The hippocampus is required for short-term topographical memory in humans

Abstract

The hippocampus plays a crucial role within the neural systems for long-term memory, but little if any role in the short-term retention of some types of stimuli. Nonetheless, the hippocampus may be specialized for allocentric topographical processing, which impacts on short-term memory or even perception. To investigate this we developed performance-matched tests of perception (match-to-sample) and short-term memory (2 s delayed-match-to-sample) for the topography and for the nonspatial aspects of visual scenes. Four patients with focal hippocampal damage and one with more extensive damage, including right parahippocampal gyrus, were tested. All five patients showed impaired topographical memory and spared nonspatial processing in both memory and perception. Topographical perception was profoundly impaired in the patient with parahippocampal damage, mildly impaired in two of the hippocampal cases, and clearly preserved in the other two hippocampal cases (including one with dense amnesia). Our results suggest that the hippocampus supports allocentric topographical processing that is indispensable when appropriately tested after even very short delays, while the presence of the sample scene can allow successful topographical perception without it, possibly via a less flexible parahippocampal representation.

Figure 1

Stimuli used for the 4 Mountains Test. A) Each landscape is comprised of four scattered hills of varying shape and size within a semicircular range of hills. Each stimulus used in the experiment used a unique configuration of hills. Smoothed interpolated 2D noise was also added to the heightfield to produce unique, naturalistic variations between landscapes. The scenes are rendered as through a virtual camera sited at one of seven points (shown by white circles) equidistant from the centre of the heightfield. B) Typical example of a rendered stimulus image based on topography shown in A C) Topography shown in A, rendered under the eight different combinations of non-spatial parameters used in the experiment (see text). These combinations vary cloud cover, lighting (apparent elevation of sun), texture and colour of vegetation: properties that combine to create an overall impression of the distinctive prevailing conditions at a particular time. For each combination, lighting direction (apparent azimuth of sun) could come from either due “north” or “south” relative to camera locations shown in A.

Figure 2

Top: Timing and layout of test items. Perceptual tests used a concurrent match to sample task. Participants had a maximum of 30s to choose one picture from four alternatives (on the lower page of the test booklet) that matched the sample image (upper page). Memory tests used a delayed match to sample task, interposing a 2s delay (during which a blank page was shown) between sample and test images. Bottom: Examples of non-spatial and topographical items. In non-spatial tests participants had to match images based solely on the non-spatial features in the scene; cloud cover, lighting, texture and colour of vegetation. The target is shown at the bottom left of the four choices. Topographical features were varied between sample and test images. In topographical tasks subjects had to match images based solely on the topographical features; viewpoint, and non-spatial features were varied between sample and test images. The target is shown at the top left of the four choices. The spatial foil is at the top right, the configural foil at the bottom left, and the elemental foil at the bottom right (2 hills changed as in the topographical memory test), see Experimental Procedures.

Figure 3

MRI images showing the location and extent of lesions in each patient. The left column shows a coronal section through the hippocampi, the right column shows a sagittal section through the right hippocampus. Patient KC3’s scan is a T2 weighted image showing abnormal high signal in the hippocampus prior to treatment. No corresponding sagittal scan is available for KC3. The other scans are T1 weighted structural images in which cerebro-spinal fluid appears dark. Brief lesion descriptions: KC3 - bilateral hippocampal damage; VC - bilateral hippocampal damage; RH - right hippocampal damage; Jon - bilateral hippocampal damage; MH - right-sided lesion including hippocampus, medial-ventral parts of occipital and temporal cortices. See Experimental Procedures for further details of the lesions and aetiology for each patient involved in the study, and Tables 1 and 2 for further information on patient age, sex and cognitive profile.

Figure 4

Results. Top: Raw scores for each patient on each of the four tests (maximum 15), for control groups, the mean score is shown by the bar, error bars indicate the standard deviation. Patient MH has damage to medial and ventral parts temporal and occipital neocortex, including parahippocampal gyrus, as well as damage to the hippocampus; the other patients have focal damage to the hippocampus (refer to text and Figure 1 for further details). Bottom: Corresponding Z-scores derived by comparing each patient to the relevant control group. Patient Jon is compared with the young control group, the other patients are compared with the older control group. The dashed line shows 1.96 standard deviations below the mean.

Figure 5

Pie charts showing qualitative breakdown of responses to topographical tests. Each response in the topographical tests could take one of four forms: In correct responses (white segments) the selected image shows the same topography as the sample image (albeit from a different viewpoint). In spatial errors (stripes) the selected foil retains the same combination of distinctive hills arranged in the same order around the origin, but with the locations of the hills disrupted. In configural errors (grey) the selected foil retains the same combination of distinctive hills arranged in a different order around the origin. In elemental errors (black) the shape and size of one (perception) or two (memory) hills is changed. The figure shows the breakdown of responses for each patient both perception (left column) and memory (right column) tasks. Patient Jon (who is substantially younger than the other patients) is shown on the right of the figure, beneath the overall results for the young control group. The other patients are shown on the left of the figure beneath the overall results for the old control group.