Abstract

Working memory (WM) is known to be impaired in patients with stroke experiencing unilateral spatial neglect (USN). Here, we examined in a systematic manner three WM components: memory of object identity, memory of object location, and binding between object identity and location. Moreover, we used two different retention intervals to isolate maintenance from other mnemonic and perceptual processes. Fourteen USN first-event stroke patients with right-hemisphere damage were tested in two different WM experiments using long and short retention intervals and an analog response scale. Patients exhibited more identification errors for items displayed on the contralesional side. Localization errors were also more prominent in the contralesional side, especially after a long retention interval. These localization errors were often a result of swap errors, that is, erroneous localizations of correctly identified contralesional objects in correctly memorized locations of ipsilesional objects. We conclude that a key WM deficit in USN is a lateralized impairment in binding between the identity of an object and its spatial tag.

INTRODUCTION

Unilateral spatial neglect (USN) is a clinical syndrome frequently encountered after right-hemisphere stroke (Karnath & Rorden, 2012; Mort et al., 2003; Halligan, Marshall, & Wade, 1989). The hallmark of this symptom complex is failure, or diminished ability, to orient, attend, and respond to salient stimuli in the contralesional side of space (Corbetta & Shulman, 2011; Husain & Rorden, 2003). The neglect syndrome is multifactorial and multifaceted, and its research gave rise to various theoretical accounts of attention and perception, leading to the development of different therapeutic modalities (Azouvi, Jacquin-Courtois, & Luauté, 2017; Luauté, Halligan, Rode, Rossetti, & Boisson, 2006; Pierce & Buxbaum, 2002). Better understanding of the mechanisms underlying the expression of USN is important for deciding on the optimal treatment strategy in each specific case and also contributes to better understanding of the neural processes involved in perception and attention.

One construct that was frequently shown to be impaired in USN is spatial working memory (WM; Malhotra et al., 2005; Pisella, Berberovic, & Mattingley, 2004; Wojciulik, Husain, Clarke, & Driver, 2001), and influential theoretical accounts of USN point to impaired spatial WM as an important component of the mechanism behind neglect (Malhotra et al., 2005; Della Sala, Logie, Beschin, & Denis, 2004; Husain & Rorden, 2003; Beschin, Cocchini, Della Sala, & Logie, 1997). However, spatial WM is only one facet of a more comprehensive mechanism that could be separated into (1) visual WM, the memory of visual information; (2) spatial WM, the memory of spatial information; and (3) a third component enabling the binding of visual and spatial information (Manohar, Pertzov, & Husain, 2017; Baddeley, Allen, & Hitch, 2011). Evidence for this segregation comes from cognitive psychology, showing interference effects that are specific for location memory, for identity, or for the binding between them (Darling, Della Sala, & Logie, 2007; Treisman & Zhang, 2006; Baddeley, 2000). Neuropsychology studies report on neurological conditions that lead to impairments in one of these components, but not in the others (Zokaei et al., 2019; Dundon et al., 2018; Darling, Della Sala, Logie, & Cantagallo, 2006). A recent pharmacological study showed a specific effect of glucose intake on object location binding and object location memory but not on memory for identity (Stollery & Christian, 2016). Lesion studies have pointed to the role of the posterior parietal cortex—the region most frequently associated with USN (Vuilleumier, 2013; Mort et al., 2003)—in the process of binding identity and location information of objects. Friedman-Hill, Robertson, and Treisman (1995) reported a patient with bilateral parieto-occipital damage who showed an increased rate of binding errors between the color or size of a letter and its spatial location. In a more recent study, impaired ability to associate identity and location data of objects was found most frequently in patients with damage to the left posterior parietal cortex (van Asselen et al., 2009).

Note that the existence of distinct processes for remembering isolated features and for binding between the features is not under scientific consensus. For example, a host of studies failed to find a selective impairment of binding because of concurrent demands on attention (e.g., Allen, Baddeley, & Hitch, 2006), nor in special populations (e.g., Rhodes, Parra, Cowan, & Logie, 2017), leading to question previous reports on a distinction between feature and binding memory. Memory for individual features may be represented, at the neural level, together with other features that belong to the same object, as well as its location (e.g., conjunctive population code model; Schneegans & Bays, 2017). So, the question about the amount of overlap between memory of independent features, and memory of which features belong to each object, is far from being settled (Schneegans & Bays, 2019).

The issue of binding of identity and location information was examined in USN research from different perspectives, but unfortunately, these studies did not address working memory. In the original description of the famous “Piazza Effect” (Bisiach & Luzzatti, 1978), patients failed to retrieve from long-term memory identity information located in the contralesional side of the representational space (alternating between the northern and southern parts of the piazza depending on patients' changing imaginary point of view). This was interpreted as evidence that conscious access to identity information requires proper binding to location information and that, in USN, such binding is damaged in the contralesional space. A recent USN study examined how the memory of objects is influenced by their spatial position at the learning phase (Moreh, Malkinson, Zohary, & Soroker, 2014). Patients were instructed to memorize the identity of visual objects, each presented in one of four quadrants of the screen. Objects displayed on the contralesional side were remembered worse than objects on the ipsilesional side, both in immediate free-recall testing and in a recognition test after a delay of 8 min. In the latter test, patients were asked also to point to the quadrant of the screen where the recognized object was originally presented. Mislocalization rate was higher for objects on the neglected side, with an increased tendency to report that objects originally presented on the contralesional side were located on the ipsilesional side.

In the realm of WM, the binding of stimulus identity and location information, and the temporary maintenance of this bound information, are crucial for constructing a clear and coherent percept of a visual scene from a series of ocular fixations on individual objects constructing the scene. Failure to properly encode location information in contralesional space, to bind identity and location information in WM, or to maintain it for short intervals of time could all contribute and explain various USN phenomena. Yet, the operation of the different components of WM in USN was not examined in a systematic manner until now. The introduction of “delayed estimation” tasks in WM research enabled an accurate quantitative characterization of types of WM failure. In such tasks, participants are asked to report the exact value of a recently displayed object in a continuous analog response space (e.g., Cohen-Dallal, Fradkin, & Pertzov, 2018; Bays, Gorgoraptis, Wee, Marshall, & Husain, 2011; Bays & Husain, 2008; Zhang & Luck, 2008; Prinzmetal, Amiri, Allen, & Edwards, 1998), thus enabling analysis of the distribution of errors around the target value as well as separating errors that may arise from different origins (Bays, Catalao, & Husain, 2009). This paradigm has led to conceptual shifts in visual-WM theorizing (Ma, Husain, & Bays, 2014). It was found to have increased sensitivity in the assessment of memory deficits in neurological conditions (Zokaei, Burnett Heyes, Gorgoraptis, Budhdeo, & Husain, 2015) and improved accuracy in assessment of fine detailed memory compared to traditional assessments of location memory (where memory of the spatial location of a given object is reported by selecting one of a number of options). In a study by van Asselen, Kessels, Kappelle, and Postma (2008), location memory tested in the traditional manner was more impaired after left-hemisphere damage compared to right-hemisphere damage, whereas testing using the analog reporting scale revealed the opposite (van Asselen et al., 2008).

A series of recent studies (Liang et al., 2016; Pertzov, Heider, Liang, & Husain, 2015; Pertzov, Bays, Joseph, & Husain, 2013; Pertzov, Dong, Peich, & Husain, 2012) have used a WM “delayed estimation” task named the “what was where” task. It examines memory of object location separately from memory of object identity, after different retention intervals and using a continuous reporting measure (see Figure 2 for a presentation of the method). Location memory is assessed by dragging the recognized object to its remembered location using a touch screen. The task provides a quantitative measure of three different types of WM errors: (1) object identification errors, (2) object localization errors measured by the absolute distance between the original and the reported location, and (3) swap errors—when the correct target object was erroneously placed close to the location of a different (nontarget) object. This third kind of error has been linked to a deficit in binding together the correct identity of an object to its location (Spotorno, Evans, & Jackson, 2018; Liang et al., 2016; Pertzov et al., 2012, 2015). Encoding and retrieval are separated by relatively short and long retention intervals (conditions interleaved) to assess the impact of maintenance time in WM on the above three measures. Using this task in individuals with medial temporal lobe damage because of limbic encephalitis (Pertzov, Miller, et al., 2013), it was possible to show that these patients have a specific impairment in binding object identity to location but no difficulty in remembering the identities and locations on their own. In another study, “asymptomatic” individuals with pathological mutations in genes related to familial Alzheimer's disease showed significantly more binding errors compared to healthy controls (Liang et al., 2016). A third study using this task in healthy individuals revealed an age-related decline in object identification and localization, but once controlling for decline in identity memory, there was no deficit in identity–location binding (Pertzov et al., 2015).

Here, we used the “what was where” task to examine USN-related lateralized deficits in the different components of WM by comparing memory reports for objects presented in the ipsilesional and contralesional sides of the memory array. In a second “delayed estimation” task, object identity memory was tested using an analog scale, asking the patients to memorize the orientation of two bars, one on the left and one on the right of fixation, and reproducing the orientation of one bar using a dial (as in Pertzov, Bays, et al., 2013; Bays et al., 2011; Gorgoraptis, Catalao, Bays, & Husain, 2011). Error quantification was based on the difference between the reported angle and the correct angle. Swap errors were defined here as errors in which the reported orientation was close to the orientation of the other bar (Ma et al., 2014; Pertzov, Bays, et al., 2013; Pertzov et al., 2012). This second paradigm may uncover effects that were missed in the first experiment and rule out an alternative explanation for selective localization errors (where stimulus identity is recalled correctly) as originating from the use of different reporting methods for identity and location data in the “what was where” paradigm.

Note that patients with USN exhibit attention and WM deficits across the entire visual space (e.g., Husain & Rorden, 2003). However, in the current study, we focus on lateralized deficits and assess them by comparing patients' performance in the contralesional versus ipsilesional sides of space. Regardless of the location of the target stimulus, memory performance of brain-damaged USN patients is expected to be worse than that of healthy individuals (patients' results can be compared to those of age-matched healthy controls tested in another study that used an almost identical experimental design; Pertzov et al., 2015; see supplementary material1).

In summary, the goal of the current study was to unravel the nature of lateralized WM deficits in USN using two different delayed estimation tasks—“what was where” and a line-orientation task. We assessed the impact of stimulus side on location memory, identity memory, and the binding between the two, after different retention intervals.

METHODS

Participants

Sixteen patients with first-ever stroke affecting the right cerebral hemisphere who showed clinical symptoms of USN and were found to be able to comply with the task demands were recruited to the study. All the patients were admitted to the Department of Neurological Rehabilitation at the Loewenstein Hospital, Raanana, Israel, shortly after the onset of stroke. To be included, patients had to comply with the following inclusion criteria: first-ever ischemic or hemorrhagic right-hemisphere stroke, manifestation of clinical symptoms of neglect, focal lesion with no evidence of earlier vascular events or another pathology in CT performed in the acute stage, negative neurological and psychiatric past history, stable clinical and metabolic state at the time of testing, and willingness to participate in the study. During the recruitment period, there were another six USN patients who met the inclusion criteria but had to be excluded because the severity of their impairment precluded compliance with task demands. Two patients had to be excluded from the analysis because their scores in the formal neglect tests were above the cutoff for normality. Stroke was ischemic in 10 patients and hemorrhagic in four. Testing commenced between 38 and 99 days after stroke onset (mean = 73.1 days, SD = 19.5 days). There were nine men and five women at an age range between 49 and 75 years (61.93 ± 6.8) and an educational level of 8–18 years of formal education (12.7 ± 2.9). Twelve patients participated in the first experiment, and 11 participated in the second experiment. The study was approved by the ethics review board of the Loewenstein Hospital, and all the patients provided a written informed consent to participate. Clinical and demographic details of each patient are shown in Table 1, and lesion data are shown in Figure 1.

Table 1. 
Patients' Demographic and Clinical Data
PatientAge/GenderEducationTAO Exp1TAO Exp2Lesion TypeBITMWCTLB MSD, Mean ± SDSNT RT (msec)
LRLR
68/M 12 99 113 127 26 29 4.4 ± 4.4 584 480 
59/F 15 – 90 122 12 30 16.6 ± 5.6 1071 644 
75/M 10 38 38 142 28 29 13.6 ± 6.2 1040 860 
49/M 16 – 25 67 15 19.8 ± 10.6 853 485 
58/F 12 99 99 130 11 22 19.5 ± 2.7 789 642 
62/M 12 47 47 141 28 30 4.8 ± 4.0 658 482 
63/F 10 71 57 122 23 29 2.2 ± 5.1 866 519 
49/M 12 89 83 139 19 28 5.8 ± 5.9 1007 831 
65/M 15 73 69 97 14 27 5.9 ± 6.8 623 457 
10 67/M 18 75 74 109 19 28 5.4 ± 3.4 1120 875 
11 58/F 68 74 101 10 19 14.1 ± 10.4 1305 1352 
12 67/F 17 96 – 81 21 21.1 ± 5.5 1645 789 
13 62/M 158 – 119 21 13.1 ± 5.5 658 471 
14 65/M 12 138 – 137 15 26 1.4 ± 4.5 638 391 
PatientAge/GenderEducationTAO Exp1TAO Exp2Lesion TypeBITMWCTLB MSD, Mean ± SDSNT RT (msec)
LRLR
68/M 12 99 113 127 26 29 4.4 ± 4.4 584 480 
59/F 15 – 90 122 12 30 16.6 ± 5.6 1071 644 
75/M 10 38 38 142 28 29 13.6 ± 6.2 1040 860 
49/M 16 – 25 67 15 19.8 ± 10.6 853 485 
58/F 12 99 99 130 11 22 19.5 ± 2.7 789 642 
62/M 12 47 47 141 28 30 4.8 ± 4.0 658 482 
63/F 10 71 57 122 23 29 2.2 ± 5.1 866 519 
49/M 12 89 83 139 19 28 5.8 ± 5.9 1007 831 
65/M 15 73 69 97 14 27 5.9 ± 6.8 623 457 
10 67/M 18 75 74 109 19 28 5.4 ± 3.4 1120 875 
11 58/F 68 74 101 10 19 14.1 ± 10.4 1305 1352 
12 67/F 17 96 – 81 21 21.1 ± 5.5 1645 789 
13 62/M 158 – 119 21 13.1 ± 5.5 658 471 
14 65/M 12 138 – 137 15 26 1.4 ± 4.5 638 391 

F = female; M = male; Education = years of formal schooling; TAO = time after onset (in days); Lesion Type: I = ischemic infarction, H = hemorrhagic infarction; BIT = Behavioral Inattention Test; MWCT = Mesulam–Weintraub Cancellation Test; LB MSD = line bisection mean signed displacement (mm); SNT = Starry Night Test; L/R = left/right; [–] = not administered. See Clinical Assessment under the Methods section for further explanations.

Figure 1. 

Normalized lesions (white) of USN patients represented on 11 standard axial slices following Damasio and Damasio (1989). The images are flipped so the right hemisphere is on the left side.

Figure 1. 

Normalized lesions (white) of USN patients represented on 11 standard axial slices following Damasio and Damasio (1989). The images are flipped so the right hemisphere is on the left side.

Clinical Assessment

All the patients showed signs of left USN in at least two of the following paper-and-pencil tests and/or prolonged RT to left-sided stimuli in a computerized visual search task (the Starry Night test [SNT]; Deouell, Sacher, & Soroker, 2005). Individual patients' scores are depicted in Table 1.

Behavioral Inattention Test

The Behavioral Inattention Test is a standardized test battery for assessing neglect in the visual modality. It includes three distinct target cancelation tasks (lines, letters, stars): figure and shape copying, line bisection, and representational drawing. Maximum total score is 146, and the cutoff for normality is 130 points (Wilson, Cockburn, & Halligan, 1987).

Line Bisection Task

Participants marked the middle of 1-mm-wide horizontal lines of three different lengths (36, 90, and 180 mm) printed on A4 paper, each presented separately 10 times in random order. Mean signed displacement (+ rightward; − leftward) in millimeters of the subjective midpoint from the true midpoint of the 180-mm lines was used as a measure of neglect severity in this task.

Mesulam and Weintraub Cancellation Task

Participants are required to circle target stimuli randomly dispersed between distracter stimuli (15 of 90 stimuli in each quadrant of A4 paper are targets; Weintraub & Mesulam, 1987). Two or more omissions are considered pathological (Lowery, Ragland, Gur, Gur, & Moberg, 2004).

SNT

SNT is a computerized visual search task. Severity of USN is reflected by the difference in RT to right- and left-sided stimuli (Deouell et al., 2005).

Lesion Analysis

Patients' brain damage was assessed using clinically motivated follow-up CT scans performed during the rehabilitation period, on average 41 days after stroke onset (SD = 18.2). Lesion analyses were performed using the “Analysis of Brain Lesions” module implemented in MEDx software (Medical Numerics). The Analysis of Brain Lesions characterizes brain lesions in MRI/CT scans of the adult human brain by spatially normalizing the damaged brain into Talairach space. It reports anatomical structures in the normalized brain by using an interface to the Talairach daemon, the automated anatomical labeling atlas, and the white matter atlas (Solomon, Raymont, Braun, Butman, & Grafman, 2007; Tzourio-Mazoyer et al., 2002; Lancaster et al., 2000). The method we used for lesion analysis was described in detail in Haramati, Soroker, Dudai, and Levy (2008). Lesion boundaries were manually outlined on the digitized CTs using the MEDx software. Registration accuracy of the scans to the Montreal Neurological Institute template (as reported by the software) was greater than 92% for all patients.

Parenchymal damage was confined to the territory of the right middle cerebral artery in most patients (the graphic presentation of normalized lesion data is shown separately for each patient in Figure 1, using 11 representative axial CT slices, in accordance with the method of Damasio & Damasio, 1989). As can be seen in Figure 1, lesion size and location differ markedly in different patients. Total lesion volume ranged from 10.72 to 203.85 cm3 (M = 95.44 cm3, SD = 56.39 cm3).

Experiment 1: Methods, Materials, and Procedure

The task was designed to assess separately the different components of WM, as in earlier research (Liang et al., 2016; Pertzov et al., 2012, 2015; Pertzov, Bays, et al., 2013). To reduce possible effects of verbal coding, the objects to be memorized were fractals (following Pertzov et al., 2012). Memory of object identity was tested by recognition (selection of the memorized object from two options), and memory of object location was tested by dragging the object on the touch screen to its remembered location.

The task was designed in MATLAB (MathWorks, Inc.) using the Psychophysics Toolbox extensions (Kleiner et al., 2007; Brainard, 1997) and Cogent toolbox (Wellcome Department of Imaging Neuroscience). It was presented on a 28.5 × 51.0 cm touch-screen computer (Dell Inspiron One 2320) with a resolution of 1080 × 1920 at a viewing distance of approximately 40 cm (39.2° × 65.0°).

The procedure (similar to Experiment 3 in Pertzov et al., 2012) is depicted in Figure 2A. In each trial, participants viewed one or three fractal objects (drawn from Sprott's Fractal Gallery; sprott.physics.wisc.edu/fractals.html), for 1000 or 3000 msec, respectively, located randomly on the screen and were asked to remember both the identity of the objects and their locations. After a delay of either 100 or 4000 msec, two fractal objects (one that appeared in the memory array [target] and one that did not [foil]) were presented above and below the central fixation point of the screen, and the patient was asked to touch the one that appeared before and to drag it on the touch screen to its remembered location. This enabled the assessment of memory for object identity separately from memory for object location. Note that, because the task consists of two stages (identification and localization), even in the short retention interval (100 msec), localization stage began at least 1 sec after the memory array disappeared. Thus, all retention interval conditions were well within the domain of WM (rather than “iconic memory”), enabling the assessment of data maintenance in WM. Localization performance was analyzed only in trials in which an object was correctly identified. Each stimulus had a maximum width and height of 120 pixels (visual angle = 4.6°), and its location was determined by a MATLAB script for random presentation with the following restrictions: Objects were never located within 9.5° of each other, and they were positioned with a minimum of 4.1° from the edges of the screen and 6.8° from the center of screen. A test block consisted of 50 trials—10 with one fractal and 40 with three fractals. Fractals were repeated three to four times in different trials within a block. The blank maintenance interval was 100 msec in half of the trials and 4000 msec in the other half, in random order. Location memory was quantified by measuring the distance (in visual angles) between the center of the target object after it had been dragged to its remembered location, from the actual center of the object in the initial memory array. The main focus of the current study was on lateralized effects of the USN condition on WM performance. For that aim, we compared trials in which the target object was the leftmost in the array, to trials where it was the rightmost in the array. In all these comparisons, trials with one object were removed from the analysis. Progress was self-paced, and termination of each trial was signaled by pressing the space bar.

Figure 2. 

Timeline of the two experiments. (A) Experiment 1: One or three fractals were simultaneously presented at random locations on a black background for 1 or 3 sec (1 sec per object displayed). After a delay of 100 or 4000 msec, two fractals appeared, one old and the other new. Participants had to report which object appeared before and to drag it to the remembered location. (B) Experiment 2: A memory array of two oriented bars appeared until the participant reported she or he saw the two bars. This was followed by a blank retention interval of 1000 or 6000 msec. When a probe appeared, the participant was required to rotate it to match the orientation of the bar of the same color that appeared before in the memory array. Afterward, a white bar provided feedback by indicating the true orientation of the target stimulus.

Figure 2. 

Timeline of the two experiments. (A) Experiment 1: One or three fractals were simultaneously presented at random locations on a black background for 1 or 3 sec (1 sec per object displayed). After a delay of 100 or 4000 msec, two fractals appeared, one old and the other new. Participants had to report which object appeared before and to drag it to the remembered location. (B) Experiment 2: A memory array of two oriented bars appeared until the participant reported she or he saw the two bars. This was followed by a blank retention interval of 1000 or 6000 msec. When a probe appeared, the participant was required to rotate it to match the orientation of the bar of the same color that appeared before in the memory array. Afterward, a white bar provided feedback by indicating the true orientation of the target stimulus.

Experiment 2: Methods, Materials, and Procedure

The aim here was to corroborate the findings of Experiment 1 using an analog scale, which was found to be more sensitive (Zokaei et al., 2015), to test memory for object identity. Participants were presented with two oriented bars in two colors (one red and one blue, randomly determined) on the two sides of the screen. After a variable retention interval, they had to rotate a central bar, in random orientation and color (red or blue), to the orientation of the bar with the same color from memory. Swap errors were trials where the orientation reported was closer to the nonprobed object rather than the probed object. Programming tools and hardware were as in Experiment 1.

Before starting the experiment, participants were trained four times in rotating a bar to align its orientation with a viewed target bar, using a response dial (PowerMate USB Multimedia controller; Griffin Technology). Those who failed to properly operate the dial at this training stage were not included in the experiment. The experimental procedure is depicted in Figure 2B. Each trial began with the appearance of a central black fixation cross displayed over a gray background until the experimenter validated verbally that the participant was ready to start the trial and clicked the space bar key. Then, a memory array was presented, consisting of two oriented bars (0.4° × 2.0°) in two different colors (blue, red), presented at an equal distance of 5.0° to the left and right of the central fixation point. The two different colors appeared randomly on either side. The orientation of each bar was independently and randomly chosen from a circular parameter space of 0°–180° (i.e., the full range of possible undirected bar orientations) at a 1° resolution. The termination of the memory array was self-paced, and participants were instructed to press the dial after they had seen both bars. Next, a blank retention interval was introduced for 1000 or 6000 msec (retention intervals were randomly assigned to each trial), followed by a single bar (probe) presented in the center of the screen in a random orientation in one of the two colors of the bars in the memory array. Participants were required to rotate the probe using the dial to match it to the remembered orientation of the bar with the probe color (i.e., the target bar). Once the participants were satisfied with the probe orientation, they pressed the dial, and then a white bar with the correct orientation appeared superimposed on the probe for 600 msec, as feedback. Participants were instructed to be as accurate as possible regardless of the time it took them to adjust the probe orientation. The experiment included 60 trials; however, patients were allowed to stop the experiment at any point if they felt tired. The overall accuracy of each participant was displayed as a feedback at the end of the experiment (average angular error was linearly transformed to a 0–100 scale; 0 was 90° error, and 100 was a perfect alignment).

RESULTS

Experiment 1

Of the 12 USN patients who participated in this experiment, four completed it in two sessions, one completed it in three sessions, and the remainder completed the experiment in one session. In trials with three objects, only those in which the target object was the leftmost object in the array (mean = 36.45%, SD = 4.94%) and those in which it was the rightmost object (mean = 31.04%, SD = 5.04%) were analyzed.

Memory of object identity (Figure 4A), signaled by the rate of correct recognition performance (i.e., the proportion of times that participants touched the correct object in the two-alternative forced choice [2AFC] phase of the experiment) was analyzed by a repeated-measures ANOVA with Presentation Side (left, right) and Retention Interval (100 msec, 4000 msec) as within-participant factors. The main effect of Presentation Side was significant, F(1, 11) = 4.96, p = .047, ηp2 = .31, pointing to disadvantage in contralesional space. The main effect of Retention Interval was not significant, F(1, 11) < 1, p = .68, ηp2 = .02, nor was the interaction between the two factors, F(1, 11) < 1, p = .85, ηp2 = .02.

Memory of object location was measured by the distance between the reported location and the true location of correctly identified objects. Figure 3 describes the overall distribution of localization reports around the target (Figure 3A), when it was the leftmost object in the array (Figure 3B) and the rightmost object in the array (Figure 3C). As can be seen, when the target was in the rightmost position, the distribution of errors was more densely centered around the target than when the target appeared in the leftmost position (contralesional side). A repeated-measures ANOVA of localization performance with Presentation Side (left, right) and Retention Interval (100 msec, 4000 msec) as within-participant factors (Figure 4B) showed a significant main effect of Retention Interval, F(1, 11) = 5.11, p = .045, ηp2 = .31, with greater errors after longer retention time. The main effect of Side was also significant, F(1, 11) = 11.16, p < .01, ηp2 = .50, pointing to disadvantage in contralesional space. There was no significant interaction between the two factors, F(1, 11) < 1, p = .56, ηp2 = .03.

Figure 3. 

The distribution of localization responses around the location of the target object (A), when it was the leftmost (B) and rightmost (C) object in the array. The distribution of localization responses around the location of the nontarget objects (C) when the target object was the leftmost (E) and rightmost (F) object in the array. The green circle marks 4.7° around the center of the object, used as the threshold for determining the occurrence of a swap error (mislocalization of the target object in the location of a nontarget object). The units of x and y axes are degrees of visual angle (DVA).

Figure 3. 

The distribution of localization responses around the location of the target object (A), when it was the leftmost (B) and rightmost (C) object in the array. The distribution of localization responses around the location of the nontarget objects (C) when the target object was the leftmost (E) and rightmost (F) object in the array. The green circle marks 4.7° around the center of the object, used as the threshold for determining the occurrence of a swap error (mislocalization of the target object in the location of a nontarget object). The units of x and y axes are degrees of visual angle (DVA).

Figure 4. 

Results of Experiment 1 in the three object conditions. (A) Identification performance: proportion of correct responses when the target object was the leftmost (contralesional) and rightmost (ipsilesional) object in the array. (B) Localization performance measured by the distance between the reported location and the true location of target object. (C) Percentage of swap errors in which the target object was erroneously localized close to an original location of a nonprobed object. (D) The nearest item control was calculated as the minimal distance between a reported location and the closest location in which an object previously appeared. Error bars represent SEMs.

Figure 4. 

Results of Experiment 1 in the three object conditions. (A) Identification performance: proportion of correct responses when the target object was the leftmost (contralesional) and rightmost (ipsilesional) object in the array. (B) Localization performance measured by the distance between the reported location and the true location of target object. (C) Percentage of swap errors in which the target object was erroneously localized close to an original location of a nonprobed object. (D) The nearest item control was calculated as the minimal distance between a reported location and the closest location in which an object previously appeared. Error bars represent SEMs.

Localization reports tended to cluster also around nontarget objects (Figure 3D), more so when the target was on the left (Figure 3E) than on the right (Figure 3F). As in previous studies (Pertzov et al., 2012), swap errors (Figure 4C) were defined as errors in which the reported location of a correctly identified object was closer to the original location of a different object in the memory array. We used a predefined threshold of 4.7° (less than half of the minimal distance between objects, i.e., 9.48°) to ensure that the reported location could never be attributed to more than one object. The proportion of swap errors was quantified as the number of trials in which a swap error occurred, out of the total number of trials with correct identification. A repeated-measures ANOVA with Presentation Side (left, right) and Retention Interval (100 msec, 4000 msec) revealed significant main effects of Retention Interval, F(1, 11) = 7.99, p = .02, ηp2 = .50, and Presentation Side, F(1, 11) = 10.88, p = .01, ηp2 = .49, pointing to a disadvantage in recall information from the contralesional space. The interaction between the two factors was not significant, F(1, 11) = 2.09, p = .18, ηp2 = .15. Separate contrasts between the retention interval conditions in each side showed that the effect of retention interval was significant only in the contralesional side, t(11) = 2.877, p = .02, d = 0.83, but not in the ipsilesional side, t(11) = 0.15, p = .88, d = 0.04, pointing to rapid degradation of the connection between identity and location data of contralesional objects.

Close examination of the localization errors reveals a pattern that is incompatible with a global rightward localization bias as an explanation of the above results. In fact, the distribution of errors (Figure 3B and C) shows a clear bias toward the center and only a weaker general bias toward the right (Figure 3B). Thus, when the target was on the leftmost position, the errors were biased toward the right, and when the target was on the rightmost position, the localization errors were biased mainly toward the left.

We controlled the contribution of swap errors to localization performance by removing swap-error trials from the calculation of the average localization error. When removing swap errors, the repeated-measures ANOVA showed an insignificant effect of Retention Interval, F(1, 11) = 1.95, p = .18, ηp2 = .15, no effect of Display Side, F(1, 11) = 2.39, p = .15, ηp2 = .18, and no interaction between the two, F(1, 11) < 1, p = .97, ηp2 = .00.

As in previous studies (Pertzov et al., 2012), we also assessed the contribution of swap errors to localization performance by examining the distance between the reported location and the location of the closest object (Figure 4D). In Figure 5, we show the distribution of erroneous localizations around the true location of the nearest object in the memory array. It can be seen that most trials fall inside the threshold of 4.7° used to determine swap errors. Note that there are trials where the reported location was close neither to the target nor to the nontarget object, falling outside the threshold radius of 4.7° used to determine swap errors (Figure 5). In such trials, we cannot determine whether location memory of the objects was too weak or whether localization performance actually reflects a random guess. Enlarging the threshold for determining swap errors may lead to inclusion of some of these trials. However, the threshold was predefined before data collection as half of the minimal distance between fractals, to preclude attribution of swap errors to two objects. A repeated-measures ANOVA on the measure of error to the nearest object revealed no significant effect of Retention Interval, F(1, 11) = 1.79, p = .21, ηp2 = .14, no effect of Display Side, F(1, 11) < 1, p = .43, ηp2 = .06, and no interaction between the two, F(1, 11) < 1, p = .45, ηp2 = .05. The diminished effect of side when swap errors are taken out of the localization errors suggests that the main lateralized WM deficit in USN is in retaining the correct spatial tag of an object placed in contralesional space, rather than in location memory per se.

Figure 5. 

The distribution of localization errors to the nearest object: the distance between the reported location of the object and the nearest location of any object in the memory array (target or nontarget). The circle marks the 4.7° radius around the center of the item.

Figure 5. 

The distribution of localization errors to the nearest object: the distance between the reported location of the object and the nearest location of any object in the memory array (target or nontarget). The circle marks the 4.7° radius around the center of the item.

Only trials with correct identification were included in the localization and swap error analyses. In cases where correct identification occurred by chance, if location memory was preserved, participants could localize objects around one of the remembered locations. Thus, trials in which the identity of the objects was lost but location information was intact could contribute to the production of swap errors. The number of trials in which the correct object was selected by chance is expected to be similar to the number of trials in which the object was identified incorrectly. Of these trials, the upper limit of the number of trials where swap errors could be explained by identification by chance coupled with preserved location memory is 66% (localized near one of the other two nontarget objects—swap error). Thus, as performed in previous studies (e.g., Pertzov et al., 2012), the estimate of swap errors attributed to chance is 66% multiplied by the rate of failed object identification (Figure 6). If the locations are not remembered perfectly (less than three are remembered or location is remembered erroneously), then the number of swaps should be lower. Thus, if this assumption is not fully supported, the number of swaps in this calculation may overestimate the true number of swaps because of “lucky guesses.” Note that, in the left (contralesional) side, the number of swap errors was larger than the calculated upper limit, especially in the long delay, t(11) = 2.12, p = .06, d = 0.61. On the other hand, the measured number of swap errors was (insignificantly) lower than this upper limit in the right (ipsilesional) side (100 msec: t(11) = −0.87, p = .40, d = −0.25; 4000 msec: t(11) = −1.03, p = .32, d = −0.29). These results attest for the occurrence of swap errors (i.e., misattribution of object location to the location of another object) in conditions where a left-sided object was correctly identified and memory of its identity was maintained across the long retention interval. The direction of swap errors in such cases is from the contralesional to the ipsilesional side.

Figure 6. 

Comparison between the measured number of swap errors and the calculated upper limit of swap errors that could arise in trials where the objects were identified by chance and location memory was preserved.

Figure 6. 

Comparison between the measured number of swap errors and the calculated upper limit of swap errors that could arise in trials where the objects were identified by chance and location memory was preserved.

To examine whether USN patients only tend to recall the location of objects displayed on the right side of the screen, rather than making binding errors, we compared the number of swap errors to the right and to the left, when the target object was located in the center. No directional bias was observed, t(11) = 1.198 p = .26, d = 0.35. We also examined the localization performance for incorrectly identified objects, to see whether there was a tendency for patients to localize the wrongly identified objects in locations close to objects that were displayed on the right or left side of the memory array, F(2, 22) < 1, p = .77, ηp2 = .02. The lack of significant effects in these analyses suggests that rightward directional reporting bias is unlikely to explain the results. However, such conclusion should be taken with caution as the statistical power of these analyses is low because of the small number of trials in each condition. We report these analyses and further discuss it in the supplementary material.

Two measures of RT were recorded (see Figure 7): (1) the time between the presentation of the two objects and the selection of one of the objects (identification time) and (2) the time between the selection and final localization of the object (localization time). Identification time was not affected by the retention interval, F(1, 11) < 1, p = .79, ηp2 = .01, nor the side of presentation, F(1, 11) = 1.55, p = .24, ηp2 = .12. The interaction between Side and Retention Interval was also insignificant, F(1, 11) = 1.43, p = .26, ηp2 = .12. Localization time was significantly longer on the contralesional left side, F(1, 11) = 5.61, p = .04, ηp2 = .34, but the effect of Retention Interval, F(1, 11) < 1, p = .98, ηp2 = .00, and the interaction between the two, F(1, 11) < 1, p = .45, ηp2 = .05, were insignificant.

Figure 7. 

Measures of RT in Experiments 1 and 2. (A) The duration between the presentation of the two objects and the selection of one of the objects (identification duration). (B) The duration between the object selection and the final localization of the object (localization duration). (C) In Experiment 2, the duration between the presentation of the probe and the report.

Figure 7. 

Measures of RT in Experiments 1 and 2. (A) The duration between the presentation of the two objects and the selection of one of the objects (identification duration). (B) The duration between the object selection and the final localization of the object (localization duration). (C) In Experiment 2, the duration between the presentation of the probe and the report.

Experiment 2

Unlike the 2AFC recognition method employed in Experiment 1 for the assessment of object-identity memory, here a more demanding analog report method was employed. The aim of this change was to examine the possibility that the findings in the first experiment reflect the relative ease of arriving to correct identification.

Participants completed an average of 54 trials (SD = 9.68), with the target bar on the left (contralesional side) in about half (51.5 ± 6.3%) of the trials. Errors in object-identity memory were defined as the angular deviation between the reported orientation and the original orientation of the target bar (see also van Ede, Chekroud, & Nobre, 2019). Errors were averaged separately for each participant, display side, and length of retention interval (Figure 8A). Repeated-measures ANOVA with Side (left, right) and Retention interval (1000 msec, 6000 msec) as within-participant factors showed a significant main effect of retention interval, F(1, 10) = 7.46, p = .02, ηp2 = .43, and side, F(1, 10) = 7.18, p = .02, ηp2 = .42, reflecting a disadvantage for objects displayed on the left and after longer retention intervals. The interaction between the two factors was not significant, F(1, 10) < 1, p = .46, ηp2 = .06.

Figure 8. 

Results of Experiment 2: (A) angular errors when the target was displayed on the left (contralesional) and right (ipsilesional) sides, after the two retention interval conditions. (B) Percentage of swap errors. (C) The angular error in each condition after controlling for swap errors. Error bars represent SEMs.

Figure 8. 

Results of Experiment 2: (A) angular errors when the target was displayed on the left (contralesional) and right (ipsilesional) sides, after the two retention interval conditions. (B) Percentage of swap errors. (C) The angular error in each condition after controlling for swap errors. Error bars represent SEMs.

Next, we assessed the amount of swap errors, that is, trials in which the reported answer was closer to the orientation of the nontarget bar rather than to the target bar (in trials where the orientation of the two bars differed by at least 20°; see Figure 8B). Repeated-measures ANOVA with Side (left, right) and Retention interval (1000 msec, 6000 msec) as within-participant factors showed a significant main effect for Retention Interval, F(1, 10) = 6.08, p = .03, ηp2 = .38, and side, F(1, 10) = 6.18, p = .03, ηp2 = .38, reflecting a disadvantage when objects were displayed on the left and after longer retention intervals. The interaction between the two factors was not significant, F(1, 10) < 1, p = .82, ηp2 = .01. The occurrence of more swap errors with left-sided target stimuli and after longer retention time (as found also for orientation performance) suggests that swap errors may play a key role in the production of raw angular errors.

To further explore the contribution of swap errors to the lateral deficit, we removed from the analysis the trials in which swap errors occurred and calculated again the repeated-measures ANOVA on the remaining angular errors (Figure 8C). After excluding swap errors, both the effect of Retention Interval, F(1, 10) < 1, p = .41, ηp2 = .07, and the effect of side, F(1, 10) = 2.77, p = .13, ηp2 = .22, disappeared. The interaction was also nonsignificant, F(1, 10) = 2.84, p = .12, ηp2 = .22. Thus, it seems that both the effects of Retention Interval and Display Side could be explained by swap errors.

The reporting time did not differ significantly between conditions (see Figure 7C; main effect of Retention Interval, F(1, 10) = 2.68, p = .13, ηp2 = .21, side, F(1, 10) < 1, p = .89, ηp2 = .00, and the interaction, F(1, 10) = 1.38, p = .27, ηp2 = .12).

DISCUSSION

Visuospatial WM was examined in USN patients using two established delayed estimation paradigms (Pertzov & Husain, 2014; Pertzov et al., 2012; Bays & Husain, 2008). These paradigms allow for (1) quantitative measurement of the different types of errors and (2) analysis of maintenance across time in WM.

The main findings in Experiment 1 can be summarized as follows: (1) Stroke patients with right-hemisphere damage and left spatial neglect exhibit a significant disadvantage in their capacity to identify objects presented on the left, although their performance in that side was above chance level. This lateralized impairment in immediate memory of object identity was not strongly affected by the length of the retention interval. (2) Memory of object location was also inferior on the left, but here, rapid degradation led to significantly lower localization accuracy after longer retention intervals. (3) The evidence that the amount of retention time in WM affects the memory of object location but not the memory of object identity points to separate forgetting properties for each of them. (4) The number of swap errors was larger on the contralesional side (left) and after longer retention intervals, with post hoc contrasts showing that the main effect of retention interval had its origin in left-side performance, thus suggesting the existence of rapid degradation of the connection between identity and location data of contralesional objects.

In Experiment 2, using a more demanding task to assess the accuracy of object-identity memory, USN patients exhibited disadvantage of contralesional (left) compared to ipsilesional (right) objects as well as inferior performance in long compared to short retention intervals. Thus, when object-identity memory was examined using an analog response method (rather than recognition as in Experiment 1), retention time and not only presentation side affected patients' performance. Swap errors (i.e., trials in which the reported answer was closer to the orientation of the nontarget object rather than to the target object) showed similar side and retention interval effects. Swap errors in this experiment are of interest because they point to data retention in WM, which is erroneously attributed to the target object (especially when the target object is placed in contralesional space), although it pertains to the nontarget object. Loose and unstable connection in WM between object identity and object location information, as shown in Experiment 1, could lead to such errors. Interestingly, the analysis of raw angular errors discarding trials with swap errors diminished the effects of both presentation side and retention interval, suggesting a key role for defective binding (between identity and location data), rather than rapid degradation of the feature memory trace, in the lateralized impairment of USN patients.

An important difference between Experiment 1 and Experiment 2 is that, in Experiment 2, the location information is not explicitly required to perform the task (probing by color and reporting the orientation of the target object). However, reduced recall performance on the contralateral side in Experiment 2 points to an effect of stimulus location on the encoding of object's features. This, however, cannot explain the decrease in precision after longer retention intervals. The diminished effect of presentation side on orientation accuracy when controlling for swap errors suggests that stimulus location plays a role in maintaining nonspatial features and their binding in memory (Schneegans & Bays, 2017; Pertzov & Husain, 2014; Treisman & Zhang, 2006). Impaired ability to use the spatial tag of left-sided objects for conjunction of different stimulus features (e.g., color and orientation) could lead to the noted effects of presentation side on recall performance.

A salient feature of the WM impairment in USN shown in both experiments is the increased amount of swap errors in objects displayed on the contralesional side, especially after longer retention intervals. Swap errors can explain the difference that was found in visual information (i.e., visual WM) and in location information (i.e., spatial WM) between the contralesional and ipsilesional sides as well as between long and short retention intervals. In both experiments, swap errors entail correct use of memory for object identity coupled with incorrect memory for the object's location, that is, a binding failure (Ma et al., 2014; Pertzov, Bays, et al., 2013). Taken together, the results of the two experiments reported here suggest that patients with USN are impaired in their ability to maintain the binding between object identity and object location in WM, when objects are displayed on the contralesional (neglect prone) side of the visual field.

As far as we know, this is the first USN study in which the impairment in WM was decomposed into three components: identity memory, location memory, and the mechanism that binds the two together. Without the specific control for swap errors, as conducted here, the results could have been explained simply as a reflection of contralesional disadvantage in remembering object identity (visual WM) and object location (spatial WM). However, after controlling for swap errors, the difference between the ipsilesional and contralesional sides diminished considerably, thus pointing to an important role of binding in lateralized USN deficits. Loss of object identity and spatial information on the neglected side could lead participants to report the remembered location of one of the objects on the nonneglected sides. However, as we analyzed only trials with correct identification, the amount of such trials is limited by the number of incorrect identifications (if a patient erroneously identified the nontarget object in 10% of the trials, he could be expected to select the target object by chance in 10% of the trials). The empirical number of swap errors on the neglected side exceeded the upper limit derived from identification errors, supporting the contribution of swap errors to the lateralized deficit in USN.

Various lines of evidence point to the fact that the spatial tag of an object plays a crucial role in lending it access to conscious awareness (for a discussion of empirical evidence for that in USN, see Deouell, 2002). The “Piazza Effect” (Bisiach & Luzzatti, 1978), discussed here in the Introduction section, is a classic example of this principle. Preserved and deranged spatial tags of ipsilesional and contralesional objects, respectively, enabled conscious recollection of data, once from the northern part and once from the southern part of the square, depending on the position from which the patients were asked to imagine themselves viewing the square. It is clear that information from both sides of the square is properly maintained in long-term memory, but access to conscious awareness was denied from objects associated with locations on the contralesional space.

A similar phenomenon could occur in this study (but for short-term rather than long-term memory). USN patients could have captured and maintained in WM the features of visual objects and possibly also their location data, even on the neglected side. However, when the patients had to make their conscious report, they failed to reach the correct location information. The lateralized memory impairment after short retention intervals, with minimal retention demands, could possibly reflect a selective lateralized impairment in retrieval of location information, whereas the additional amount of swap errors after longer retention intervals could reflect the added lateralized deficit in maintaining the data in WM. The longer localization time when target objects were displayed on the neglected side lends some support for a lateralized deficit in retrieving information on the contralesional side, rather than a pure binding deficit. In case of a pure lateralized binding deficit, we would expect to see no difference between ipsilateral and contralateral sides in localization time.

Of the different possible interpretations of the results, we think that a lateralized binding deficit explains the main findings in a more coherent manner. First, the analysis of localization performance when discarding trials with swap errors revealed no significant lateralized deficit in localization precision. Second, most of the location swaps in Experiment 1 had a left-to-right direction (i.e., correctly identified objects in contralesional space were localized erroneously at or near the memorized locations of objects in the ipsilesional space). Patients hardly placed objects in random locations but rather in object locations correctly maintained on the ipsilesional side. A lateralized deficit in location memory is also expected to lead to a systematic shift in location reports toward the right. However, such a report bias was not observed, but rather a systematic shift toward the center of the array, presumably because of some coarse representation of the stimuli that rely on ensemble statistics (Katshu & d'Avossa, 2014; Brady & Alvarez, 2011).

The occurrence of left-to-right swap errors, as documented here within the realm of WM impairment in USN, is known from research of other aspects of the neglect syndrome. In certain conditions, misattribution of left-sided identity information to right-sided locations was shown to play a “positive” role, enabling conscious access to contralesional stimuli that otherwise would have been neglected. For example, in a study aimed to assess the effect of stimulation side on the ability of left-USN patients to identify consonant–vowel auditory stimuli, a clear disadvantage for contralesional (left-sided) stimuli disappeared by application of the ventriloquist illusion, whereby the patients were led to believe that the stimuli were emitted from a (fictitious) loudspeaker viewed on the ipsilesional side (right; Calamaro, Myslobodsky, & Soroker, 1995; Calamaro, Soroker, & Myslobodsky, 1995). Likewise, USN patients were shown to form McGurk illusory blend percepts in a normal rate when they heard consonants–vowels emitted from the left (thus normally prone to be neglected) in conjunction with lipreading of discordant stimuli in front of them (Soroker, Calamaro, & Myslobodsky, 1995; see also Deouell, Deutsch, Scabini, Soroker, & Knight, 2008, for another example of swap errors aiding perception in USN). Naturally occurring left-to-right swaps are frequent also in conditions of bilateral simultaneous stimulation in USN and may help patients perceive contralesional stimuli prone to extinction (Deouell & Soroker, 2000).

Beyond the crucial role of impaired binding of identity and location data in the WM pathology of USN, the findings of Experiment 1 shed light on other important WM attributes. The contralesional relative inferiority in patients' capacity to recognize previously shown objects was essentially similar 100 and 4000 msec after the end of the memory display phase. Moreover, the proportion of correct recognition of objects remained stable throughout this time interval, both in the contralesional and ipsilesional sides. Thus, beyond the initial attenuation of object-identity registration in the contralesional neglect-prone space, there is no evidence here for a further degradation of the memory trace neither in the ipsilesional nor in the contralesional sides. Object-identity memory remained stable, with no signs of forgetting up to 4000 msec, in the entire field. When object-identity memory was tested using a more demanding and sensitive delayed estimation task (Experiment 2), performance was again inferior in the contralesional side, but here the accuracy with which patients remembered the inclination of the target bar deteriorated (in both sides) with the lengthening of the retention interval. However, this effect of retention interval disappeared when swap errors were controlled, thus corroborating the conclusion that object identity memory in itself may be maintained in USN at a stable rate throughout the retention intervals used in this study.

In contrast, location memory of correctly identified objects was shown to deteriorate along the retention interval. The temporal degradation occurred at an equal rate for stimuli located in the contralesional and ipsilesional visual fields, thus keeping the initial difference between the contralesional and ipsilesional sides at a more or less similar level across retention times. Inasmuch as the spatially lateralized component of the USN syndrome is concerned (Husain & Rorden, 2003), we did not find evidence that it affects the forgetting rate, in neither visual (object-identity) nor spatial (object location) WM. It did seem to have an effect, however, on the likelihood of an object representation becoming separated from its corresponding spatial tag during data maintenance in WM. This in turn led to an increased rate of contralesional objects being erroneously located, after a long retention time, at the memorized location of ipsilesional objects (swap errors).

The finding that the binding between location and identity memory may be impaired in USN echoes core theoretical accounts of visual information processing, recognizing a division between identity (what) processing in the temporo-occipital cortex and location (where) processing in the parieto-occipital cortex (Schneegans, Spencer, & Schöner, 2015; Ungerleider & Mishkin, 1982; but see Kravitz, Saleem, Baker, Ungerleider, & Mishkin, 2013; Kravitz, Kriegeskorte, & Baker, 2010). Concomitant processing of identity and location data becomes crucial at retrieval, when access to conscious awareness is required. On the basis of the findings of the current study, we propose that the association between identity and location representations of objects, presented in the contralesional side of space, is impaired in USN.

A limitation of the procedure in Experiment 1 that is worth mentioning is the fact that the identification and localization stages of the experiment pose different processing demands. The identification stage is a 2AFC recognition task, which could lead to the correct response on the basis of familiarity, with a 50% chance of reaching the correct answer by guessing. Thus, the information provided by patients' performance on the quality of their visual memory is limited. Remembering the correct object in the localization stage is more demanding, as there are three alternatives, and memory of the object's locations becomes relevant to the task. Thus, any comparison between the identification and localizations stages should be taken with caution.

In conclusion, USN patients were found to express contralesional disadvantage in both visual and spatial WM (i.e., memory for object identity and object location, respectively). However, careful analysis disclosed an increased tendency to erroneously report the feature of the ipsilesional object when asked about the contralesional object. These “swap” errors point to degraded binding between identity and location data of contralesional objects. When swap errors were excluded from the analyses, side and retention time effects on memory performance disappeared. Inasmuch as the spatially lateralized component of the USN syndrome is concerned, we did not find evidence that it affects forgetting rate in neither visual nor spatial WM. It did seem to attenuate registration in the contralesional neglect-prone space, and it affected the likelihood of an object representation becoming separated from its corresponding spatial tag in WM. We conclude that defective binding between identity and location information plays a key role in the WM pathology of USN patients.

Acknowledgments

This work was supported by Israel Science Foundation grant 1747/14 to Y. P.

Reprint requests should be sent to Yoni Pertzov, Department of Psychology, The Hebrew University of Jerusalem, Mount Scopus, Jerusalem 9190501, Israel, or via e-mail: yoni.pertzov@mail.huji.ac.il.

Note

REFERENCES

REFERENCES
Allen
,
R. J.
,
Baddeley
,
A. D.
, &
Hitch
,
G. J.
(
2006
).
Is the binding of visual features in working memory resource-demanding?
Journal of Experimental Psychology: General
,
135
,
298
313
.
Azouvi
,
P.
,
Jacquin-Courtois
,
S.
, &
Luauté
,
J.
(
2017
).
Rehabilitation of unilateral neglect: Evidence-based medicine
.
Annals of Physical and Rehabilitation Medicine
,
60
,
191
197
.
Baddeley
,
A. D.
(
2000
).
The episodic buffer: A new component of working memory?
Trends in Cognitive Sciences
,
4
,
417
423
.
Baddeley
,
A. D.
,
Allen
,
R. J.
, &
Hitch
,
G. J.
(
2011
).
Binding in visual working memory: The role of the episodic buffer
.
Neuropsychologia
,
49
,
1393
1400
.
Bays
,
P. M.
,
Catalao
,
R. F. G.
, &
Husain
,
M.
(
2009
).
The precision of visual working memory is set by allocation of a shared resource
.
Journal of Vision
,
9
,
7
.
Bays
,
P. M.
,
Gorgoraptis
,
N.
,
Wee
,
N.
,
Marshall
,
L.
, &
Husain
,
M.
(
2011
).
Temporal dynamics of encoding, storage, and reallocation of visual working memory
.
Journal of Vision
,
11
,
6
.
Bays
,
P. M.
, &
Husain
,
M.
(
2008
).
Dynamic shifts of limited working memory resources in human vision
.
Science
,
321
,
851
854
.
Beschin
,
N.
,
Cocchini
,
G.
,
Della Sala
,
S.
, &
Logie
,
R. H.
(
1997
).
What the eyes perceive, the brain ignores: A case of pure unilateral representational neglect
.
Cortex
,
33
,
3
26
.
Bisiach
,
E.
, &
Luzzatti
,
C.
(
1978
).
Unilateral neglect of representational space
.
Cortex
,
14
,
129
133
.
Brady
,
T. F.
, &
Alvarez
,
G. A.
(
2011
).
Hierarchical encoding in visual working memory: Ensemble statistics bias memory for individual items
.
Psychological Science
,
22
,
384
392
.
Brainard
,
D. H.
(
1997
).
The Psychophysics Toolbox
.
Spatial Vision
,
10
,
433
436
.
PMID:9176952
Calamaro
,
N.
,
Myslobodsky
,
M. S.
, &
Soroker
,
N.
(
1995
).
Ventriloquist effect reinstates responsiveness to auditory stimuli in the ‘ignored’ space in patients with hemispatial neglect
.
Journal of Clinical and Experimental Neuropsychology
,
17
,
243
255
.
Calamaro
,
N.
,
Soroker
,
N.
, &
Myslobodsky
,
M. S.
(
1995
).
False recovery from auditory hemineglect produced by source misattribution of auditory stimuli (the ventriloquist effect)
.
Restorative Neurology and Neuroscience
,
7
,
151
156
.
Cohen-Dallal
,
H.
,
Fradkin
,
I.
, &
Pertzov
,
Y.
(
2018
).
Are stronger memories forgotten more slowly? No evidence that memory strength influences the rate of forgetting
.
PLoS One
,
13
,
e0200292
.
Corbetta
,
M.
, &
Shulman
,
G. L.
(
2011
).
Spatial neglect and attention networks
.
Annual Review of Neuroscience
,
34
,
569
599
.
Damasio
,
H.
, &
Damasio
,
A. R.
(
1989
).
Lesion analysis in neuropsychology
.
New York
:
Oxford University Press
.
Darling
,
S.
,
Della Sala
,
S.
, &
Logie
,
R. H.
(
2007
).
Behavioural evidence for separating components within visuo-spatial working memory
.
Cognitive Processing
,
8
,
175
181
.
Darling
,
S.
,
Della Sala
,
S.
,
Logie
,
R. H.
, &
Cantagallo
,
A.
(
2006
).
Neuropsychological evidence for separating components of visuo-spatial working memory
.
Journal of Neurology
,
253
,
176
180
.
Della Sala
,
S.
,
Logie
,
R. H.
,
Beschin
,
N.
, &
Denis
,
M.
(
2004
).
Preserved visuo-spatial transformations in representational neglect
.
Neuropsychologia
,
42
,
1358
1364
.
Deouell
,
L. Y.
(
2002
).
Pre-requisites for conscious awareness: Clues from electrophysiological and behavioral studies of unilateral neglect patients
.
Consciousness and Cognition
,
11
,
546
567
.
Deouell
,
L. Y.
,
Deutsch
,
D.
,
Scabini
,
D.
,
Soroker
,
N.
, &
Knight
,
R. T.
(
2008
).
No disillusions in auditory extinction: Perceiving a melody comprised of unperceived notes
.
Frontiers in Human Neuroscience
,
1
,
15
.
Deouell
,
L. Y.
,
Sacher
,
Y.
, &
Soroker
,
N.
(
2005
).
Assessment of spatial attention after brain damage with a dynamic reaction time test
.
Journal of the International Neuropsychological Society
,
11
,
697
707
.
Deouell
,
L. Y.
, &
Soroker
,
N.
(
2000
).
What is extinguished in auditory extinction?
NeuroReport
,
11
,
3059
3062
.
Dundon
,
N. M.
,
Katshu
,
M. Z. U. H.
,
Harry
,
B.
,
Roberts
,
D.
,
Leek
,
E. C.
,
Downing
,
P.
, et al
(
2018
).
Human parahippocampal cortex supports spatial binding in visual working memory
.
Cerebral Cortex
,
28
,
3589
3599
.
Friedman-Hill
,
S. R.
,
Robertson
,
L. C.
, &
Treisman
,
A.
(
1995
).
Parietal contributions to visual feature binding: Evidence from a patient with bilateral lesions
.
Science
,
269
,
853
855
.
Gorgoraptis
,
N.
,
Catalao
,
R. F. G.
,
Bays
,
P. M.
, &
Husain
,
M.
(
2011
).
Dynamic updating of working memory resources for visual objects
.
Journal of Neuroscience
,
31
,
8502
8511
.
Halligan
,
P. W.
,
Marshall
,
J. C.
, &
Wade
,
D. T.
(
1989
).
Visuospatial neglect: Underlying factors and test sensitivity
.
Lancet
,
334
,
908
911
.
Haramati
,
S.
,
Soroker
,
N.
,
Dudai
,
Y.
, &
Levy
,
D. A.
(
2008
).
The posterior parietal cortex in recognition memory: A neuropsychological study
.
Neuropsychologia
,
46
,
1756
1766
.
Husain
,
M.
, &
Rorden
,
C.
(
2003
).
Non-spatially lateralized mechanisms in hemispatial neglect
.
Nature Reviews Neuroscience
,
4
,
26
36
.
Karnath
,
H.-O.
, &
Rorden
,
C.
(
2012
).
The anatomy of spatial neglect
.
Neuropsychologia
,
50
,
1010
1017
.
Katshu
,
M. Z. U. H.
, &
d'Avossa
,
G.
(
2014
).
Fine-grained, local maps and coarse, global representations support human spatial working memory
.
PLoS One
,
9
,
e107969
.
Kleiner
,
M.
,
Brainard
,
D.
,
Pelli
,
D.
,
Ingling
,
A.
,
Murray
,
R.
, &
Broussard
,
C.
(
2007
).
What's new in Psychtoolbox-3?
Perception
,
36
,
1
16
.
Kravitz
,
D. J.
,
Kriegeskorte
,
N.
, &
Baker
,
C. I.
(
2010
).
High-level visual object representations are constrained by position
.
Cerebral Cortex
,
20
,
2916
2925
.
Kravitz
,
D. J.
,
Saleem
,
K. S.
,
Baker
,
C. I.
,
Ungerleider
,
L. G.
, &
Mishkin
,
M.
(
2013
).
The ventral visual pathway: An expanded neural framework for the processing of object quality
.
Trends in Cognitive Sciences
,
17
,
26
49
.
Lancaster
,
J. L.
,
Woldorff
,
M. G.
,
Parsons
,
L. M.
,
Liotti
,
M.
,
Freitas
,
C. S.
,
Rainey
,
L.
, et al
(
2000
).
Automated Talairach atlas labels for functional brain mapping
.
Human Brain Mapping
,
10
,
120
131
.
Liang
,
Y.
,
Pertzov
,
Y.
,
Nicholas
,
J. M.
,
Henley
,
S. M. D.
,
Crutch
,
S.
,
Woodward
,
F.
, et al
. (
2016
).
Visual short-term memory binding deficit in familial Alzheimer's disease
.
Cortex
,
78
,
150
164
.
Lowery
,
N.
,
Ragland
,
J. D.
,
Gur
,
R. C.
,
Gur
,
R. E.
, &
Moberg
,
P. J.
(
2004
).
Normative data for the symbol cancellation test in young healthy adults
.
Applied Neuropsychology
,
11
,
216
219
.
Luauté
,
J.
,
Halligan
,
P.
,
Rode
,
G.
,
Rossetti
,
Y.
, &
Boisson
,
D.
(
2006
).
Visuo-spatial neglect: A systematic review of current interventions and their effectiveness
.
Neuroscience and Biobehavioral Reviews
,
30
,
961
982
.
Ma
,
W. J.
,
Husain
,
M.
, &
Bays
,
P. M.
(
2014
).
Changing concepts of working memory
.
Nature Neuroscience
,
17
,
347
356
.
Malhotra
,
P.
,
Jäger
,
H. R.
,
Parton
,
A.
,
Greenwood
,
R.
,
Playford
,
E. D.
,
Brown
,
M. M.
, et al
(
2005
).
Spatial working memory capacity in unilateral neglect
.
Brain
,
128
,
424
435
.
Manohar
,
S. G.
,
Pertzov
,
Y.
, &
Husain
,
M.
(
2017
).
Short-term memory for spatial, sequential and duration information
.
Current Opinion in Behavioral Sciences
,
17
,
20
26
.
Moreh
,
E.
,
Malkinson
,
T. S.
,
Zohary
,
E.
, &
Soroker
,
N.
(
2014
).
Visual memory in unilateral spatial neglect: Immediate recall versus delayed recognition
.
Journal of Cognitive Neuroscience
,
26
,
2155
2170
.
Mort
,
D. J.
,
Malhotra
,
P.
,
Mannan
,
S. K.
,
Rorden
,
C.
,
Pambakian
,
A.
,
Kennard
,
C.
, et al
(
2003
).
The anatomy of visual neglect
.
Brain
,
126
,
1986
1997
.
Pertzov
,
Y.
,
Bays
,
P. M.
,
Joseph
,
S.
, &
Husain
,
M.
(
2013
).
Rapid forgetting prevented by retrospective attention cues
.
Journal of Experimental Psychology: Human Perception and Performance
,
39
,
1224
1231
.
Pertzov
,
Y.
,
Dong
,
M. Y.
,
Peich
,
M.-C.
, &
Husain
,
M.
(
2012
).
Forgetting what was where: The fragility of object-location binding
.
PLoS One
,
7
,
e48214
.
Pertzov
,
Y.
,
Heider
,
M.
,
Liang
,
Y.
, &
Husain
,
M.
(
2015
).
Effects of healthy ageing on precision and binding of object location in visual short term memory
.
Psychology and Aging
,
30
,
26
35
.
Pertzov
,
Y.
, &
Husain
,
M.
(
2014
).
The privileged role of location in visual working memory
.
Attention, Perception, & Psychophysics
,
76
,
1914
1924
.
Pertzov
,
Y.
,
Miller
,
T. D.
,
Gorgoraptis
,
N.
,
Caine
,
D.
,
Schott
,
J. M.
,
Butler
,
C.
, et al
(
2013
).
Binding deficits in memory following medial temporal lobe damage in patients with voltage-gated potassium channel complex antibody-associated limbic encephalitis
.
Brain
,
136
,
2474
2485
.
Pierce
,
S. R.
, &
Buxbaum
,
L. J.
(
2002
).
Treatments of unilateral neglect: A review
.
Archives of Physical Medicine and Rehabilitation
,
83
,
256
268
.
Pisella
,
L.
,
Berberovic
,
N.
, &
Mattingley
,
J. B.
(
2004
).
Impaired working memory for location but not for colour or shape in visual neglect: A comparison of parietal and non-parietal lesions
.
Cortex
,
40
,
379
390
.
Prinzmetal
,
W.
,
Amiri
,
H.
,
Allen
,
K.
, &
Edwards
,
T.
(
1998
).
The phenomenology of attention: I. Color, location, orientation, and spatial frequency
.
Journal of Experimental Psychology: Human Perception and Performance
,
24
,
261
282
.
Rhodes
,
S.
,
Parra
,
M. A.
,
Cowan
,
N.
, &
Logie
,
R. H.
(
2017
).
Healthy aging and visual working memory: The effect of mixing feature and conjunction changes
.
Psychology and Aging
,
32
,
354
366
.
Schneegans
,
S.
, &
Bays
,
P. M.
(
2017
).
Neural architecture for feature binding in visual working memory
.
Journal of Neuroscience
,
37
,
3913
3925
.
Schneegans
,
S.
, &
Bays
,
P. M.
(
2019
).
New perspectives on binding in visual working memory
.
British Journal of Psychology
,
110
,
207
244
.
Schneegans
,
S.
,
Spencer
,
J. P.
, &
Schöner
,
G.
(
2015
).
Integrating “what” and “where”: Visual working memory for objects in a scene
. In
Dynamic thinking: A primer on dynamic field theory
(pp.
197
226
).
New York
:
Oxford University Press
.
Solomon
,
J.
,
Raymont
,
V.
,
Braun
,
A.
,
Butman
,
J. A.
, &
Grafman
,
J.
(
2007
).
User-friendly software for the analysis of brain lesions (ABLe)
.
Computer Methods and Programs in Biomedicine
,
86
,
245
254
.
Soroker
,
N.
,
Calamaro
,
N.
, &
Myslobodsky
,
M.
(
1995
).
“McGurk illusion” to bilateral administration of sensory stimuli in patients with hemispatial neglect
.
Neuropsychologia
,
33
,
461
470
.
Spotorno
,
S.
,
Evans
,
M.
, &
Jackson
,
M. C.
(
2018
).
Remembering who was where: A happy expression advantage for face identity–location binding in working memory
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
44
,
1365
1383
.
Stollery
,
B.
, &
Christian
,
L.
(
2016
).
Glucose improves object–location binding in visual–spatial working memory
.
Psychopharmacology
,
233
,
529
547
.
Treisman
,
A.
, &
Zhang
,
W.
(
2006
).
Location and binding in visual working memory
.
Memory & Cognition
,
34
,
1704
1719
.
Tzourio-Mazoyer
,
N.
,
Landeau
,
B.
,
Papathanassiou
,
D.
,
Crivello
,
F.
,
Etard
,
O.
,
Delcroix
,
N.
, et al
(
2002
).
Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain
.
Neuroimage
,
15
,
273
289
.
Ungerleider
,
L. G.
, &
Mishkin
,
M.
(
1982
).
Two cortical visual systems
. In
D. J.
Ingle
,
M. A.
Goodale
, &
R. J. W.
Mansfield
(Eds.),
Analysis of visual behavior
(pp.
549
586
).
Cambridge
:
MIT Press
.
van Asselen
,
M.
,
Kessels
,
R. P. C.
,
Frijns
,
C. J. M.
,
Jaap Kappelle
,
L.
,
Neggers
,
S. F. W.
, &
Postma
,
A.
(
2009
).
Object–location memory: A lesion–behavior mapping study in stroke patients
.
Brain and Cognition
,
71
,
287
294
.
van Asselen
,
M.
,
Kessels
,
R. P. C.
,
Kappelle
,
L. J.
, &
Postma
,
A.
(
2008
).
Categorical and coordinate spatial representations within object–location memory
.
Cortex
,
44
,
249
256
.
van Ede
,
F.
,
Chekroud
,
S. R.
, &
Nobre
,
A. C.
(
2019
).
Human gaze tracks attentional focusing in memorized visual space
.
Nature Human Behaviour
,
3
,
462
470
.
Vuilleumier
,
P.
(
2013
).
Mapping the functional neuroanatomy of spatial neglect and human parietal lobe functions: Progress and challenges
.
Annals of the New York Academy of Sciences
,
1296
,
50
74
.
Weintraub
,
S.
, &
Mesulam
,
M.-M.
(
1987
).
Right cerebral dominance in spatial attention: Further evidence based on ipsilateral neglect
.
Archives of Neurology
,
44
,
621
625
.
Wilson
,
B.
,
Cockburn
,
J.
, &
Halligan
,
P.
(
1987
).
Development of a behavioral test of visuospatial neglect
.
Archives of Physical Medicine and Rehabilitation
,
68
,
98
102
.
PMID:3813864
Wojciulik
,
E.
,
Husain
,
M.
,
Clarke
,
K.
, &
Driver
,
J.
(
2001
).
Spatial working memory deficit in unilateral neglect
.
Neuropsychologia
,
39
,
390
396
.
Zhang
,
W.
, &
Luck
,
S. J.
(
2008
).
Discrete fixed-resolution representations in visual working memory
.
Nature
,
453
,
233
235
.
Zokaei
,
N.
,
Burnett Heyes
,
S.
,
Gorgoraptis
,
N.
,
Budhdeo
,
S.
, &
Husain
,
M.
(
2015
).
Working memory recall precision is a more sensitive index than span
.
Journal of Neuropsychology
,
9
,
319
329
.
Zokaei
,
N.
,
Nour
,
M. M.
,
Sillence
,
A.
,
Drew
,
D.
,
Adcock
,
J.
,
Stacey
,
R.
, et al
(
2019
).
Binding deficits in visual short-term memory in patients with temporal lobe lobectomy
.
Hippocampus
,
29
,
63
67
.