Thus far we have provided neuropsychological evidence to argue that visuospatial working memory is best viewed as an active workspace, rather than a passive memory store, and that it is somewhat distanced from perceptual processes. A range of patient data speak to the notion that visual and spatial working memory might be seen as two distinct but linked components of the cognitive system. For the purposes of this argument, it is important to clarify what we mean by "visual" as opposed to "spatial" properties. By "visual", we refer to the visual appearance of an object or scene, its colour, shape, contrast, size, visual texture and the location of objects relative to one another with respect to a particular viewpoint in a static array. By "spatial" we refer to pathways or sequences of movements from one location to another in the scene, or the processes of change in the perceived relative locations of objects that occur when an observer moves (physically or in a mental image) from one viewpoint to another. There is some ambiguity in the literature as to the use of the word "spatial", which sometimes is used to refer to relative locations or layouts of objects. In the description of neuropsychological impairments that we discuss, it should become clear that it is more useful to think of the "spatial" as referring to the dynamic properties of a scene or representation (e.g. Logie, 1995; Quinn & Ralston, 1986; Smyth & Pendleton, 1990).
The idea that visual and spatial working memory, according to our definitions above, comprise dissociable components of the cognitive system has been supported by a range of studies of healthy participants. For example, in healthy adults visual immediate memory tasks appear to be sensitive to disruption by secondary tasks different from those that disrupt immediate memory for movement sequences (Della Sala et al., 1999; Hecker & Mapperson, 1997; Logie & Marchetti, 1991; Tresch et al., 1993).
This separation in the representations held within working memory is sometimes thought to be similar to a distinction between perceiving object identity and object location derived from single-cell recording in studies of monkeys (Ungerleider & Mishkin, 1982). These studies demonstrated that monkeys attempting to identify an object rely on ventral/temporal pathways, while identifying the location of an object involves dorsal/parietal pathways. There are several problems with the apparently compelling link between the cognitive functions of visuospatial working memory and the neuroanatomical pathways associated with the "what" and the "where" of visual perception. One difficulty arises from the fact that the visual/location/movement distinction within working memory applies to the representation held and manipulated within working memory. It does not refer to the processes of perceiving and identifying an object and its current location. We have already argued that the processes of perception and the operation of working memory are less closely linked than has been widely assumed. Moreover, the ability to detect the location of objects and orientate attention towards them appears to be a fundamental, built-in property of the perception and attention systems. It can be performed by infants who have very limited knowledge and experience of objects. In stark contrast, object identification requires prior experience with objects, their associated labels, uses and properties. However, the information held in working memory incorporates these associated properties along with information about location, again arguing that a separation between "what" and "where" might be relevant for perception but not for visuospatial working memory.
A further problem is that the concept of the so-called "what" and "where" pathways is overly simplistic at a neuroanatomical level. There are multiple connections and pathways involved following initial processing of sensory input within the primary visual cortex (e.g. Hof et al., 1990; Stein, 1992; Zeki & Shipp, 1988; Zeki, cited in Della Sala, 1998). Moreover, the representations that we hold in working memory incorporate information from several sensory modalities (auditory, haptic, kinaesthetic, and possibly even olfactory and gustatory) in addition to elements of prior knowledge not immediately available from the perceived scene.
The apparent similarity has been taken further in linking visual appearance and location information with activation patterns in brain imaging studies using PET and fMRI. For example, Jonides et al. (1993) tested two groups of participants on visual and location immediate memory tasks while undergoing functional brain imaging using PET. In the location task, volunteers were briefly shown dots at random positions on a screen. After a short interval they were shown a visual cue and had to indicate whether or not the cue identified the location of one of the previously presented dots. In the visual task, participants were presented with unfamiliar shapes, and were later shown a second shape. The task was to indicate whether or not the initial and the comparison shape were identical. Both tasks were performed in a memory condition as described above, and in a perceptual condition, in which the target locations or shapes remained on the screen while the comparison took place. When subtracting out the perceptual conditions from the brain activation data, there were clearly different neuroanatomical networks associated with memory for object shape, primarily in the left hemisphere, and memory for object location, primarily in the right hemisphere. Jonides and colleagues (e.g. see Smith & Jonides, 1995) interpreted these different activation patterns as reflecting the operation of the "what" and "where" pathways, and equated different components of working memory with particular sensory input channels.
Subsequent studies with PET and fMRI (e.g. Courtney et al., 1996, 1998; Haxby et al., 2000) using different stimuli, such as faces, have likewise shown a neuroanatomical segregation between memory for object identity (primarily ventral prefrontal cortex), and memory for object location (primarily dorsal prefrontal cortex). It is striking that, although the neu-roanatomical correlates show dissociations that echo those found in behavioural data, the locations involved appear to vary from study to study, possibly as a result of the use of rather different materials and procedures that have rather different general cognitive demands. In more recent studies (e.g. Owen et al., 1998; 1999; Petrides, 1996), the general cognitive demands of visual and location tasks were separately manipulated. The more demanding tasks appeared to be associated with the more dorsal areas, while the less demanding tasks were linked with activation in the more ventral areas. However, the same prefrontal areas in both hemispheres were associated with visual and location tasks when the overall task demands were equated, i.e. the neuroanatomical segregation appears to be associated with task demand, and not with a contrast between object identity and object location.
This conclusion raises an important issue with regard to the choice of tasks that are intended to explore distinctions among specialized cognitive resources. Tasks vary between studies, and rarely is there a formal task analysis carried out to ascertain the precise nature as well as the extent of the demands on the cognitive system. Tasks are described as "spatial" or "visual", but there is no guarantee that they are using the cognitive resources that are assumed, and volunteers may adopt a range of cognitive strategies, including verbal labelling of parts of a visual array, or visual encoding of a visually presented verbal sequence for recall (e.g. Logie et al., 2000). The pattern of brain activation will reflect the cognitive processes that volunteers bring to bear when performing each task, and the cognitive strategy that they choose could result in gross differences in the activation patterns observed. In purely behavioural studies, this difficulty is often addressed by testing large numbers of volunteers on multiple trials, so that individual differences in strategy are less influential in the overall data pattern. However, brain imaging studies often use rather small numbers of volunteers, and therefore individual differences in strategies could have a very large influence on the activation patterns observed (Logie et al., in press; Savage et al., 2001). Therefore, we have to be cautious when interpreting brain activation patterns, in the absence of converging evidence from other sources, such as behavioural data fromhealthy volunteers or from braindamaged patients for whom the lesion site is known. The brain imaging data are consistent with the notion that locational and visual working memory might be separable, but it is important not to focus too literally on the specific neuroanatomical areas that appear to be active when people perform these tasks. Moreover, it remains unclear precisely what a spatial or visual working memory system might comprise. For example, a requirement to retain locational or movement information might be simply more demanding of cognitive resources than is retention of the visual appearance of an object or shape. The visual/spatial distinction could then reflect a visual memory system coupled with an amodal executive resource that supports retention of novel layouts and sequences of movements (Logie et al., 2001).
The dissociation between visual and spatial working memory has been reported also in studies of nonhuman primates (Goldman-Rakic, 1996; Meunier et al., 1997), electrophysiological studies in healthy adults (Ruchkin et al., 1997; MartinLoeches, 1997) and neuropsychological studies of brain-damaged patients, on whom we will focus.
The experimental dissociations found in healthy volunteers mirror different patterns of impairment and sparing of visual and spatial working memory function found in neuropsychological patients. The dissociations have been shown in contrasts between single cases as well as between groups of patients. The group studies have contrasted performance on a test of visual pattern memory (Phillips & Baddeley, 1971), with retention of targeted movement sequences (Milner, 1971), sometimes referred to as 'Corsi blocks' (Corsi, 1972). Pattern memory involves presenting a matrix of squares, half of which are filled in at random, while avoiding recognizable patterns, such as letter shapes or canonical shapes (see Figure 13.4). The participants are then required to retrieve the patterns. The Corsi task involves the experimenter pointing to a sequence of blocks arranged randomly on aboard (see Figure 13.5). The participant has then to repeat from memory the sequence of movements in the same order. One of the group studies (Grossi et al., 1993) examined 39 Alzheimer patients and found two patients, one of whom showed very poor performance on visual pattern memory but had normal memory for retention of targeted movement sequences, while the other showed the converse pattern. Another group study (Della Sala et al., 1999) reported performance
patterns for 69 patients. Two of the patients performed poorly on retention of targeted movements but were above the median on visual pattern memory, while a further patient showed the converse. All three patients had a lesion in the left hemisphere.
Across the whole group of patients, there was no clear link between right or left hemisphere damage and performance on either of the tasks. The same is true of other studies reporting performance on versions of the Corsi block task. De Renzi & Nichelli (1975) acknowledged that the size of the lesion, as measured by the presence or absence of a visual field defect, "not the hemispheric side of the lesion, was a significant factor in impairing the performance" (p. 346). The results were replicated in a second group study (De Renzi et al., 1977), in which the authors reported that spatial span, as measured by a version of Corsi blocks, was "affected by involvement of either side of the posterior region of the brain" (p. 430). More recently, Feigenbaum et al. (1996) reported performance on a set of three location and spatial manipulation tasks in a group of 40 neurosurgical patients. Patients with right hemisphere lesions performed well below the levels observed for healthy controls on all three tasks, whereas patients with left hemisphere lesions performed poorly overall on only one of the tasks. However, overall performance levels between the two groups of patients did not differ statistically when they were compared with each other rather than with controls.
In summary, the behavioural data from the literature indicate a clear dissociation in performance of the visual pattern memory and immediate memory for location or mental spatial manipulation, but there is no clear lateralization of function for movement sequence memory or for visual pattern memory. This again suggests that a simple mapping of well-established behavioural dissociations onto neuroanatomy might be too simplistic.
It is worth noting, too, that different patients could perform poorly on a task for very different reasons. In addition to being open to the use of different strategies, most tests of cognitive function require several components of cognition for successful performance, and damage to any one component could disrupt overall performance on the tasks. For example, the Corsi blocks task is often assumed to assess a spatial working memory system; however, retaining a sequence of targeted movements requires memory for observed movement, and for a pathway between objects. It might also be crucial whether the target positions are coded relative to body position or relative to one another (e.g. Postma et al., 2000). Depending on the nature of the cognitive impairment following brain damage, performance could be poor because of deficits in any or all of the above cognitive functions. Performance could also be poor because of the application of possible compensatory strategies that patients have developed as a result of their impairments. The compensatory strategy that they adopt may be suboptimal for the task concerned. In other words, Corsi blocks cannot be a "pure" test of a spatial working memory system, unless we consider that spatial working memory comprises visual and movement information as well as retention of sequential information, and a decision process that is required as participants choose which block to touch next in the sequence.
Despite the equivocal nature of the possible link between specific neuroanatomical sites and performance on spatial or visual tasks, a number of authors have interpreted patterns from group studies, as well as from single cases, as indicating that damage to the posterior areas of the right hemisphere are linked to poor performance on Corsi blocks and on other location/movement memory tasks (De Renzi, 1982; Nichelli, 1999; Vallar & Papagno, 1995). This contrasts with the brain-imaging data from healthy volunteers, described above, which on the whole indicated involvement of the prefrontal cortex in location memory tasks or in the mental representation of movement. This form of discrepancy is not unique. Analogous discrepancies are apparent when comparing the neuroanatomical correlates of word comprehension and reading in the healthy brain with those identified in neuropsychological patients (Abbott, 2001).
The dissociation between visual- and location/movement-based working memory gains support from the patterns of impairment observed in a number of individual case studies. Farah et al. (1988) reported patient L.H. who, as a result of a closed head injury in an automobile accident, suffered damage in both temporal/occipital areas, in the right temporal lobe and in the right inferior frontal lobe. He performed well on tasks concerned with memory for locations and for pathways, such as letter rotation, 3-D form rotation, mental scanning, and recalling a recently described pathway, but was severely impaired in his ability to remember colours, the relative size of objects and shapes of States in the map of the USA. A similar case was reported more recently by Wilson et al. (1999). Their patient, L.E., was a professional sculptress who, following systemic lupus erythematosus, resulting in diffuse damage to both the cortex and the white matter, was unable to generate visual images of possible sculptures, and had a severe visual short-term memory deficit. Her deficits included very poor performance on the Doors test (Baddeley et al., 1994), which involves recognizing a door amongst very similar distracters, and on retention of visual matrix patterns. However she could draw complex figures that did not rely on memory, and performed within the low normal range for Corsi block span.
Contrasting cases have been reported by Luzzatti et al. (1998) and by Carlesimo et al. (2001). The Luzzatti et al. patient, E.P., mentioned earlier, was affected by a slowly progressive deterioration of the brain and showed a focal atrophy of the anterior part of the right temporal lobe, including the hippocampus. Her performance was flawless on visual imagery tasks, such as making judgements about relative animal size, or the relative shapes or colours of objects. On the other hand, she was impaired on a range of topographical tasks, such as describing from memory the relative locations of landmarks in her home town, as well as in mentally manipulating the contents of images. A similar pattern was reported for the Carlesimo et al. (2001) patient, M.V., who had an ischaemic lesion in the cortical area supplied by the pericallosal artery affecting a diffuse area of the right dorsolateral frontal cortex. The patient performed within the normal range on judging from memory the shapes, colours and sizes of objects and animals, but had pathologically poor performance on mental rotation tasks, on Corsi block span, and on immediate memory for an imagined path around a matrix.
The contrasting patterns shown in the patients above echo dissociations that have been reported for patients with syndromes that are referred to as topographical amnesia and landmark agnosia (for review, see Aguirre & D'Esposito, 1999). Patients with topographical amnesia appear unable to generate or remember routes around towns that were previously very familiar, despite having no difficulty in recognizing familiar buildings and other landmarks. Landmark agnosia is characterized by an inability to recognize familiar buildings, but with an apparently intact ability to remember, generate and follow routes. One crucial point to note, however, is that in the literature describing such patients, the contrast is essentially between a perceptual impairment (landmark agnosia) and an impairment of a mental representation (topographical amnesia). Therefore, whether landmark agnosia might bear some similarities to the visual imagery disorders reported for patients L.H. (Farah et al., 1988) and L.E. (Wilson et al., 1999) remains an area for investigation.
The combination of visual and spatial working memory has a long history in neuropsy-chology as well as in cognitive psychology. A well-established syndrome in classic neuropsychology was known as Charcot-Wilbrand syndrome, and refers essentially to a deficit in visual imagery (for discussion, see Solms et al., 1996). Charcot & Bernard (1883; Young & Van De Wal, 1996) reported a patient, "Monsieur X", who had sudden onset of a clear deficit in forming visual images of objects, such as monuments and buildings, and of familiar people, such as close relatives. He also was unable to use an imagery mnemonic that he had used prior to the brain damage for remembering and reciting poetry. Wilbrand (1887) reported another patient, "Fraulein G", who suffered from an abrupt onset (probably a stroke) of a severe topographical amnesia, e.g. she was unable to report details of locations or routes in her native city of Hamburg, neither could she navigate around this city, in which she had lived for many years. Although the limited formal testing carried out allows only for an educated guess as to the nature of the problems faced by these patients, they do appear to comprise early examples of a dissociation between visual- and location/route-based imagery. However, the characteristic deficits in these patients were considered sufficiently similar for them to be accorded the same eponym. In an analogous fashion, visual and location/movement-based cognition have been combined under the concept of visuospatial working memory. The evidence now available allows us to consider that fractionation of the concept might be useful, theoretically and clinically.
Was this article helpful?
The comprehensive new ebook All About Alzheimers puts everything into perspective. Youll gain insight and awareness into the disease. Learn how to maintain the patients emotional health. Discover tactics you can use to deal with constant life changes. Find out how counselors can help, and when they should intervene. Learn safety precautions that can protect you, your family and your loved one. All About Alzheimers will truly empower you.