Detecting age‐dependent deterioration of spatial working memory with a virtual water maze paradigm in human subjects

Background Virtual‐Morris Water Maze (V‐MWM) has been developed to assess potential hippocampal‐dependent memory decline. Previous attempts showed the V‐MWM was sensitive in detecting age‐related reference memory decline, but our pilot study indicated that elderly often could not complete the tasks within 60s. Therefore, we have investigated further the use of a spatial working memory paradigm to evaluate one‐trial learning task allowing for up to 120s per trial. Methods Three age groups were recruited from the community with a cross‐sectional design. The Young Adult (YA,n = 22), Young Elderly (YE,n = 30) and older elderly (OE,n = 19) groups defined according to the subjects’ age in years: 18‐45, 60‐69, and 70 or above, respectively. The V‐MWM was programmed using MazeWaker. Subjects navigated in a 3D circular arena, defined by uniform walls and presented on a 14‐inch screen using an arrow key. Eight distal 3D‐cues located outside the arena. A total of 8 possible locations assigned randomly to every two consecutive trials. The two trials were separated by 5s or interspersed by a comic strip for 15s. The target become visible after 120s. Working memory was indexed by more efficient navigation from trial‐1 to trial‐2 in terms of latency or path‐length, indexed by the ratio: T2/(T1+T2) and subjected to a 3×2×3 (AgexInterference typesxLocations) ANOVAs. Two 60s‐nonrevealing probe tests were conducted on trials 8th and 16th. Percent path‐length near the 8 possible target‐zones using separate 2×2×8 (Age groupsxInterference typesxTarget‐zones) ANOVAs. Results The YA group outperformed both the YE and OE groups (F(2,68) = 8.17, p = .001), but no significant group by interference interaction was revealed. However, the 15s‐interference tended to increase the latency in both YA and YE groups (F(1,68) = 9.05, p