crossref Search ADS WorldCat Fan J , McCandliss BD, Fossella J, Flombaum JI, Posner MI. 2005 . The activation of attentional networks . NeuroImage. 26 : 471 – 479 . Google Scholar Crossref Search ADS PubMed WorldCat Fukuda K , Awh E, Vogel EK. 2010a . Discrete capacity limits in visual working memory . Curr Opin Neurobiol. 20 : 177 – 182 . Google Scholar Crossref Search ADS WorldCat Fukuda K , Mance I, Vogel EK. 2015 . α Power modulation and event-related slow wave provide dissociable correlates of visual working memory . J Neurosci. 35 : 14009 – 14016 . Google Scholar Crossref Search ADS PubMed WorldCat Fukuda K , Vogel E, Mayr U, Awh E. 2010b . Quantity, not quality: the relationship between fluid intelligence and working memory capacity . Psychon Bull Rev. 17 : 673 – 679 . Google Scholar Crossref Search ADS WorldCat Gratton G . 1998 . The contralateral organization of visual memory: a theoretical concept and a research tool . Psychophysiology. 35 : 638 – 647 . Google Scholar Crossref Search ADS PubMed WorldCat Husain M , Rorden C. 2003 . Non-spatially lateralized mechanisms in hemispatial neglect . Nat Rev Neurosci. 4 : 26 – 36 . Google Scholar Crossref Search ADS PubMed WorldCat Kamitani Y , Tong F. 2005 . Decoding the visual and subjective contents of the human brain . Nature Neuroscience. 8 : 679 – 685 . Google Scholar Crossref Search ADS PubMed WorldCat Kanai R , Rees G. 2011 . The structural basis of inter-individual differences in human behaviour and cognition . Nat Rev Neurosci. 12 : 231 – 242 . Google Scholar Crossref Search ADS PubMed WorldCat Konstantinou N , Constantinidou F, Kanai R. 2017 . Discrete capacity limits and neuroanatomical correlates of visual short-term memory for objects and spatial locations . Hum Brain Mapp. 38 : 767 – 778 . Google Scholar Crossref Search ADS PubMed WorldCat Luck SJ , Vogel EK. 1997 . The capacity of visual working memory for features and conjunctions . Nature. 390 : 279 – 281 . Google Scholar Crossref Search ADS PubMed WorldCat Luck SJ , Vogel EK. 2013 . Visual working memory capacity: from psychophysics and neurobiology to individual differences . Trends Cogn Sci. 17 : 391 – 400 . Google Scholar Crossref Search ADS PubMed WorldCat Ma WJ , Husain M, Bays PM. 2014 . Changing concepts of working memory . Nat Neurosci. 17 : 347 – 356 . Google Scholar Crossref Search ADS PubMed WorldCat Machizawa MG , Driver J. 2011 . Principal component analysis of behavioural individual differences suggests that particular aspects of visual working memory may relate to specific aspects of attention . Neuropsychologia. 49 : 1518 – 1526 . Google Scholar Crossref Search ADS PubMed WorldCat Machizawa MG , Goh CCW, Driver J. 2012 . Human visual short-term memory precision can be varied at will when the number of retained items is low . Psychol Sci. 23 : 554 – 559 . Google Scholar Crossref Search ADS PubMed WorldCat Malhotra P , Coulthard EJ, Husain M. 2009 . Role of right posterior parietal cortex in maintaining attention to spatial locations over time . Brain. 132 : 645 – 660 . Google Scholar Crossref Search ADS PubMed WorldCat Malhotra P , Jäger HR, Parton A, Greenwood R, Playford ED, Brown MM, Driver J, Husain M. 2005 . Spatial working memory capacity in unilateral neglect . Brain. 128 : 424 – 435 . Google Scholar Crossref Search ADS PubMed WorldCat Mance I , Vogel EK. 2013 . Visual working memory . Wiley Interdiscip Rev Cogn Sci. 4 : 179 – 190 . Google Scholar Crossref Search ADS PubMed WorldCat Matsuyoshi D , Ikeda T, Sawamoto N, Kakigi R, Fukuyama H, Osaka N. 2012 . Differential roles for parietal and occipital cortices in visual working memory . PLoS One. 7 : e38623 . Google Scholar Crossref Search ADS PubMed WorldCat McCollough AW , Machizawa MG, Vogel EK. 2007 . Electrophysiological measures of maintaining representations in visual working memory . Cortex. 43 : 77 – 94 . Google Scholar Crossref Search ADS PubMed WorldCat McNab F , Klingberg T. 2008 . Prefrontal cortex and basal ganglia control access to working memory. Nature Neuroscience. 11 : 103 – 107 . Google Scholar Crossref Search ADS PubMed WorldCat Mevorach C , Humphreys GW, Shalev L. 2006 . Opposite biases in salience-based selection for the left and right posterior parietal cortex . Nat Neurosci. 9 : 740 – 742 . Google Scholar Crossref Search ADS PubMed WorldCat Mevorach C , Humphreys GW, Shalev L. 2009 . Reflexive and preparatory selection and suppression of salient information in the right and left posterior parietal cortex . J Cogn Neurosci. 21 : 1204 – 1214 . Google Scholar Crossref Search ADS PubMed WorldCat Moran RJ , Campo P, Maestu F, Reilly RB, Dolan RJ, Strange BA. 2010 . Peak frequency in the theta and alpha bands correlates with human working memory capacity . Front Hum Neurosci. 4 : 200 . Google Scholar Crossref Search ADS PubMed WorldCat Murray AM , Nobre AC, Stokes MG. 2011 . Markers of preparatory attention predict visual short-term memory performance . Neuropsychologia. 49 : 1458 – 1465 . Google Scholar Crossref Search ADS PubMed WorldCat Nguyen VT , Breakspear M, Hu X, Guo CC. 2016 . The integration of the internal and external milieu in the insula during dynamic emotional experiences . Neuroimage. 124 : 455 – 463 . Google Scholar Crossref Search ADS PubMed WorldCat Pashler H . 1988 . Familiarity and visual change detection . Percept Psychophys. 44 : 369 – 378 . Google Scholar Crossref Search ADS PubMed WorldCat Pessoa L , Gutierrez E, Bandettini P, Ungerleider L. 2002 . Neural correlates of visual working memory: fMRI amplitude predicts task performance . Neuron. 35 : 975 – 987 . Google Scholar Crossref Search ADS PubMed WorldCat Robitaille N , Marois R, Todd J, Grimault S, Cheyne D, Jolicœur P. 2010 . Distinguishing between lateralized and nonlateralized brain activity associated with visual short-term memory: fMRI, MEG, and EEG evidence from the same observers . Neuroimage. 53 : 1334 – 1345 . Google Scholar Crossref Search ADS PubMed WorldCat Rouder JN , Morey RD, Morey CC, Cowan N. 2011 . How to measure working memory capacity in the change detection paradigm . Psychon Bull Rev. 18 : 324 – 330 . Google Scholar Crossref Search ADS PubMed WorldCat Ruff CC , Bestmann S, Blankenburg F, Bjoertomt O, Josephs O, Weiskopf N, Deichmann R, Driver J. 2008 . Distinct causal influences of parietal versus frontal areas on human visual cortex: evidence from concurrent TMS-fMRI . Cereb Cortex. 18 : 817 – 827 . Google Scholar Crossref Search ADS PubMed WorldCat Salomon R , Ronchi R, Dönz J, Bello-Ruiz J, Herbelin B, Martet R, Faivre N, Schaller K, Blanke O. 2016 . The insula mediates access to awareness of visual stimuli presented synchronously to the heartbeat . J Neurosci. 36 : 5115 – 5127 . Google Scholar Crossref Search ADS PubMed WorldCat Sauseng P , Klimesch W, Heise KF, Gruber WR, Holz E, Karim AA, Glennon M, Gerloff C, Birbaumer N, Hummel FC. 2009 . Brain oscillatory substrates of visual short-term memory capacity . Curr Biol. 19 : 1846 – 1852 . Google Scholar Crossref Search ADS PubMed WorldCat Seth AK , Friston KJ. 2016 . Active interoceptive inference and the emotional brain . Philos Trans R Soc Lond B Biol Sci. 371 : 20160007 – 20160010 . Google Scholar Crossref Search ADS PubMed WorldCat Shulman GL , Pope DLW, Astafiev SV, McAvoy MP, Snyder AZ, Corbetta M. 2010 . Right hemisphere dominance during spatial selective attention and target detection occurs outside the dorsal frontoparietal network . J Neurosci. 30 : 3640 – 3651 . Google Scholar Crossref Search ADS PubMed WorldCat Todd JJ , Marois R. 2004 . Capacity limit of visual short-term memory in human posterior parietal cortex . Nature. 428 : 751 – 754 . Google Scholar Crossref Search ADS PubMed WorldCat Todd JJ , Marois R. 2005 . Posterior parietal cortex activity predicts individual differences in visual short-term memory capacity . Cogn Affect Behav Neurosci. 5 : 144 – 155 . Google Scholar Crossref Search ADS PubMed WorldCat Tombu M , Jolicœur P. 2005 . Testing the predictions of the central capacity sharing model . J Exp Psychol Hum Percept Perform. 31 : 790 – 802 . Google Scholar Crossref Search ADS PubMed WorldCat Umemoto A , Drew T, Ester EF, Awh E. 2010 . A bilateral advantage for storage in visual working memory . Cognition. 117 : 69 – 79 . Google Scholar Crossref Search ADS PubMed WorldCat Vogel EK , Awh E. 2008 . How to exploit diversity for scientific gain: using individual differences to constrain cognitive theory . Curr Dir Psychol Sci. 17 : 171 – 176 . Google Scholar Crossref Search ADS WorldCat Vogel EK , Machizawa MG. 2004 . Neural activity predicts individual differences in visual working memory capacity . Nature. 428 : 748 – 751 . Google Scholar Crossref Search ADS PubMed WorldCat Vogel EK , McCollough AW, Machizawa MG. 2005 . Neural measures reveal individual differences in controlling access to working memory . Nature. 438 : 500 – 503 . Google Scholar Crossref Search ADS PubMed WorldCat Wilken P , Ma WJ. 2004 . A detection theory account of change detection . J Vis. 4 : 11 – 11 . Google Scholar Crossref Search ADS WorldCat Wright IC , McGuire PK, Poline JB, Travere JM, Murray RM, Frith CD, Frackowiak RS, Friston KJ. 1995 . A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia . Neuroimage. 2 : 244 – 252 . Google Scholar Crossref Search ADS PubMed WorldCat Xu Y . 2007 . The role of the superior intraparietal sulcus in supporting visual short-term memory for multifeature objects . J Neurosci. 27 : 11676 – 11686 . Google Scholar Crossref Search ADS PubMed WorldCat Xu Y . 2017 . Reevaluating the sensory account of visual working memory storage . Trends Cogn Sci. 21 : 794 – 815 . Google Scholar Crossref Search ADS PubMed WorldCat Xu Y , Chun MM. 2006 . Dissociable neural mechanisms supporting visual short-term memory for objects . Nature. 440 : 91 – 95 . Google Scholar Crossref Search ADS PubMed WorldCat Xu Y , Chun MM. 2009 . Selecting and perceiving multiple visual objects . Trends Cogn Sci. 13 : 167 – 174 . Google Scholar Crossref Search ADS PubMed WorldCat Zhang W , Luck SJ. 2008 . Discrete fixed-resolution representations in visual working memory . Nature. 453 : 233 – 235 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2020. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permission@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Gray Matter Volume in Different Cortical Structures Dissociably Relates to Individual Differences in Capacity and Precision of Visual Working Memory
JF - Cerebral Cortex
DO - 10.1093/cercor/bhaa046
DA - 2020-07-30
UR - https://www.deepdyve.com/lp/oxford-university-press/gray-matter-volume-in-different-cortical-structures-dissociably-GLbzjQtGnT
SP - 4759
EP - 4770
VL - 30
IS - 9
DP - DeepDyve
ER -