Skip to main content
NCBI home page
Brain logoLink to Brain
. 2013 Feb 11;136(3):751–760. doi: 10.1093/brain/awt003

Parietal substrates for dimensional effects in visual search: evidence from lesion-symptom mapping

Sandra Utz 1,2,✉, Glyn W Humphreys 3, Magdalena Chechlacz 2,3
PMCID: PMC3580271  PMID: 23404335

Abstract

In visual search, the detection of pop-out targets is facilitated when the target-defining dimension remains the same compared with when it changes across trials. We tested the brain regions necessary for these dimensional carry-over effects using a voxel-based morphometry study with brain-lesioned patients. Participants had to search for targets defined by either their colour (red or blue) or orientation (right- or left-tilted), and the target dimension either stayed the same or changed on consecutive trials. Twenty-five patients were categorized according to whether they showed an effect of dimensional change on search or not. The two groups did not differ with regard to their performance on several working memory tasks, and the dimensional carry-over effects were not correlated with working memory performance. With spatial, sustained attention and working memory deficits as well as lesion volume controlled, damage within the right inferior parietal lobule (the angular and supramarginal gyri) extending into the intraparietal sulcus was associated with an absence of dimensional carry-over (P < 0.001, cluster-level corrected for multiple comparisons). The data suggest that these regions of parietal cortex are necessary to implement attention shifting in the context of visual dimensional change.

Keywords: visual search, dimension weighting, brain lesions, voxel-based morphometry

Introduction

In traditional visual search experiments, participants are required to detect the presence of a predefined target item (e.g. a red bar) surrounded by varying numbers of irrelevant distracter items (e.g. green bars). Müller and colleagues. (Müller et al., 1995; Found and Müller, 1996; Müller and O’Grady, 2000) observed that in such tasks, reaction times are significantly faster when the target was defined within the same dimension in consecutive trials (e.g. odd one out in red → odd one out in red) compared with when the dimension of the target changed (e.g. odd one out in red → odd left-tilted item). In contrast with this, a change of the feature value within one dimension (e.g. from a right- to a left-tilted target) did not affect reaction times (although see Maljkovic and Nakayama, 1994; Wolfe et al., 2003, for evidence that feature repetition can also affect search). Treisman (1988) investigated several different feature targets. Participants either knew in advance which target would be presented or not. Relative to performance when there was a single known target (e.g. a blue bar in the colour condition), there were few costs to performance when the disjunctive targets were all defined within the same dimension (e.g. a blue, red or white bar among green bars), but there were costs when the disjunctive targets were defined by values on different dimensions. Following such results, Müller et al. (1995) and Found and Müller (1996) proposed that in visual search, participants could selectively attend to the dimension defining the target (the target’s dimension could be allocated ‘attentional weight’; see Duncan and Humphreys, 1989; Bundesen, 1990). When consecutive targets were defined along the same dimension, there did not need to be any change in the attentional weights for detection to take place. In contrast, reaction times were slowed to targets defined along different dimensions due to the need for attentional weights to be shifted from one dimension to the other.

Dimensional effects on visual search are assumed to occur early in visual processing (Müller et al., 1995; Töllner et al., 2008), enhancing the accrual of information from targets when the target-defining dimension is carried over across trials relative to when it changes. The electrophysiological correlates of these dimensional carry-over effects in search were investigated by Gramann et al. (2007). On dimensional change trials, the amplitude of the fronto-centrally distributed N2 component was increased relative to dimensional repeat trials. This systematic variation likely reflects the rapid detection of dimensional changes. The P3 component was also delayed on dimension change trials. Gramann et al.’s (2007) results support the idea that relatively early, perceptual processes (N2), as well as later attention-related processes (P3), are associated with dimensional change (see also Töllner et al., 2008 for confirmatory evidence for effects on N2). Gramann et al. (2010) further investigated the modulation of early visual processes through dimension-based attention using a singleton search task. The search display was preceded by an identical cue array with a spatially (not dimensionally) predictive singleton. In addition to replicating the modulation of the N2 component, Gramann et al. (2010) found modulations in the (early) P1 component. P1 amplitudes were enhanced for targets defined in the same dimension as the cue (independent of whether the target carried the same or a different feature). The enhanced amplitude of the P1 component is consistent with attentional weight being applied to early visual input modules (Gramann et al. 2010).

Pollmann et al. (2000) using functional MRI found that in comparison with trials where the target dimension remained the same in search, dimension-change trials were associated with increased activation in the left fronto-polar cortex and the right superior parietal lobule and intraparietal sulcus, as well as along the cuneus/precuneus border in the right hemisphere. Increased activation was also found in the left middle-temporal gyrus and anterior prefrontal areas (in the fronto-polar cortex) along with the supramarginal gyrus. Also, when a dimensional change occurred, the activation in the striate cortex was decreased in comparison with no-change trials. Pollmann et al. (2000) argued that attentional shifts between different dimensions are controlled by the fronto-polar cortex.

In a subsequent functional MRI experiment, Weidner et al. (2002) used both (i) a conjunction search task with targets defined by a unique combination of size and either colour or motion direction; and (ii) a feature search task in which targets were defined by a simple feature contrast in one of the dimensions. Changes of the secondary target dimension (colour or motion) across trials in the conjunction task increased reaction times to a higher degree than in the feature search task. The left fronto-polar cortex showed increased activation when stimulus-driven dimensional changes occurred in the feature search task, while fronto-medial cortex increased activity when dimensional changes occurred in conjunction search (Weidner et al., 2002).

Though the functional MRI studies implicate both prefrontal and posterior parietal brain regions as being important for dimensional carry-over effects in search, the results are correlatory in nature and do not demonstrate the necessary role of any implicated brain regions. Our understanding of the brain regions necessary for dimensional carry-over effects can be enhanced by study of the effects of brain lesions, which can show behavioural changes linked to the site of the lesion (lesion-symptom mapping). In the current study, we use lesion-symptom mapping to evaluate the brain areas necessary for dimensional carry-over. Specifically, the approach we take is to use voxel-based morphometry, where we evaluate the correlations between behavioural changes and alterations to individual brain voxels in an unselected group of patients. In voxel-based morphometry, patients without the behavioural deficit can serve as controls for patients where there is behavioural change, controlling for effects of damage due (e.g.) to the vascular territory that might be affected. Here, we evaluate which brain areas are necessary for dimensional carry-over effects by testing for lesion-symptom correlations in our unselected patient sample.

Previously, using the lesion-symptom mapping approach, Pollmann et al. (2007) tested the hypothesis that the left fronto-polar cortex is involved in shifts of attention between different dimensions by testing patients with left fronto-polar cortex lesions in comparison with patients with fronto-medial lesions. Dimensional changes caused significantly higher costs in patients with left fronto-polar cortex lesions. These results support the idea of the left fronto-polar cortex mediating dimension weighting in feature search. This positive evidence for an involvement of frontal cortex is in contrast with negative neuropsychological evidence on the involvement of other regions. For example, Kristjánsson et al. (2005) reported that dimensional carry-over effects in search were preserved in two patients with left neglect after damage to the inferior parietal lobule and superior temporal gyrus. This result is potentially important, since it suggests that the posterior parietal cortex may not play a general role in modulating non-spatial (feature-based) as well as spatial visual attention (Kanwisher and Wojciulik, 2000), but, given that only two patients were tested, this conclusion should be tentative.

In the present study, we took a different approach to examining neural substrates in shifting attention across stimulus dimensions. Rather than testing patients with specific brain lesions, we examined an unselected group of patients and distinguished them according to whether they did or did not demonstrate a dimensional carry-over effect in search.

We applied observer-independent voxel-based morphometry based on a robust statistical method within the framework of a general linear model (Ashburner and Friston, 2000) to assess the association between behavioural performance in visual search and the brain lesions. Our study differs from previous patient reports in terms of both the approach to image analysis that we adopt (voxel-based morphometry versus simple lesion comparison) and our use of an unbiased patient sample (patients were not pre-selected based on any clinical, anatomical or neuropsychological criteria), and in using this sample, we were able to examine common structure–function relationships across the whole brain. Furthermore, the whole brain voxel-based morphometry analysis provided us with the opportunity to control in our analyses for age-related changes (reflecting the level of atrophy), lesion volume and presence of visuospatial attention deficits across the patient sample, all factors that could mediate performance in individual patients.

Materials and methods

Participants

Patients

Twenty-five patients participated (22 males and three females), with ages ranging from 37 to 79 years (mean = 64.44 years; SD = 12.40 years). All patients had acquired brain lesions (22 stroke, two carbon monoxide poisoning, one aneurism). Omitting the patients with carbon monoxide poisoning made little difference to the results, while including them enables us to generalize across different aetiologies. Patients were at a chronic stage (>9 months post-injury). All were recruited from the panel of neuropsychological volunteers established in the Behavioural Brain Sciences Centre at the School of Psychology, University of Birmingham, and all participated in the Birmingham Cognitive Screen (BCoSTM; Humphreys et al., 2012). Clinical, demographic and dimensional effect data for all the patients are shown in Table 1. Spatial deficits were assessed using the Apple Cancellation and the Visual Extinction test of the Birmingham Cognitive Screen (Humphreys et al., 2012). All participants provided written informed consent in agreement with ethics protocols at the School of Psychology and Birmingham University Imaging Centre. All patients had normal or corrected-to-normal vision. Some of the patients had prior experience with visual search tasks but were naïve as to the purpose of this experiment.

Table 1.

Patient details: clinical and demographic data

Patient ID Sex/age/handedness Aetiology Spatial deficits Lesion side Dimensional effects
PM M/69/L S L B 0
MH M/56/R CM R B 0
RG M/60/R S R B 0
MD M/78/R S R L 0
MP M/63/L A L R 0
AK M/75/R S – B 0
TM M/73/R S L R 0
JB F/75/R S L R 0
PH M/38/R S R L 0
JW M/78/R S L R 1
CM F/38/R S – B 1
JE M/62/L S – L 1
GA M/56/R HSE – B 1
FL M/76/R CM – B 1
JQ M/55/R S L R 1
PF F/61/R S L R 1
RH M/77/L S R L 1
DT M/69/R S – L 1
JQ M/65/R S – R 1
RP M/55/R S L R 1
TH M/63/L S R L 1
MC M/76/R S L R 1
DS M/78/R S – L 1
MH M/79/R S – B 1
ST M/54/R S L R 1
Open in a new tab

A = aneurism; B = bilateral; CM = carbon monoxide poisoning; F = female; HSE = herpes simplex encephalitis; L = left; M = male; R = right; S = stroke.

Healthy control subjects

We also acquired T1-weighted images from 100 healthy control subjects (55 males and 45 females, mean age 54.5 years, range 20–87 years) with no history of stroke, brain damage or neurological disorders. The structural scans from the healthy control subjects were used to estimate lesion volume in the patients. All the controls provided written informed consent in agreement with ethics protocols at the School of Psychology and the Birmingham University Imaging Centre.

A small group of six control participants (three males and three females) with ages ranging from 58 to 80 years (mean = 72 years; SD = 9.06 years) was also tested with the visual search task.

All control subjects had normal or corrected-to-normal vision. Some of the controls had prior experience with visual search tasks but were naïve as to the purpose of this experiment.

Cognitive assessment

Behavioural task: dimensional effects

Stimuli were presented on a black background on a 17-inch monitor with an 800 × 600-pixel screen resolution. Patients were positioned ∼70 cm from the screen. Responses were recorded from a keyboard. Only right and left arrow keys were used to enable patients to perform the task with one hand.

The display consisted of a 5 × 5 or 6 × 6 matrix of bar stimuli. Each bar subtended an area of ∼1.34° of visual angle in height and of ∼0.34° of visual angle in width. The bars in the matrix were slightly jittered with the horizontal distance between neighbouring bars varying between 1.7° and 3.56° of visual angle and the vertical distance between 0.98° and 2.33° of visual angle. Target bars were either defined by colour (red vertical or blue vertical) or by orientation (45° tilted to the left, 45° tilted to the right). Distractor items were green vertical bars. Targets were presented at one randomly selected location of the inner 4 × 4 matrix of the 5 × 5 matrix or the inner 5 × 5 matrix of the 6 × 6 matrix to avoid edge effects (i.e. targets at the edges of the display are selected more slowly than targets at the centre of the display).

Within blocks, the trials were presented in a pseudo-randomized order to ensure that trials from each condition were preceded equally often by a trial from each of the other conditions. Participants completed four sessions of 12 blocks, each consisting of 40 trials resulting in 1920 trials in total. At the start of each session, 24 practice trials were presented containing all possible target types and display sizes. Data from this block were then discarded. Each session consisted of four cross-dimension search blocks, which contained both colour- and orientation-defined targets. Of the remaining eight within-dimension search blocks, half contained only colour-defined targets and the other half only orientation-defined targets. In consecutive target present trials, targets were either defined by the same dimension and the same feature (red → red; same dimension same feature), by the same dimension but a different feature (blue → red; same dimension different feature) or by different dimensions (blue → orientation; different dimensions). To take account of the spatial deficits in the patients, we collected enough trial pairs for each of the three different intertrial transition conditions to assess performance separately in the left and right hemifields. On 80% of the trials in the within (same dimension same feature, same dimension different feature) as well as in the cross dimension (different dimensions) blocks, a target was present; on 20%, it was absent. For each condition (same dimension same feature, same dimension different feature and different dimension), at least 50 trial pairs were presented.

A trial started with the presentation of a centrally presented fixation cross for 500 ms, followed by a blank for 500 ms and then the simultaneous presentation of either 25 or 36 bars. The display stayed on the screen until the observer responded. During each session, observers had to press one arrow key for target present and the other arrow key for target absent responses. We did not change the response key between blocks (only between sessions) to avoid confusing the patients. A white fixation hash was displayed during the intertrial interval, which lasted either 800, 1000 or 1200 ms. During the practice block, feedback was provided about the accuracy of the answer.

Visuo-spatial attention and working memory battery

In order to measure the visuo-spatial deficits associated with spatial neglect and visual extinction, we used an assessment based on tests from the Birmingham Cognitive Screen battery including the Apples Cancellation task (a measure of neglect), the Visual Extinction Test and an Auditory Attention task (providing measures of working memory and sustained attention) (Chechlacz et al., 2010; Bickerton et al., 2011; Humphreys et al., 2012). Patients were classed as having a clinical deficit in spatial neglect and visual extinction if their scores on the Apples and Visual Extinction tests fell outside the norms for the tests taken from 86 control participants with no history of neurological diseases (35 males and 51 females, mean age 67 years, range 47–88 years). The resulting spatial deficits of the patients are listed in Table 1. Visual working memory of all patients was tested using the Corsi block test (Corsi, 1972), and sustained attention was furthermore measured with the digit span forward and backward (contrasting backward against forward span; see Robertson, 1990).

Neuroimaging assessment

All patients and healthy control subjects were scanned at the Birmingham University Imaging Centre on a 3 T Philips Achieva MRI system with 8-channel phased array SENSE head coil. We acquired all anatomical scans using a sagittal T1-weighted sequence (echo time/repetition time = 3.8/8.4 ms, voxel size 1 × 1 × 1 mm3).

Image preprocessing

All T1 scans (both from patients and healthy control subjects) were first converted and reoriented using MRICro (Chris Rorden, Georgia Tech). This was followed by image preprocessing with Statistical Parametric Mapping 5 (SPM, Wellcome Department of Cognitive Neurology, London, UK). All scans were then transformed into the standard MNI space using the unified-segmentation procedure (Ashburner and Friston, 2005). The unified-segmentation protocol includes tissue classification based on the signal intensity in each voxel as well as on a priori knowledge of the expected localization of grey matter, white matter and CSF in the brain. To further improve tissue classification and spatial normalization of damaged brains, we used a modification in the segmentation procedure based on the recently published approach by Seghier et al. (2008). This method was specifically designed to resolve problems with misclassification of damaged tissue by including an additional prior for an atypical tissue class (an added ‘extra’ class) to account for the ‘abnormal’ voxels within lesions and thus allowing classification of the outlier voxels. While earlier versions of SPM struggled with normalizing and segmenting brains containing large lesions (e.g. Stamatakis and Tyler, 2005), the unified-segment procedure as implemented in SPM5 has been shown to be optimal for spatial normalization of lesioned brains (Crinion et al., 2007). The segmented images were next smoothed with 8-mm full-width at half-maximum Gaussian filter to accommodate the assumption of random field theory used in the statistical analysis (Worsley, 2003). The choice of intermediate smoothing of 8 mm full-width at half-maximum was previously shown to be optimal for lesion detection and further analysis of segmented images (Seghier et al., 2008; Leff et al., 2009).

The preprocessed grey matter and white matter images were used for automated lesion definition using fuzzy clustering (Seghier et al., 2008). Furthermore, segmented grey matter images were entered into voxel-based analyses to determine the neural substrates of dimensional effects. Previous studies (Leff et al., 2009; Price et al., 2010; Chechlacz et al., 2010, 2011, 2012) have demonstrated that the modified segmentation protocol combined with voxel-based morphometry is successful in facilitating the understanding of brain behaviour relationships in neurological patients.

Automated lesion identification

Lesion maps from individual patients were reconstructed using a modified segmentation procedure (as described above) and an outlier detection algorithm based on fuzzy clustering (see Seghier et al., 2008 for a full protocol as method validation based on real and simulated lesions on T1-weighted scans). Seghier et al.’s (2008) approach identifies voxels that are different in the lesioned brain compared with a set of healthy control subjects (here we used a set of 100 healthy control subjects) using normalized grey and white matter segments. The grey matter and white matter outlier voxels were then combined into a single outlier image and thresholded to generate a binary map of the lesion. The results of lesion reconstruction were verified against the patient’s T1 scans. The lesion volumes for each patient were calculated using Matlab 7.5 (The MathWorks) based on individual lesions from automated lesion-identification procedure. The estimated lesion volumes of all individual patients were used as covariates in the statistical analyses (see below).

Voxel-based morphometry

Grey matter maps from all patients, obtained from their segmented T1 scans, were used in further statistical analyses with SPM5, to assess the relationship between grey matter damage and dimensional effects on a voxel-by-voxel basis based on the voxel-based morphometry approach (Ashburner and Friston, 2000). We used parametric statistics within the framework of the general linear model (Ashburner and Friston, 2000; Kiebel and Holmes, 2003) to carry out all analyses. To ensure that we could control for various confounding factors that could potentially affect cognitive performance, each statistical model included age, handedness, gender, time since diagnosis, type of lesion and lesion volume as covariates. The analysis included all 25 patients, and we compared 16 patients showing dimensional carry-over effects with nine patients who showed no dimensional effects at all. We controlled for spatial deficits in our patients, as poor performance on the task examining dimensional effects could be affected by either visual extinction or visual neglect, both of which may co-vary with our main task of interest. Furthermore we controlled for deficits in working memory and sustained attention, which co-vary with the neglect syndrome (e.g. Malhotra et al., 2005). We report only results that showed significant effects at P < 0.001 family-wise error (FWE)-corrected threshold at the cluster level with amplitude of voxels surviving P < 0.001 (uncorrected) across the whole brain and an extended threshold of 800 mm3 (>100 voxels). The anatomical localization of the lesion sites was based on the Duvernoy Human Brain Atlas (Duvernoy et al., 1991) and the Brain Atlas by Woolsey et al. (2008). The brain coordinates are presented in the standardized MNI space (Evans et al., 1993).

Results

The data were analysed by assessing the data on the non-lesioned control group relative to coarsely defined patient groups (based on coarse lesion descriptors: unilateral left, unilateral right, bilateral). Subsequently we classified the patients according to whether they showed a dimensional carry-over effect in their search behaviour, and we then used this classification in the voxel-based morphometry analysis.

Behavioural performance

Since no differences in reaction times and errors could be found between the two different set sizes (25 or 36 items) (P > 0.05), the data were analysed together.

Data for the two within-dimension conditions (same dimension same feature and same dimension different feature) were taken from the within-dimension blocks and data for the dimensional change condition (different dimensions) were taken from the cross-dimensional blocks. Firstly, errors and reaction times were analysed with a repeated measures ANOVA with two within-subject factors, intertrial transition (same dimension same feature, same dimension different feature and different dimensions) and the two hemifields (left and right); and one between-subjects factor, lesion (unilateral left, unilateral right and bilateral). The patient group consisted of seven patients with unilateral left, 10 patients with unilateral right and eight patients with bilateral lesions.

Accuracy

Control group

The overall error rate of the non-lesioned control group was 0.52% (SD = 0.91). Hemifield, F(1,5) = 0.01; P = 0.940; partial η2 = 0.001, and intertrial transition, F(2,10) = 3.53; P = 0.069; partial η2 = 0.414, did not have a significant effect on the error rates of the control group. Furthermore, the factors did not interact, F(2,10) = 0.21; P = 0.815; partial η2 = 0.040.

Coarsely-defined patient groups

Overall error rates were 2.51% (SD = 5.14%). There was neither a significant effect of hemifield, F(1,22) = 3.88; P = 0.062; partial η2 = 0.150, nor of the different intertrial transitions on error rates, F(2,44) = 0.68; P = 0.511; partial η2 = 0.030, and there were no overall differences between the patient groups when defined coarsely by lesion (unilateral left, unilateral right and bilateral), F(2,22) = 0.63; P = 0.544; partial η2 = 0.054. There were no interactions, all P-values > 0.050.

Reaction times

Control group

The analysis of the control group’s reaction times (Fig. 1) showed a significant effect of the different intertrial transitions, F(2,10) = 38.17; P < 0.001; η2 = 0.884. Reaction times were significantly slower in the dimensional change condition (mean = 667.58 ms; SD = 74.08 ms) in comparison with both within-dimension conditions [same dimension same feature: mean = 592.14 ms; SD = 57.06 ms; same dimension different feature: mean = 610.39 ms; SD = 61.14 ms; F(2,4) = 21.68; P = 0.002; η2 = 0.916]. The mean reaction time difference between the within-dimension conditions and cross-dimension condition was 65 ms. The data are presented in Fig. 1.

Figure 1.

Figure 1

Open in a new tab

Reaction times of the control group are plotted separately for both hemifields as a function of the different intertrial transitions. RTs = reaction times; sdsf = same dimension same feature; sddf = same dimension different feature; dd = different dimensions.

Coarsely-defined patient groups

There was a significant effect of hemifield, F(1,22) = 9.26; P = 0.006; partial η2 = 0.296, and of the different intertrial transitions (same dimension same feature, same dimension different feature, different dimensions) on reaction times, F(2,44) = 14.19; P < 0.001; partial η2 = 0.392. There was no significant main effect of lesion, F(1,22) = 0.47; P = 0.632; partial η2 = 0.041. The factors hemifield and lesion interacted significantly, F(2,22) = 6.51; P = 0.006; partial η2 = 0.372. Patients with right lesions showed faster reaction times for targets in the right hemifield (mean = 827.62 ms; SD = 258.79 ms) compared with the left hemifield (mean = 990.89 ms; SD = 344.99 ms). Patients with bilateral or left lesions did not show reaction time differences between targets in the two hemifields. There were no other interactions, all P-values > 0.05.

Redefining the patient groups

An inspection of individual data sets indicated that a subset of patients showed no dimensional effects on reaction times (n = 9; mean age = 65.00 years; SD = 12.98 years), while another subset did show significant dimensional carry-over effects (n = 16; mean age = 64.13 years; SD = 12.49 years).

A cluster-analysis revealed that a two-cluster model best fit the data. Cluster 1 described the subgroup with significant dimensional carry-over effects and Cluster 2 the subgroup showing no dimensional effects at all (see dendrogram in the Supplementary material).

In the group not showing dimensional effects, there were two left, three right and four bilateral patients. In the group showing effects, there were five left, seven right and four bilateral patients. To statistically confirm our classification of the patients, we conducted a repeated-measures ANOVA with the within-subject factors being hemi-field (left and right) and intertrial transition (same dimension same feature, same dimension different feature and different dimensions), and with a between-subject factor of dimensional carry-over group (dimensional effect and no dimensional effect). There were significant effects of hemifield, F(1,23) = 8.89; P = 0.007; partial η2 = 0.28, intertrial transition, F(2,46) = 12.78; P < 0.001; partial η2 = 0.36 and group, F(1,23) = 6.82; P = 0.001; partial η2 = 0.23, on reaction times. Most importantly, group and intertrial transition interacted significantly, F(2,46) = 8.47; P = 0.001; partial η2 = 0.27. Reaction times in the no-dimensional carry-over group did not differ across the within- and across-dimension conditions [within-dimension condition: mean = 1052.69 ms, SD = 349.41 ms; across dimension condition: mean = 1061.97 ms, SD = 382.21 ms; t(8) = 0.47; P = 0.649]. In contrast, the other patients showed slower reaction times when the dimension changed (mean = 806.29 ms; SD = 247.79 ms) compared with when the dimension was repeated [mean = 716 ms; SD = 208.97 ms; t(15) = 5.77; P < 0.001]. The results are presented in Fig. 2. Reaction times of the patient groups were also analysed separately in relation to the hemifield effect.

Figure 2.

Figure 2

Open in a new tab

Reaction time differences between the dimensional change and dimensional repetition condition in the dimensional effect group are shown in comparison with a lack of reaction time differences in the no-dimensional effect group. RTs = reaction times.

No-dimensional carry-over group

There was a significant effect of hemifield on reaction times, F(1,8) = 6.14; P = 0.038; partial η2 = 0.434. The different intertrial transitions (same dimension same feature, same dimension different feature and different dimensions) did not show a significant effect, F(2,16) = 0.18; P = 0.839; partial η2 = 0.022, and the factors did not interact, F(2,16) = 0.94; P = 0.414; partial η2 = 0.105. Reaction times can be seen in Fig. 3. The patients were considerably slower when targets were presented in the left hemifield (mean = 1102.56 ms; SD = 409.01 ms) compared with the right hemifield (mean = 1009.01 ms; SD = 372.98 ms). To make sure that the lack of dimensional effects was not due to overall reaction times, the group was split (median split) into faster (n = 4; left: mean = 744.04 ms; SD = 232.55 ms; right: mean = 719.54 ms; SD = 202.09 ms) and slower patients (n = 5; left: mean = 1389.37 ms; SD = 243.28 ms; right: mean = 1240.59 ms; SD = 308.32 ms). The effects of intertrial transition were not reliable [F(2,6) = 0.11; P = 0.900; partial η2 = 0.035, and F(2,8) = 0.14; P = 0.870; partial η2 = 0.034, for the faster and slower groups, respectively].

Figure 3.

Figure 3

Open in a new tab

Reaction times of the group without (left) and of the group with (right) dimensional effects are plotted separately for both hemifields as a function of the different intertrial transitions. RTs = reaction times; sdsf = same dimension same feature; sddf = same dimension different feature; dd = different dimensions.

Dimensional carry-over group

Reaction times of the second patient group (Fig. 3) were significantly affected by the different intertrial transitions, F(2,30) = 28.83; P < 0.001; partial η2 = 0.658, but only marginally by hemifield, F(1,15) = 3.59; P = 0.077; partial η2 = 0.193. These factors did not interact, F(2,30) = 0.18; P = 0.839; partial η2 = 0.012. In both hemifields, reaction times were considerably higher [left hemifield: 97.66 ms; F(2,14) = 10.93; P = 0.001; partial η2 = 0.609; right hemifield: 89.50 ms; F(2,14) = 13.79; P < 0.001; partial η2 = 0.663] when the target’s dimension changed across consecutive trials compared with when it was repeated (in the same- and different-feature conditions); dimensional carry-over effects were symmetrical, i.e. the costs of switching dimensions were equally large for colour and orientation targets for those participants who showed the dimensional carry-over effects.

Overall reaction times of the group not showing a dimensional carry-over effect were on average higher (left field: mean = 1102.56 ms; SD = 409.01 ms; right field: mean = 1009.01 ms; SD = 372.98 ms) than reaction times in the group showing a dimensional effect (left: mean = 777.29 ms; SD = 253.24 ms; right: mean = 714.98 ms; SD = 190.58 ms). To make sure that the dimensional carry-over effects were not due to the overall faster reaction times, the dimensional effect group was split (median split) into the faster (n = 8; left: mean = 617.26 ms; SD = 102.21 ms; right: mean = 584.73 ms; SD = 86.98 ms) and slower patients (n = 8; left: mean = 937.32 ms; SD = 256.47 ms; right: mean = 845.23 ms; SD = 173.53 ms). The effects of intertrial transition were reliable [F(2,14) = 13.61; P = 0.001; partial η2 = 0.660 and F(2,14) = 19.85; P < 0.001; partial η2 = 0.739, for the faster and slower groups, respectively].

The behavioural analysis suggested a clear difference between patients who did and who did not show dimensional effects in search. To look at potential connections between dimensional carry-over effects and working memory deficits, we looked for correlations between dimensional carry-over effects and several working memory measures.

Visuo-spatial attention and working memory battery

The magnitude of the dimensional carry-over effects in the patients (the difference between the mean reaction times of the dimensional change conditions minus the mean reaction times of the dimensional repetition conditions) was correlated with several measures of working memory: the score from the auditory attention task in the Birmingham Cognitive Screen (Humphreys et al., 2012), a measure of each patient’s Corsi block span and the digit span (forward and backward). None of the working memory measures correlated with the dimensional effects auditory attention task (Birmingham Cognitive Screen), r = −0.056; P = 0.792; Corsi block span: r = −0.103; P = 0.625; digit span forward: r = 0.098; P = 0.642; digit span backward: r = 0.002; P = 0.792). As a measure for sustained attention, we subtracted the scores from the digit span forward from the digit span (Robertson, 1990). The difference between backward and forward span—as a test for sustained attention—also did not correlate with the dimensional effects, r = −0.161; P = 0.44. Consistent with this, there were no significant differences on the working memory measures between the patients who did or who did not show reliable dimensional carry-over effects: for the auditory attention task (Birmingham Cognitive Screen), t(23) = 1.05; P = 0.305; for Corsi block span, t(23) = 0.96; P = 0.347; for digit span forward, t(23) = 1.84; P = 0.078; for digit span backward, t(23) = 1.28; P = 0.212; backward minus forward span, t(23) = 1.54; P = 0.137.

To assess the neuroanatomical bases of the dimensional carry-over effects, we conducted voxel-based morphometry analyses assessing correlations between the dimensional carry-over effects and the underlying lesions.

Neuroimaging findings

We used whole brain statistical analyses (voxel-based morphometry) based on a general linear model to evaluate common structure–function relationships across the whole brain for grey matter in an unbiased group of chronic neuropsychological patients (patients were not selected based on either lesion location or behaviour).

Using the voxel-based morphometry approach, we controlled for visuo-spatial deficits, working memory and sustained attention problems in the patients by adding covariates for left and right spatial deficits (either visual neglect or visual extinction), working memory and sustained attention. We found that grey matter damage within the right inferior parietal lobule (the angular and supramarginal gyri) extending into the intraparietal sulcus was associated with reported lack of dimensional effects in patients (Table 2 and Fig. 4).

Table 2.

Grey matter substrates of dimensional effects (controlled for visuospatial deficits, working memory and sustained attention problems)

Contrast Cluster level
 Voxel level Coordinates
 Brain structure (location)
P-valuesFWE Size Z-score x y z
Dimensional effects
Voxel-based morphometry: analysis P < 0.001 1528 4.84 62 −50 42 Right inferior parietal lobule (angular and supramarginal gyri) extending into intraparietal sulcus
3.88 60 −38 50
3.77 60 −28 40
Open in a new tab

Results are reported at significance level P < 0.001 family-wise error (FWE) cluster-level corrected for multiple comparisons.

Figure 4.

Figure 4

Open in a new tab

Voxel-wise statistical analysis of grey matter damage: dimensional effect versus no dimensional effects. The figure shows the areas associated with a lack of dimensional effects in patients (grey matter lesions within the right inferior parietal lobule extending into the intraparietal sulcus) controlled for left and right spatial deficits, working memory and sustained attention index. VBM = voxel-based morphometry.

Discussion

We examined the neuronal substrates for dimensional biases in visual search by testing for dimensional carry-over effects in a non-biased sample of brain-injured patients. Behavioural performance was assessed in one group of patients with substantial dimensional carry-over effects (an average 90 ms difference between trials where the target was defined along the same dimension relative to trials where the target dimension differed), a group with no dimensional effects and a healthy control group (dimensional costs on average ∼65 ms). For all the groups, there was no effect of changing across consecutive trials the feature value within each critical dimension (from one colour to another, or one orientation to another). This result fits with findings by Müller and colleagues (Müller et al., 1995; Found and Müller, 1996; Gramann et al., 2007) and confirms their proposal that attentional weighting is dimension-specific, rather than feature-specific, in nature. In the study by Gramann et al. (2007), this was also reflected in amplitude modulations of event-related potential components, which were unaffected by whether or not the target-defining feature changed within the repeated dimension. These data contrast with other findings where feature-specific effects have been found (Maljkovic and Nakayama, 1994; Wolfe et al., 2003). At present, it is unclear why such differences occur. For example, studies showing feature effects have typically used component search tasks, where participants have to make a further discrimination after selecting the target, and these studies have also used smaller display sizes than studies showing a lack of feature carry-over. It may be that, with larger display sizes participants respond to the feature disparity rather than the feature value, resulting in a lack of feature-specific carry-over. Whatever the case, the current displays did support strong dimensional carry-over effects.

Voxel-based morphometry analyses (Ashburner and Friston, 2000) were applied to investigate the neural substrates of the dimensional carry-over effects by examining across the whole brain the relationship between grey matter damage and dimensional carry-over effects. The data showed that, controlling for potentially co-varying factors (e.g. visual neglect or extinction, working memory and sustained attention), grey matter damage within the right inferior parietal lobule (the angular and supramarginal gyri), extending into the intraparietal sulcus, was associated with a lack of dimensional carry-over effects in the patient group. Our data suggest that these regions of parietal cortex are necessary to attention shifting in the context of visual dimensional change (see also Pollmann et al., 2006). The present results contrast with the effects reported by Kristjánsson et al. (2005). In the two cases of Kristjánsson et al. (2005), however, the lesions were more ventral than the critical site revealed in our study, and it may be that preservation of more dorsal regions of the angular and supramarginal gyrus, extending to the intraparietal sulcus, may have enabled these patients to still manifest dimensional carry-over effects. This would support the case for parcellation of the posterior parietal cortex (see also Gillebert et al., 2012) between ventral regions more critical for the detection of contralesional stimuli [e.g. both patients studied by Kristjánsson et al. (2005) had unilateral neglect] and more dorsal regions supporting memory-based carry-over effects and biases in non-spatial selection across trials.

One possible reason for the lack of dimensional effects in the patients with damage to the angular gyrus, supramarginal gyrus and intra-parietal sulcus is that these regions support visual working memory (Malhotra et al., 2005; Xu and Chun, 2006), and this may be critical to sustain effects across consecutive trials. Against this, we found neither overall differences in working memory between the patient groups who did or did not show dimensional carry-over effects nor correlations between the carry-over effects and the measures of working memory for individual patients. There was no evidence that differences in working memory were critical. We also note that in studies such as that of Kristjánsson et al. (2005), the dimensional carry-over effects could even be observed when the patients had no conscious awareness of prior targets appearing on their affected side. This suggests that the dimensional carry-over effects do not necessarily reflect explicit memory across trials as opposed to the existence of some more automatic process of spatio-temporal updating that is performed by the posterior parietal cortex. Olivers and Humphreys (2004) reported data from another form of spatio-temporal updating based on preview effects in visual search, where earlier presented distractors are ignored (rather than targets being positively weighted for search, as here). Patients with posterior parietal damage showed a poor ability to update search so that the old items had the same attentional impact as new items. We propose that the posterior parietal cortex may play a general role in spatio-temporal updating, and that damage to the neural structures supporting this process limit the impact of prior targets on subsequent visual search. The present results also differ from previous reports (Pollmann et al., 2007), which have stressed the role of the left fronto-polar cortex and that damage to this brain region disrupts dimensional effects. Though we found no evidence for this, fronto-polar lesions were not strongly represented in our sample. However, of the six patients with fronto-polar lesions, only three were in the no carry-over group, and the others did show carry-over effects. Hence, the evidence we have does not strongly support the necessary role of fronto-polar cortex in dimensional carry-over effects.

It may also be that the left fronto-polar cortex plays a different computational role in the dimensional effects than the parietal areas we found critical to performance. For example, the left fronto-polar cortex may signal a change when the target dimension shifts across trials while the posterior parietal cortex implements that shift by the updating of attentional weights. Gillebert et al. (2012) have argued that the middle section of the intraparietal sulcus is involved in setting attentional ‘weights’ to the locations of stimuli in order to bias visual selection. Our results indicate that this might be a more general function not confined to spatial selection but also reflecting selection through other stimulus features (e.g. colour, orientation). Whether the brain regions differ for the setting of weights for spatial and feature-based selection remains a question for future research.

Methodological considerations

The number of patients included in our study was relatively small and, being unselected, the patients had a range of lesions. We note, however, that the power of voxel-based morphometry is to use lesion diversity to enable different patients to act as controls for factors such as vascular territory, which can impact on the location of brain lesions (e.g. in stroke patients; Mort et al., 2003). However, in relatively small patient samples the approach may be insensitive for lesion-symptom mapping if the diversity of lesions masks some critical brain areas. This potentially means that such methods in general may detect regions necessary, but not sufficient, for specific cognitive functions, i.e. detecting some, but not all, components of the neuronal network involved in a given task. This is in contrast to functional neuroimaging studies using a correlatory approach, which may detect all regions activated during a task, but these might not be necessary for performance. In this respect, approaches such as voxel-based morphometry may be complementary to functional brain imaging in normal populations.

Funding

This work was supported by grants from the Biotechnology and Biology Research Council (BB/E006175/1), the Stroke Association (UK): Attentional and executive problems following stroke (PROG6), and by the National Institute of Health Research: Improving the diagnosis and treatment of cognitive problems after stroke (RP-DG-0610-10041).

Supplementary material

Supplementary material is available at Brain online.

Supplementary Data
supp_136_3_751__index.html (842B, html)

References

  1. Ashburner J, Friston KJ. Voxel-based morphometry—the methods. NeuroImage. 2000;11:805–21. doi: 10.1006/nimg.2000.0582. [DOI] [PubMed] [Google Scholar]
  2. Ashburner J, Friston KJ. Unified segmentation. NeuroImage. 2005;26:839–51. doi: 10.1016/j.neuroimage.2005.02.018. [DOI] [PubMed] [Google Scholar]
  3. Bickerton WL, Samson D, Williamson J, Humphreys GW. Separating forms of neglect using the Apples Test: validation and functional prediction in chronic and acute stroke. Neuropsychology. 2011;25:567–80. doi: 10.1037/a0023501. [DOI] [PubMed] [Google Scholar]
  4. Bundesen C. A theory of visual attention. Psychol Rev. 1990;97:523–47. doi: 10.1037/0033-295x.97.4.523. [DOI] [PubMed] [Google Scholar]
  5. Chechlacz M, Rotshtein P, Bickerton W-L, Hansen PC, Deb S, Humphreys GW. Separating neural correlates of allocentric and egocentric neglect: distinct cortical sites and common white matter disconnections. Cognit Neuropsychol. 2010;27:277–303. doi: 10.1080/02643294.2010.519699. [DOI] [PubMed] [Google Scholar]
  6. Chechlacz M, Rotshtein P, Hansen PC, Riddoch JM, Deb S, Humphreys GW. The neural underpinings of simultanagnosia: disconnecting the visuospatial attention network. J Cogn Neurosci. 2012;24:718–35. doi: 10.1162/jocn_a_00159. [DOI] [PubMed] [Google Scholar]
  7. Chechlacz M, Rotshtein P, Hansen PC, Deb S, Riddoch MJ, Humphreys GW. The central role of the temporo-parietal junction and the superior longitudinal fasciculus in supporting multi-item competition: Evidence from lesion-symptom mapping of extinction. Cortex. 2011 doi: 10.1016/j.cortex.2011.11.008. , doi:10.1016/j.cortex.2011.11.008. [Epub ahead of print] [DOI] [PubMed] [Google Scholar]
  8. Corsi PM. Human memory and the medial temporal region of the brain. Unpublished doctoral dissertation. Montreal, Canada: McGill University; 1972. [Google Scholar]
  9. Crinion J, Ashburner J, Leff A, Brett M, Price C, Friston K. Spatial normalization of lesioned brains: performance evaluation and impact on fMRI analyses. NeuroImage. 2007;37:866–75. doi: 10.1016/j.neuroimage.2007.04.065. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Duncan J, Humphreys GW. Visual search and stimulus similarity. Psychol Rev. 1989;96:433–58. doi: 10.1037/0033-295x.96.3.433. [DOI] [PubMed] [Google Scholar]
  11. Duvernoy HM, Cabanis EA, Vannson JL. The Human brain surface, three-dimensional sectional anatomy and MRI. Wien: Springer-Verlag; 1991. [Google Scholar]
  12. Evans AC, Collins DL, Mills SR, Brown RD, Kelly RL, Peters TM. 3D statistical neuroanatomical models from 305 MRI volumes. IEEE Nucl Sci Symp Med Imag Conf. 1993;3:1813–17. [Google Scholar]
  13. Found A, Müller HJ. Searching for unknown feature targets on more than one dimension: further evidence for a dimension weighting account. Percept Psychophys. 1996;58:88–101. doi: 10.3758/bf03205479. [DOI] [PubMed] [Google Scholar]
  14. Gillebert CR, Dyrholm M, Vangkilde S, Kyllingsbæk S, Peeters R, Vandenberghe R. Attentional priorities and access to short-term memory: parietal interactions. NeuroImage. 2012 doi: 10.1016/j.neuroimage.2012.05.038. [DOI] [PubMed] [Google Scholar]
  15. Gramann K, Töllner T, Krummenacher J, Eimer M, Müller HJ. Brain electrical correlates of dimensional weighting: an ERP study. Psychophysiology. 2007;44:277–92. doi: 10.1111/j.1469-8986.2007.00496.x. [DOI] [PubMed] [Google Scholar]
  16. Gramann K, Töllner T, Müller HJ. Dimension-based attention modulates early visual processing. Psychophysiology. 2010;47:968–78. doi: 10.1111/j.1469-8986.2010.00998.x. [DOI] [PubMed] [Google Scholar]
  17. Humphreys GW, Bickerton W-L, Samson D, Riddoch M. Birmingham Cognitive Screen (BCoS) Hove: Psychology Press; 2012. [DOI] [PubMed] [Google Scholar]
  18. Kanwisher N, Wojciulik E. Visual attention: insights from brain imaging. Nat Rev Neurosci. 2000;1:91–100. doi: 10.1038/35039043. [DOI] [PubMed] [Google Scholar]
  19. Kiebel S, Holmes A. The General Linear Model. In: Frackowiak RSJ, Friston KJ, Frith C, et al., editors. Human brain function. London: Academic Press; 2003. pp. 881–6. [Google Scholar]
  20. Kristjánsson Á, Vuilleumier P, Malhotra P, Husain M, Driver J. Priming of color and position during visual search in unilateral spatial neglect. J Cognit Neurosci. 2005;1:859–73. doi: 10.1162/0898929054021148. [DOI] [PubMed] [Google Scholar]
  21. Leff AP, Schofield TM, Crinion JT, Seghier ML, Grogan A, Green DW, et al. The left superior temporal gyrus is a shared substrate for auditory short-term memory and speech comprehension: evidence from 210 patients with stroke. Brain. 2009;132:3401–10. doi: 10.1093/brain/awp273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Malhotra P, Jäger HR, Parton A, Greenwood R, Playford ED, Brown MM, et al. Spatial working memory capacity in unilateral neglect. Brain. 2005;128:424–35. doi: 10.1093/brain/awh372. [DOI] [PubMed] [Google Scholar]
  23. Maljkovic V, Nakayama K. Priming of popout: I. Role of features. Mem Cognit. 1994;22:657–72. doi: 10.3758/bf03209251. [DOI] [PubMed] [Google Scholar]
  24. Mort DJ, Malhotra P, Mannan SK, Rorden C, Pambakian A, Kennard C, et al. The anatomy of visual neglect. Brain. 2003;126:1986–97. doi: 10.1093/brain/awg200. [DOI] [PubMed] [Google Scholar]
  25. Müller HJ, Heller D, Ziegler J. Visual search for singleton feature targets within and across feature dimensions. Percept Psychophys. 1995;57:1–17. doi: 10.3758/bf03211845. [DOI] [PubMed] [Google Scholar]
  26. Müller HJ, O’Grady RB. Dimension-based visual attention modulates dual-judgment accuracy in Duncan’s (1984) one- versus two-object report paradigm. J Exp Psychol Hum. 2000;26:1332–51. doi: 10.1037//0096-1523.26.4.1332. [DOI] [PubMed] [Google Scholar]
  27. Olivers C, Humphreys GW. Spatiotemporal segregation in visual search: evidence from parietal lesions. J Exp Psychol Hum. 2004;30:667–8. doi: 10.1037/0096-1523.30.4.667. [DOI] [PubMed] [Google Scholar]
  28. Pollmann S, Mahn K, Reimann B, Weidner R, Tittgemeyer M, Preul C, et al. Selective visual dimension weighting deficit after left lateral frontopolar lesions. J Cogn Neurosci. 2007;19:365–75. doi: 10.1162/jocn.2007.19.3.365. [DOI] [PubMed] [Google Scholar]
  29. Pollmann S, Weidner R, Müller HJ, von Cramon DY. A fronto-posterior network involved in visual dimension changes. J Cognit Neurosci. 2000;12:480–94. doi: 10.1162/089892900562156. [DOI] [PubMed] [Google Scholar]
  30. Pollmann S, Weidner R, Müller HJ, von Cramon DY. Neural correlates of visual dimension weighting. Vis Cogn. 2006;14:877–97. [Google Scholar]
  31. Price CJ, Seghier ML, Leff AP. Predicting language outcome and recovery after stroke: the PLORAS system. Nat Rev Neurol. 2010;6:202–21. doi: 10.1038/nrneurol.2010.15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Robertson IH. Digit span and visual neglect: a puzzling relationship. Neuropsychologia. 1990;28:217–22. doi: 10.1016/0028-3932(90)90103-u. [DOI] [PubMed] [Google Scholar]
  33. Seghier ML, Ramlackhansingh A, Crinion J, Leff AP, Price CJ. Lesion identification using unified segmentation-normalisation models and fuzzy clustering. NeuroImage. 2008;41:1253–66. doi: 10.1016/j.neuroimage.2008.03.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Stamatakis EA, Tyler LK. Identifying lesions on structural brain images—Validation of the method and application to neuropsychological patients. Brain Lang. 2005;94:167–77. doi: 10.1016/j.bandl.2004.12.010. [DOI] [PubMed] [Google Scholar]
  35. Töllner T, Gramann K, Müller HJ, Kiss M, Eimer M. Electrophysiological markers of visual dimension changes and reponse changes. J Exp Psychol Hum. 2008;34:531–42. doi: 10.1037/0096-1523.34.3.531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Treisman AM. Features and objects. The fourteenth Bartlett memorial lecture. Q J Exp Psychol. 1988;40A:201–37. doi: 10.1080/02724988843000104. [DOI] [PubMed] [Google Scholar]
  37. Weidner R, Pollmann S, Müller HJ, von Cramon DY. Top–down controlled visual dimension weighting: an event-related fMRI study. Cereb Cortex. 2002;12:318–28. doi: 10.1093/cercor/12.3.318. [DOI] [PubMed] [Google Scholar]
  38. Wolfe JM, Butcher SJ, Lee C, Hyle M. Changing your mind: on the contributions of top-down and bottom-up guidance in visual search for feature singletons. J Exp Psychol Hum. 2003;29:483–502. doi: 10.1037/0096-1523.29.2.483. [DOI] [PubMed] [Google Scholar]
  39. Woolsey TA, Hanaway J, Gado MH. The brain atlas a visual guide to the human central nervous system. Hoboken, NJ: John Wiley & Sons; 2008. [Google Scholar]
  40. Worsley KJ. Developments in random field theory. In: Frackowiak RSJ, Friston KJ, Frith C, et al., editors. Human brain function. San Diego, CA: Academic Press; 2003. pp. 881–6. [Google Scholar]
  41. Xu Y, Chun MM. Dissociable neural mechanisms supporting visual short-term memory for objects. Nature. 2006;440:91–5. doi: 10.1038/nature04262. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
supp_136_3_751__index.html (842B, html)
supp_awt003_brain-2012-01257-File008.docx (108.8KB, docx)

Articles from Brain are provided here courtesy of Oxford University Press

RESOURCES