Revisiting the Landmark Task as a tool for studying hemispheric specialization: What’s really right?

Abstract

The “Landmark Task” (LT) is a line bisection judgment task that predominantly activates right parietal cortex. The typical version requires observers to judge bisections for horizontal lines that cross their egocentric midline and therefore may depend on spatial attention as well as spatial representation of the line segments. To ask whether the LT is indeed right-lateralized regardless of spatial attention (for which the right hemisphere is known to be important), we examined LT activation in 26 neurologically healthy young adults using vertical (instead of horizontal) stimuli, as compared with a luminance control task that made similar demands on spatial attention. We also varied task difficulty, which is known to affect lateralization in both spatial and language tasks. Despite these changes to the task, we observed right-lateralized parietal activations similar to those reported in other LT studies, both at group level and in individual lateralization indices. We conclude that LT activation is robustly right-lateralized, perhaps uniquely so among commonly-studied spatial tasks. We speculate that the unique properties of the LT reside in its requirement to judge relative magnitudes of the two line segments, rather than in the more general aspects of spatial attention or visual-spatial representation.

1. Introduction

It is generally assumed that the brain’s two hemispheres are specialized in different ways (see Hugdahl & Westerhausen, 2010, for an overview). The most common attributions ascribe language to the left hemisphere (Broca, 1861; Wernicke, 1874) and visual-spatial skills to the right hemisphere. Evidence for the right lateralization of visual-spatial skills comes from multiple sources. For example, patients with “split brains,” whose hemispheric connections were severed to reduce seizures, performed better on visual-spatial construction tasks with their left hand (controlled by the right hemisphere) than with their right hand (controlled by the left hemisphere), even though they were right-handed (Bogen & Gazzaniga, 1965). Patients with right-hemisphere lesions (especially those involving the parietal lobe) often exhibit striking visual-spatial impairments, such as hemispatial neglect (Bisiach & Luzzatti, 1978; Vallar & Perani, 1986), apraxia for dressing (Brain, 1941), and marked difficulties in visual-spatial construction tasks (Hecaen et al., 1956). Temporary disruptions of right parietal function by repetitive transcranial magnetic stimulation (rTMS) can induce biases in spatial judgments that mimic those observed in hemispatial neglect (Fierro et al., 2000): when asked to identify the center of a horizontal line, people with disrupted right parietal function deviate to the right side of the line, presumably because they neglect the contralesional left side of space (Schenkenberg et al., 1980). Moreover, neuroimaging studies investigating the neural basis of such line bisection judgments show strong activations in right parietal cortex (Fink et al., 2000; Çiçek et al, 2009; Cavézian et al., 2012).

Among the various tasks that have been used to examine lateralization of function in the spatial domain, line bisection has played an especially prominent role. In part, this is because it appears to regularly engage the right hemisphere and clearly shows effects of disruption due to real or temporary, experimentally induced lesions. Studies seeking to determine whether cerebral lateralization patterns for language and visual-spatial attention are independent or complementary (i.e., whether left-dominance for language coincides with right-dominance for visual-spatial attention and vice versa) use line bisection judgments to probe visual-spatial functions (Flöel et al., 2005; Jansen et al., 2005; Badzakova-Trajkov et al., 2010; Rosch et al., 2012; Cai et al., 2013; see Badzakova-Trajkov et al., 2016, for a review). These studies use variations of the “Landmark Task” (LT, see illustrations in Figure 1), which was introduced by Fink and colleagues (2000) and has been used (often in slightly modified form) by many researchers since. The task is modeled after the line bisection test used to assess hemispatial neglect in clinical settings, which requires participants to judge whether a horizontal line is correctly bisected by a short vertical line, or whether the bisector deviates from the midpoint. In a classic fMRI subtraction design, this spatial condition is usually compared to a control condition that has lower spatial demands but similar visual stimulation and motor response demands. For example, in the control task participants might have to judge whether there is a bisector (Fink et al., 2001, Figure 1A), whether the bisector touches the line (Çiçek et al., 2009, Figure 1B), whether a wavy line is above or below the horizontal line (Flöel et al., 2005b, Figure 1C), or whether the line is red or blue (Longo et al., 2015, Figure 1D). A recent study comparing the robustness and reliability of several fMRI tasks for assessing visuospatial processing concluded that, while not perfect in test-retest reliability of activations at the single-voxel level, the LT was the most robust and reliable, reproducibly determining hemispheric dominance in 93% of participants (Schuster et al., 2017).

Figure 1: Different versions of the Landmark Task.

Illustrations of stimuli used in different versions of the Landmark Task. (A) After Fink et al., 2001. (B) After Çiçek et al., 2009. (C) After Flöel et al., 2005b. (D) After Longo et al., 2015. Note that in all cases the experimental condition requires comparison of the left and right line segments, whereas the control condition does not. Also note that the control task appears easier than the experimental task in all cases, although reaction time data revealing faster responses in the control task were only reported by Fink et al. (2001) and Çiçek et al. (2009).

Notably, different researchers ascribe LT activations to somewhat different visual-spatial functions, using terms that include “(visuo)spatial attention” (Flöel et al., 2005; Cai et al., 2013), “spatial processing” (Jansen et al., 2005, Badzakova-Trajkov et al., 2010), “visuospatial information processing” (Waberski et al., 2008), “spatial awareness” (Marshall et al., 2002), “visuo-spatial function” (Lux et al., 2008), “visual attention”, “visuospatial processing”, and “visuospatial functions” (Rosch et al., 2012). While the difference among these terms may seem slight, we would argue that spatial attention can be characterized as separate from spatial representation. Spatial attention refers to the allocation of attention in space and the prioritization of processing information from that location. In contrast, spatial representation refers to the nature of the spatial content that is encoded. The LT, in its original form with horizontal lines, requires both: to determine whether the line is bisected correctly, one has to compare the lengths of the line segments on either side of the bisector (spatial representation, i.e. content), shifting attention in space between the left and the right line segment (spatial attention).

These joint requirements of the original LT create a problem for interpreting right-lateralized activation. It is well-known that the right hemisphere is involved in allocating attention to both left and right hemispace, whereas the left hemisphere predominantly allocates attention to right hemispace (Heilman & van den Abell, 1980; Mesulam, 1981; Corbetta et al., 1993; Corbetta & Shulman, 2011). Therefore right-lateralized activation might be observed in any task requiring attention to left hemispace, regardless of whether the task requires representation of spatial content. Moreover, shifts of attention in space are associated with parietal cortex in both functional neuroimaging (Yantis et al., 2002) and lesion studies (Posner et al., 1984), so right-lateralized parietal activation during the LT could be driven by shifts in spatial attention between the left and right line segments. Alternatively, right-lateralized parietal activation could be driven by representation of spatial content, that is, by the comparison of the magnitude or spatial extent of the two line segments. Such magnitude comparisons, whether pertaining to line length, quantity, or physical size, are also associated with right parietal activation in monkeys (Tudusciuc & Nieder, 2007) and humans (Faillenot et al., 1999; Pinel et al., 2004; Piazza et al., 2006; Cantlon et al., 2006) and are impaired after right parietal lesions (Warrington & James, 1967).

In the present study we designed a version of the LT that allowed us to focus on activations associated with the encoding of spatial content in the task (line lengths) while minimizing activations associated with spatial attention. To do this, we used a vertical rather than horizontal line for the line bisection judgment task (Space task) and devised a control task (Luminance task) in which participants compared the relative luminance of the top and bottom ends of the line (see Figure 2A for example stimuli). Using a vertical line presented at screen center minimized the need for attention to left hemispace. Using a control task that also requires comparison of the ends of the line ensured that shifts of spatial attention should contribute to fMRI activations for experimental and control tasks to a similar degree.

Figure 2: Experimental Design.

(A) Conditions. Rest periods (gray), during which participants were instructed to simply rest their eyes on a small horizontal line at screen center, were interleaved throughout the functional runs. Each block of an experimental condition was preceded by an instruction period (black) during which the type of the upcoming block (Space or Luminance) was announced in writing. In the Space conditions (hot colors), participants had to indicate whether the top or bottom part of a bisected vertical line was longer; in the Luminance conditions (cold colors), they had to indicate whether the top or bottom tip of a vertical line was brighter. In the Easy instances of the conditions, the top-bottom differences were larger than during the Hard instances of the conditions. (B) Time-course. Each participant completed two functional runs of 5 min duration. In each functional run, 24-s periods of the different experimental conditions were separated by 12-s periods of resting fixation (9 s) followed by an instruction screen (3 s). Space and Luminance conditions were always interleaved. Order of conditions was counterbalanced across the two runs.

We also manipulated task difficulty by varying the length difference between the top and bottom line segments (for the Space task) and the luminance difference between the top and bottom tip of the line (for the Luminance task). The purpose of this additional manipulation was twofold. First, varying difficulty in both tasks allowed us to disentangle general difficulty effects from the (space-specific) activations of interest. This sets our study apart from many others in which the control task may have been easier than the spatial task. While not all of the imaging studies mentioned above report behavioral data, Fink et al. (2000) and Çiçek et al. (2009) reported faster reaction times for their control tasks compared to their experimental tasks. This leaves open the possibility that some of the observed activations were not specific to the spatial task, but instead reflected its higher cognitive demands. In the present study, any brain area specifically involved in spatial processing should show an activation increase with increased task difficulty for the spatial but not the luminance task. Such a space-specific activation increase, in addition to stronger activation in the spatial compared to the luminance condition, would corroborate the brain area’s involvement in spatial processing.

Our second motivation for manipulating task difficulty was that there are multiple reports of lateralized activation becoming more bilateral when tasks become harder. This has been reported for language (Just et al., 1996; Bookheimer, 2002; Xu et al., 2005; Jung-Beeman, 2005), vigilance (Helton et al., 2010), and working memory (Klingberg et al., 1997; Höller-Wallscheid et al., 2017). A similar shift towards bilaterality might occur for LT activations if task difficulty is increased. Indeed, bilateral parietal activation has been observed for some complex spatial tasks, e.g. mental rotation and visual-spatial construction (Cohen et al., 1996; Richter et al., 1997, 2000; Vingerhoets et al., 2002). In our own study of visual-spatial construction, we found bilateral parietal activation, with the strongest left parietal activations in those participants for whom the task was most difficult (Seydell-Greenwald et al., 2017). We thus hypothesized that a similar increase in left parietal activation with an increase in difficulty might occur for the LT, resulting in more bilateral activation with increased task difficulty.

To preview the results obtained with our modified version of the LT, activation remained right-lateralized despite the experimental manipulations. Given these findings, we will conclude that the LT activation observed here is driven predominantly by the content of spatial processing (i.e., comparison of the lengths of the two line segments), not by shifts in spatial attention. As our activations resemble those reported in the literature, the same may be true for LT versions using horizontal lines and control conditions that are less well-matched for difficulty and shifts of attention. The specific areas of activation we obtained resemble those observed in approximate number tasks, suggesting that the task may engage the magnitude estimation system, a point to which we return in our conclusions.

2. Materials and methods

2.1. Participants

Participants were 26 right-handed neurologically healthy young adults (16 female, 10 male) recruited via flyers on the Georgetown campus and word of mouth. Ages ranged from 19 to 37 (mean: 22, se: 0.91). All provided informed consent prior to study participation and were compensated for their time.

2.2. Stimuli and design

The experiment included two 5-minute functional MRI runs, in which five experimental conditions were interleaved in a block design (Figure 2). During Rest, participants were instructed to rest their eyes on a short horizontal line at screen center and to refrain from pushing any buttons. The remaining conditions crossed the factors of Task (Space vs. Luminance) and Difficulty (Easy vs. Hard). In the Space task, participants indicated whether the top or bottom part of a bisected vertical line was longer. In the Luminance task, they indicated whether the top or bottom tip of a (non-bisected) line was brighter. Difficulty was manipulated by increasing (Easy) or decreasing (Hard) the difference between the top and bottom parts of the line.

2.2.1. Stimuli

Visual stimuli were rendered on a PC, projected onto a screen at the back of the MRI scanner bore and viewed by the participants through a slanted mirror mounted on the head coil. Stimuli (examples in Figure 2A) were vertical black lines (length: 8 cm, thickness: 1 mm on the PC, corresponding to a length of 6.4° visual angle and a thickness of 0.08° visual angle given the projection magnification and effective viewing distance). For the Space task, each vertical line was bisected by a short horizontal line (length: 1.28°, thickness: 0.08°). In the Easy Space condition, the bisector was located 0.8° (12.5% of the line length) above or below the veridical line center. In the Hard Space condition, the bisector was located 0.24° (3.8% of the line length) above or below the veridical line center. For the Luminance task, there was no bisecting line, but the uppermost and lowermost tips (1.6° length) of each vertical line were lighter than the rest of the line. One of the tips was white (RGB 255, 255, 255) and the other was gray. In the Easy Luminance condition, the gray tip was much darker (RGB 190, 190, 190), whereas in the Hard Luminance condition, it was just slightly darker (RGB 230, 230, 230) than the white tip. Stimuli were presented on a gray background (RGB 127, 127, 127). To prevent participants from basing their line bisection judgments on the location of the line tips on the screen, the vertical position of the stimuli on the screen was varied randomly within a range of 2.4° around screen center. To prevent participants from comparing lines across trials, a longer, dashed line (length: 12.64°) was presented during the 200 ms inter-stimulus interval to visually mask the preceding stimulus and any potential afterimages. Mask and location randomization were applied to all conditions to keep visual stimulation as similar as possible across tasks.

2.2.2. Stimulus timing and response collection

Participants held two fiber-optic button boxes in their right hand, velcroed together in a vertical arrangement so that there was a “top” and a “bottom” button. During the Space task, they pushed the button corresponding to where they perceived the longer segment of the line to be (above or below the bisector); during the Luminance task, they pushed the button corresponding to which tip of the line they perceived as brighter (top or bottom). Stimuli were drawn randomly on each trial, with the constraint that there could be no more than four “top” or “bottom” stimuli in a row. To ensure that participants worked continuously throughout all task blocks, stimulus presentation was terminated upon response, or after a maximum duration of 2800 ms if no response occurred within that time window. Participants thus completed more trials per block if the condition was easy for them, so that lower activations in easier conditions cannot be attributed to idle time between trials. Successive stimulus presentations were separated by an inter-trial interval of 200 ms during which the experimental stimulus was replaced by the mask as described above but responses were still recorded, leaving participants a total of 3 s to respond.

The experimental conditions were interleaved in a blocked design (Figure 2B), with blocks lasting around 24.5 s (mean: 24.553 ms, std: 438 ms, with exact block duration depending on how long the participant took to respond to the last trial that had begun before the minimum block duration of 24 s had elapsed). Blocks were separated by periods of rest, which lasted around 8.5 s, with the exact duration adjusted so that the previous block and the rest block together took 33 s. Each block was preceded by an instruction screen displayed for 3 s, which read either “Line LENGTH block. Where is the LONGER line?” before Space blocks or “Line COLOR block. Where is the BRIGHTER tip?” before Luminance blocks. During each block, participants completed as many trials as they could, and each correct response earned them a point. At the end of each run, participants were shown their point total for each condition. This was intended to serve as a motivator for continued focus on the subsequent run. Each functional run comprised two blocks of each of the four experimental conditions and nine rest periods, for a run duration of 5 minutes. Each participant completed two functional runs, with condition blocks presented in counterbalanced order.

2.3. Imaging procedures

Neuroimaging was performed on a research-dedicated Siemens Trio Tim 3-Tesla magnetic resonance imaging scanner with a 12-channel birdcage head coil. Participants lay in supine position with their heads at the center of the magnet. They wore headphones mounted in Bilsom ear defenders, which allowed them to hear instructions from the control room while at the same time being shielded from the scanner noise. Visual stimuli were projected onto a screen at the back of the scanner via an Epson PowerLite 5000 projector. Participants viewed the stimuli through a slanted mirror mounted on the head coil. Head position inside the head coil was stabilized by inserting foam pillows between the head coil and the ear defenders. Responses were recorded with two Cedrus fiber optic button response boxes, velcroed together so that participants could hold them in their right hand and operate both buttons by moving their thumb back and forth between them. Stimulus presentation and logging of behavioral responses was handled by E-Prime 2.0.

Scanning sessions began with a 1-minute “Localizer” scan to acquire a low-resolution anatomical scan to aid volume placement for the subsequent scans. Participants then performed two 5-minute functional runs to assess blood-oxygen-level-dependent (BOLD) signal changes associated with the experimental and control conditions (see Figure 2B). Lastly, we acquired a high-resolution anatomical scan (MPRAGE) on which to superimpose the functional data and to aid transformation of individual data into a standard stereotactic coordinate system. Scanning parameters were as follows:

Functional runs (T2*-weighted): Gradient echo-planar imaging (EPI), 50 horizontal slices acquired in descending order, voxel size 3×3×2.8 mm³ with a distance factor of 7% between slices, repetition time (TR) of 3 s, echo time (TE) of 30 ms, flip angle of 90 degrees, matrix 64 × 64, duration 5 minutes (100 volume acquisitions).

MPRAGE (T1-weighted): 176 sagittal slices, voxel size 1×1×1 mm³, TR of 2530 ms, TE of 3.5 ms, inversion time (TI) of 1100 ms, flip angle of 7 degrees, matrix 256 × 256, duration 8 minutes.

2.4. Imaging Analysis

2.4.1. Preprocessing

Imaging analyses were performed with BrainVoyager 20.2 for Mac. Anatomical data underwent inhomogeneity correction, brain extraction, and transformation into Talairach space using 9-parameter affine transformation. Manual corrections were performed where automated brain extraction failed. Landmarks for the Talairach transformation were identified manually. Functional imaging data from both runs underwent the following pre-processing steps: removal of the first two volume acquisitions to allow for T1 saturation, slice-scan time correction, removal of linear trends, 3D motion correction to the first volume of the run using rigid-body transformation, co-registration to the anatomical data using 9-parameter gradient-based alignment, transformation into Talairach space using the same transformation applied to the anatomical data, and spatial smoothing with an 8 mm FWHM Gaussian kernel.

2.4.2. Statistical analysis

For statistical data analysis, voxel time courses from both runs were combined and fitted with a general linear model. The model contained five stimulation-related predictors: one for the Instruction periods, one for Easy Space blocks, one for Hard Space blocks, one for Easy Luminance blocks, and one for Hard Luminance blocks. Rest periods served as the model’s baseline. Each predictor’s time course was convolved with a standard hemodynamic response function (two gamma HRF, time to peak 5 seconds, time to undershoot peak 15 seconds). The model also included the z-transformed motion estimates and a constant predictor for each functional run as nuisance regressors. Analyses were constrained to voxels inside the brain (mask based on anatomical image). Voxel time courses were normalized (percent signal change transformation) and corrected for serial autocorrelations (second-order model). For group-level analyses, all participants’ beta maps were combined into a random effects (RFX) analysis treating each participant’s results as a random sample from a larger population.

Activation maps were thresholded using a single-voxel threshold of p < 0.001 (which was stricter than the single-voxel threshold required to keep the false discovery rate below 5%) in combination with a cluster-size threshold of k < 0.05 (i.e., showing only activation clusters whose size is unlikely to occur by chance in a dataset of similar extent and smoothness as determined by a Monte-Carlo simulation with 1000 iterations, implemented in the “Cluster-Level Statistical Threshold Estimator” plugin for BrainVoyager).

2.4.3. Lateralization indices

To quantify the extent to which brain activation was lateralized, we employed a measure commonly used in studies of hemispheric specialization: the lateralization index (LI). While there are different ways of computing LIs, the basic idea is always the same: Quantify activation on the left and on the right, then compute the (left-right)/(left+right) ratio such that an LI of −1 indicates complete right-lateralization, an LI of 1 indicates complete left-lateralization, and an LI of 0 indicates perfect bilaterality. Because LIs are strongly dependent on the specific threshold at which activation is quantified (Wilke & Lidzba, 2007), we used a bootstrapping approach to compute a weighted mean of LIs obtained at different thresholds (Wilke & Schmithorst, 2006). Specifically, we applied 25 different t-thresholds to our activation maps, spanning the range from 0.1 to the maximum t-value in the map in equal steps. The single-voxel threshold was always combined with a cluster size threshold of 5 voxels, which, in combination with the spatial smoothing applied to the data, should prevent single-voxel activation outliers from distorting the results. LI computation was aborted for thresholds at which there were fewer than 10 active voxels on either side; otherwise, 10,000 LI estimates were computed from the sum of t-values of randomly drawn active voxels on each side (sampling ratio: 0.25), and a trimmed mean (using only the central 50% of estimates) of these 10,000 estimates was used as the robust LI estimate for this threshold. Following LI computation for each threshold, each LI was multiplied with its t-threshold value, such that LIs for stricter thresholds received higher weights. Finally, the sum of all weighted LIs was divided by the sum of all t-thresholds. LI computation was constrained to an anatomically defined bilateral parietal ROI comprising BAs 7, 40, and 39. LI computation and mask generation were accomplished with an in-house script using Matlab R2015b and BVQXtools v0.8d as downloaded from the BrainVoyager support site.

3. Results

3.1. Behavioral

Figure 3 illustrates average accuracy and reaction times for the different conditions. Participants failed to respond within 2800 ms of stimulus onset on only 20 out of 12,010 trials, no participant timed out more than 3 times, 20 of 26 participants never timed out, and across participants who did time out, timeouts were equally common for the Space and the Luminance Task (Easy Space = 1; Easy Luminance = 1; Hard Space = 9; Hard Luminance = 9). Repeated measures ANOVAs for both accuracy and RT on non-timeout trials revealed a main effect of Difficulty (F(1,25) = 58.48 and 194.94, respectively; both p < 0.001). We performed post-hoc paired t-tests for the Space and the Luminance Task, and in both cases found that the Easy condition yielded significantly higher accuracy and faster RTs than the Hard condition (all t(25) > 3.36, all p < 0.003). With respect to Task (Space vs. Luminance), only accuracy revealed a significant main effect (F(1,25) = 24.33, p < 0.001), whereas reaction times did not (F(1,25) = 0.10, p = 0.76). Post-hoc paired t-tests revealed that there were no significant differences for either measure between the Easy instances of the Space and the Luminance task (accuracy: t(25) = 0.58, p = 0.56; reaction time: t(25) = 1.01, p = 0.32), nor between the Hard instances of the Space and the Luminance task for reaction time (t(25) = 0.97, p = 0.34), but accuracy was significantly higher in the Hard Space compared to the Hard Luminance condition (t(25) = 5.33, p < 0.001). This was also reflected in a significant Task x Difficulty Interaction for accuracy (F(1,25) = 29.91, p < 0.001), whereas the interaction term did not reach significance for RT (F(1,25) = 3.84, p = 0.06). The number of trials participants completed in each condition was negatively correlated with RT (r < −0.97 for all conditions) and, due to this near-perfect inverse relationship, is not separately analyzed here.

Figure 3: Behavioral results.

(A) Accuracy and (B) Average Reaction Time in the different conditions. Error bars represent standard error across all 26 participants.

Given these behavioral results, we can be sure that the Easy condition of each task was indeed easier than the Hard condition, so our experimental manipulation worked as intended. In addition, the Luminance task was at least as difficult, if not more difficult, than the Space task, but was neither associated with slower RTs nor with a smaller number of trials. This is important for the interpretation of the fMRI results, as the Space>Luminance activations we report below cannot be driven by difficulty (which, judging from the lower accuracy, was greater for the Hard Luminance compared to the Hard Space condition) or by RTs (i.e., also the number of trials), which was equal for both.

3.2. Imaging

3.2.1. Whole-brain activation maps for the Space vs. Luminance tasks

We first analyzed these data in the way that is traditionally done for the Landmark Task, i.e., by comparing activation in the Space task to that in the control (Luminance) task (Figure 4A, Table 1). Contrasting activation in the Space task (Easy and Hard conditions combined) with that in the Luminance task (Easy and Hard conditions combined) revealed right-lateralized activation in the parietal lobe (both inferior, IPL, BA 40; and superior, SPL, BA 7) and the fusiform gyrus (FG, BA 37). Thus, our results overall agree with the right-lateralized parietal activations commonly reported for the LT (Fink et al., 2000; Foxe et al., 2003; Waberski et al., 2008; Çiçek et al., 2009; Cavézian et al., 2012).

Figure 4: Space > Luminance activations.

(A) Surface views (top) and select horizontal slices (bottom) illustrating activation differences between the Space and the Luminance task, Easy and Hard conditions combined. Compared to the Luminance task, the Space task evoked stronger activation in right inferior and superior parietal lobe (BA 40, IPL; and BA 7 SPL) as well as in the right fusiform gyrus (FG, BA 37). For Talairach coordinates and cluster extent, see Table 1. (B) Compared to the Easy Luminance condition, the Easy Space condition evoked stronger activation in right IPL (BA 40). (C) Increasing task difficulty increased right-sided activation: The Hard Space condition evoked significantly stronger activation in right IPL (BA 40), SPL (BA 7) and FG (BA 37) than the Hard Luminance condition. Activation maps are overlaid on the Colin27 brain template transformed into Talairach space, and thresholded at a p<0.001 single-voxel threshold combined with a k<0.05 cluster-size threshold. Activation details can be found in Table 1, and unthresholded t-maps are available at https://osf.io/m3r5x/.

Table 1.

Activation clusters from the whole-brain analysis

	Peak Talairach coordinates	Center of Gravity	Cluster extent (mm3)	Average t-value *	Average p-value *
Right IPL, BA 40
Space > Lum (Easy and Hard)	51, −28, 43	45, −33, 43	6753	8.76	< 0.000001
Space > Lum (Easy)	57, −35, 43	53, −32, 42	1185	5.67	0.000007
Space > Lum (Hard)	54, −28, 43	45, −33, 44	5305	7.13	< 0.000001
Right SPL, BA 7
Space > Lum (Easy and Hard)	15, −73, 55	17, −72, 52	1037	4.92	0.000046
Space > Lum (Easy)
Space > Lum (Hard)	12, −73, 52	17, −74, 48	2189	5.64	0.000007
Right FG, BA 37
Space > Lum (Easy and Hard)	48, −58, −8	47, −59, −8	1448	4.80	0.000063
Space > Lum (Easy)
Space > Lum (Hard)	45, −55, 52	17, −74, 48	1109	4.70	0.000080

BA - Brodmann area; IPL - inferior parietal lobule; SPL - superior parietal lobule; FG - fusiform gyrus.

^*Note that t- and p-values are inflated, as voxels are in the cluster specifically because they show a significant effect.

Consistent with many other studies, the strongest activations were found in IPL (BA40); however, the peak activation observed here was more lateral and anterior compared to those reported by some others (especially Çiçek et al., 2009; but also Fink et al., 2001; Cavézian et al., 2012; Cai et al., 2013). In contrast to these studies, we compared the Space task with a Luminance comparison task which, in agreement with other studies (Pinel et al., 2004; Cohen Kadosh, 2008), also evoked IPL activation. The Luminance activation’s more medial/posterior peak pushed the peak of the Space > Luminance maps (Figure 4) in the anterior-lateral direction. Compared to baseline, Space task activations peaked at coordinates comparable to those reported in other LT studies (Talairach x, y, z = 39, −43, 43).

SPL activation is less frequently reported in LT studies but has been demonstrated to be stronger if task instructions stress comparison of the relative lengths of the two line segments, rather than asking participants to judge whether the line is bisected at its center (Fink et al., 2002), and if bisection judgments are made for vertical rather than horizontal lines (Fink et al., 2001). The SPL activation reported here is in agreement with these prior observations.

Activation in BA 37 has only been reported in one prior LT study (Fink et al., 2002). This may be due to the fact that task difficulty in other studies may have been lower, thus evoking less activation. In keeping with this, we find no significant BA 37 activation in our study when comparing only the Easy Space and the Easy Luminance conditions (see Figure 4B).

3.2.2. Whole-brain activation maps for the two difficulty levels

Figures 4B and 4C illustrate the effect of increasing task difficulty. For the easy conditions, activation differences between Space and Luminance tasks reached significance only in a relatively small cluster of voxels in right BA 40 (Figure 4B, Table 1); however, for the hard conditions, right BA 40 activation became more extensive, and additional clusters appeared in right BA 7 and right BA 37 (Figure 4C, Table 1). Note that at both difficulty levels, reaction times (and thus number of button pushes) were equivalent for Space and Luminance (Figure 3). Thus, the activation differences shown here cannot be driven by differences in motor activation. Also, as indicated by the accuracy data (Figure 3A), the Luminance task was at least as hard as the Space task at both difficulty levels. We can therefore be confident that the greater activations observed in the Space compared to the Luminance condition are in fact driven specifically by the Space task’s increased demands on spatial processing, rather than on difficulty more generally. Importantly, activations were right-lateralized at both difficulty levels.

3.2.3. Lateralization and activation in an anatomically defined parietal region of interest

While activations certainly look right-lateralized in Figure 4, these are only snapshots of group-level activations at a particular threshold. To obtain a more comprehensive picture of activation lateralization in an unbiased region of interest at the individual participant level, we anatomically defined a parietal ROI (Figure 5A) and computed activation and lateralization indices (LIs) using a method that evaluates the data across a range of thresholds (as described in section 2.4.3). As can be seen in Figure 5B, the left and right hemisphere ROIs showed markedly different activation patterns. Only the right ROI showed significantly stronger activation in the Space compared to the Luminance conditions (t(25) = 2.31, p = 0.030 for Easy Space vs. Easy Luminance; t(25) = 4.71, p < 0.001 for Hard Space vs. Hard Luminance), whereas the left ROI did not (both t (25) < 1.34, both p > 0.19). In addition, only the right ROI showed a significant increase in activation from the Easy to the Hard Space condition (t(25) = 5.34, p < 0.001), whereas the left showed a non-significant decrease (t(25) = −1.47, p = 0.15).

Figure 5: Parietal ROI analysis.

(A) A bilateral posterior parietal ROI was anatomically defined and encompassed Brodmann areas 7, 40, and 39. (B) Activation was determined in a threshold-independent way as the weighted sum of active voxels across thresholds. Only the right ROI showed significantly stronger activation for the Space compared to the Luminance conditions at both difficulty levels, and only the right ROI showed an activation increase from the Easy to the Hard Space task. (C) Lateralization indices (indicated by ‘x’s) were negative (indicating right-lateralization) for most participants, and group means (black circles, shown with standard error bars) were significantly below zero for all contrasts, indicating stronger right-lateralization for the Space compared to the Luminance conditions at both difficulty levels.

Lateralization indices (Figure 5C) were negative (indicating right-lateralization) for the vast majority of participants, and group means were significantly different from zero for all contrasts (Hard Space > Hard Luminance: t(25) = −5.17, p < 0.001; Easy Space > Easy Luminance: t(25) = −3.80, p < 0.001; Hard Space > Easy Luminance: t(25) = −10.55, p < 0.001). A comparison of LIs for the different contrasts using a repeated-measures t-test revealed no significant differences in lateralization between Easy Space > Easy Luminance and Hard Space > Hard Luminance (t(25) = −0.50, p = 0.62), indicating that activation did not become more bilateral as the difficulty of the Space task increased.

4. Summary and Discussion

The Landmark Task (LT), a neuroimaging version of the line bisection task commonly used to diagnose hemispatial neglect, has often been used to investigate lateralization of visual-spatial functions and/or visual-spatial attention. The task is known to elicit right-lateralized activation, but because most studies have used horizontal lines as stimuli and have held difficulty levels fairly low and constant to ensure high performance, it is unclear to what extent lateralization may be influenced by spatial attention and/or effort, rather than the spatial computations underlying the required line bisection judgments. In the present study we investigated which aspects of the task contribute to the pattern of right lateralization by carefully controlling individual task parameters.

First, we asked whether right-lateralization of LT activation would persist when using a design that minimized activation related to spatial attention, which is known to activate the right hemisphere. We did this by using vertical lines (instead of horizontal) and by using a control task that, like the bisection judgment task, required comparing the top and bottom line segments, but regarding luminance rather than length. Thus, while spatial attention shifts were needed, their influence on the fMRI results is minimized because they were necessary in both the experimental and the control condition, and right-lateralization of findings cannot be attributed to the need for attention to left hemispace. In addition, we manipulated the level of difficulty of the LT (and control task) to determine whether right-lateralization persists as the task becomes more difficult, since previous literature has shown that some complex tasks show more bilateral activation (Just et al., 1996; Cohen et al., 1996; Klingberg et al., 1997; Richter et al., 1997, 2000; Vingerhoets et al., 2002; Bookheimer, 2002; Xu et al., 2005; Jung-Beeman, 2005; Helton et al., 2010; Seydell-Greenwald et al., 2017; Höller-Wallscheid et al., 2017).

Despite these task changes we still observed persistent and strong right-lateralization for the LT task. Thus, consistent with many other investigations of the LT, we observed right-lateralized parietal activations when contrasting a Spatial task (line bisection judgment) with a non-spatial control (luminance judgment). We observed this both at the group level (Figure 4) and at the individual level for most participants (Figure 5B), and over both levels of difficulty. Our results mirror those reported in studies of traditional horizontal line bisection (Fink et al., 2000, 2002; Foxe et al., 2003; Waberski et al., 2008; Çiçek et al., 2009; Cavézian et al., 2012) as well as one that tested vertical line bisection (Fink et al., 2001).

Our findings are most closely related to and expand upon two other studies. Rosch et al. (2012) studied the effect of difficulty increases on LT lateralization using functional transcranial Doppler ultrasound (fTCD) and also found right-lateralization at both difficulty levels. However, fTCD cannot localize findings to any particular brain area. Moreover, because the study assessed lateralization relative to a passive “clear mind” baseline rather than to a difficulty-matched non-spatial control condition, it is unclear to what extent the results were driven by spatial attention, visual-spatial processing, response preparation, or general executive function demands. The results of our experiment allow us to confirm right parietal activations as the site of the lateralization effect. In addition, because we included a control task that was identical to the LT except for the content of the comparison judgment, we can also rule out factors such as spatial attention, motor responses, and general executive function demands as explanations for the activation patterns. The other closely related study, Fink et al. (2001), directly compared fMRI activation to the LT performed on horizontal vs. vertical lines within the same participants. They observed right-lateralized activation in posterior parietal cortex irrespective of stimulus orientation, with no significant interactions between task (bisection judgment vs. control) and stimulus orientation (horizontal vs. vertical). They concluded that “orientation did not differentially affect the neural mechanisms underlying the visuospatial judgment per se.” We fully agree with this conclusion and now turn to discussing why the bisection task does so reliably elicit right-lateralized activation.

As we noted in the Introduction, several studies have observed bilateral parietal activations for complex visual-spatial tasks, including mental rotation and spatial construction (Cohen et al., 1996; Richter et al., 1997, 2000; Vingerhoets et al., 2002; Seydell-Greenwald et al., 2017). These findings show that not all “spatial” tasks are necessarily right-lateralized. In contrast, we have shown here that the LT does evoke right-lateralized activation, even with vertical stimuli and regardless of task difficulty. Our findings converge with those of many other studies that have shown robust right-lateralized activity for the LT with horizontal stimuli. The LT thus appears to be a clear case of a spatial task that is truly right-lateralized. Given that not all spatial tasks are reliably right-lateralized, a remaining question is why this particular task drives such a strong right lateralization profile.

We speculate that the LT, while easily construed as a “spatial” task, actually engages a parietal-based general purpose magnitude estimation system that has been shown to be involved in processing not only spatial extent, but also temporal extent and numerosity (for reviews, see Walsh, 2003; Hubbard et al., 2005; Dehaene & Brannon, 2011; Cantlon, 2012). Consider the requirements of the LT task. In our task, participants were required to compare the length of the two line segments on either side of the bisector and then indicate which was longer (top or bottom). Thus it comprises two “spatial processing” aspects: a comparison of magnitude or spatial extent of the two line segments, and localization of one of them relative to the other or to the bisector (is the longer segment above or below the bisector). Because our control task also required localization (which line segment has the brighter tip), the activation observed when contrasting the two tasks is unlikely to reflect localization alone. In addition, our results mirror those of other LT studies, many of which did not require localization of the longer line segment, but simply asked whether the line was correctly bisected. What is common to the experimental conditions of all LT studies is the need to compare the spatial extent (magnitude) of the two line segments in the experimental condition (but not in the control condition). This suggests that magnitude comparison is the source of the observed parietal activation¹.

The existence of a general magnitude system has ample experimental support (e.g. Walsh, 2003; Hubbard et al., 2005; Dehaene & Brannon, 2011; Cantlon 2012). Faillenot et al. (1999) and Pinel et al. (2004) report right-lateralized parietal activations when people compare the physical size of stimuli, and several neuroimaging studies have shown right-lateralized parietal activations for approximate number tasks (Piazza et al., 2006; Cantlon et al., 2006). A right-hemisphere dominance for magnitude estimation is also consistent with behavioral studies using tachistoscopic presentation of stimuli to either hemisphere, which show a left visual field (right hemisphere) advantage (Kimura, 1966; McGlone & Davidson, 1973), and by the neuropsychological finding that lesions to the right parietal cortex, more than anywhere else, result in impairments of numerosity estimation (Warrington & James, 1967). Further support for the idea that line bisection and number sense tap the same general magnitude estimation system comes from the observation that right parietal lesions result in characteristic biases to the right not only in line bisection, but also when participants are asked which number lies halfway between two others (e.g., patients with hemispatial neglect might state that the number halfway between two and six is five – Zorzi et al., 2002). Lastly, a study with rhesus monkeys showed that line length and numerosity comparisons activate overlapping populations of parietal neurons (Tudusciuc & Nieder, 2007).

5. Conclusion and Implications

We conclude that the Landmark Task indeed evokes right-lateralized parietal activations, supported by evidence both at the group and at the single-participant level, and that this pattern of activation is robust to variations in stimulus orientation and task difficulty. We speculate that it reflects engagement of a right-lateralized magnitude estimation system that is also activated in other tasks requiring comparisons of spatial or temporal extent as well as numerosity. It is noteworthy that this is not true of all visual-spatial tasks: Although magnitude estimation is crucial for visual-spatial tasks such as line bisection or size comparisons, it is neither necessary nor sufficient for others, such as mental rotation or block construction. Thus it should come as no surprise that some complex visual-spatial tasks are more bilateral, that they do not share the strongly right-lateralized activation pattern observed for the LT, and that performance on these tasks can be affected by left as well as by right parietal lesions.

These findings have implications for using the Landmark Task as a tool for investigating hemispheric specialization. For example, its robust right-lateralization in the majority of neurologically healthy adults and its consistency over task difficulty make it an excellent tool for probing atypical brain organization following focal brain injury, as well as for tracking developmental changes in lateralization among healthy children. However, researchers and clinicians should keep in mind that the robust and consistent pattern of right-lateralization for the Landmark Task may be rather unusual among those tasks commonly considered to be spatial.

Highlights.

• The Landmark Task is often used to probe lateralization of visual-spatial functions

• Most LT versions confound line length judgment with spatial attention

• Our LT assessed line length judgment only; activation remained right-lateralized

• LT activation was in parietal regions also associated with magnitude judgment

• LT lateralization to RH parietal regions reflects line length comparisons