Patients and methods

Patients were recruited from the Department of Neurology of the University Medical Centre, Utrecht. All patients had undergone neurosurgical removal of an intracranial tumour more than 1 year before this study. None of the patients received radiotherapy after the operation. A group of 24 healthy subjects of comparable age and education served as controls (see the table for a detailed description of the patients and control group).

To test object location memory, all patients studied computer displays with a frame containing 10 easy to name objects at different locations, which had to be relocated afterwards. Three task conditions were used. Firstly, to measure object location binding, all different objects were shown at different locations, which had to be reassigned to their correct positions that were premarked by black dots. Secondly, to assess positional memory, objects that were all equal were shown at different locations, and had to be relocated as accurately as possible in an empty frame (without premarked dots). Thirdly, to measure the combined process, different objects were shown at different locations that had to be relocated in an empty frame (without premarked dots).12

As control tasks, conditions were included that assessed visuospatial construction and perception (the patient had to copy a frame containing 10 different objects at different locations without a memory component) and object identity memory (10 objects were shown, which had to be recognised subsequently from a set of 20 objects, containing the ones that were shown previously). Each condition consisted of two different displays. Additionally, the subject had to relocate the objects both immediately after the presentation as well as after a 3 minute delay. Performance measures were percentage incorrect items in the object location binding and object identity memory condition, and deviation (mm) in the positional memory task, the combined process, and the visuospatial construction and perception condition. Neuropsychological evaluation on the basis of single tasks may produce spurious data, as patients often fail on tasks for reasons unrelated to the specific function of interest. Therefore, combined error scores were used based on both the immediate and the delayed performance. To measure procedural spatial memory, two standard neuropsychological spatial memory tasks that are widely used in clinical practice were included, The Corsi block tapping task to determine the spatial memory span13 and the Oxford stylus maze to assess spatial learning.14 All spatial tasks can be presumed to assess allocentric spatial processes, due to the presence of external frames of reference.2 The results of the control group were used to calculate standardised Z scores for the individual patients. Subsequently, cut off scores were determined for a performance that differed from the control mean at 0.05 (Z>1.65), 0.01 (Z>2.33), and 0.001 (Z>3.10) levels of statistical significance.15

Results

The table shows the results for the 10 patients on the experimental spatial memory and control tasks, as well as the mean scores for the control group. Five patients (P1–P5) were unimpaired on all spatial memory tasks, whereas one patient (P10) displayed impairments on all of these. Four patients were impaired on object location memory. Of these, two patients (P6, P7) showed a selective impairment on positional memory, and two other patients (P8, P9) performed worse than the controls on object location binding and the combined process. For the non-spatial memory control tasks, two patients (P2, P10) performed poorly on visuospatial construction and perception, and three patients (P8–P10) had impaired object identity memory. Analysis of the performance on the standard neuropsychological spatial memory tasks showed a lowered performance on the Oxford stylus maze in three patients (P6, P9, P10; p<0.001), whereas one patient was impaired on the Corsi block tapping task (P10; p<0.001).

For lesion localisation, it was found that the occurrence of impaired object location memory was more frequent in right hemispheric patients than in left hemispheric patients (Mann-WhitneyU=3.00, Z=1.96, p=0.03). Furthermore, object location memory dysfunction was more often found in patients with posterior (parietal or occipital) lesions than in patients with anterior (temporal or frontal) lesions (Mann-WhitneyU=3.00, Z=1.96, p=0.03).

Discussion

The purpose of the current study was to find selective neuropsychological deficits in specific components within object location memory. Our results showed a double dissociation between the performance on positional memory (the ability to remember precise, metric information) on the one hand, and the object location binding and the combined task conditions on the other hand. These findings provide further evidence that object location memory consists of multiple cognitive mechanisms. It seems that one process exists (positional memory) that is responsible for the encoding and retrieval of euclidean positional information, whereas a separate process is involved in the assignment of item information to (relative or precise) spatial locations (as measured with the object location binding and combined process conditions).

No selective impairments in either the object location binding or the combined process were seen in this study (in line with previous findings4), which suggests that these conditions may measure the same underlying process. Moreover, the patients that have slight problems in memory for the identity of objects (P8, P9) are also impaired at these two spatial memory conditions. However, these general memory deficits cannot fully explain the selective spatial memory findings, as it is especially spatial mnemonic processing that is tested, rather than memory for object identity (patients do not have to recall the objects). Notably, patients with deficits in object location memory did not always display impairments on maze learning or block span memory. This corroborates the distinction between memory for spatiotemporal relations (procedural learning of routes or sequences) and declarative spatial memory (remembering the locations of objects).3

Interestingly, the results show that all three right hemispheric patients are impaired on the spatial memory tasks, but only one left hemispheric patient (out of seven). This is in line with the idea that spatial deficits are commonly (but not solely) the result of right hemispheric damage.16 Furthermore, all patients with posterior lesions performed poorly, regardless of side affected. This clearly shows that the posterior parietal cortices are important in spatial memory processing.9 11 14 One patient (P6) had a lesion in the (medial) right temporal lobe (with complete removal of the right hippocampus). This brain area is also involved in object location memory.8

In summary, the results of the present study show a double dissociation between positional memory and the assignment of object identities to (relative or precise) locations. Although strong conclusions about the functional neuroanatomical circuits of spatial memory cannot be made on the basis of single cases, it seems that in this group of patients spatial memory dysfunction is especially present in those with either right hemispheric damage or posterior lesions.