Sleep facilitates spatial memory but not navigation using the Minecraft Memory and Navigation task

Abstract

Sleep facilitates hippocampal-dependent memories, supporting the acquisition and maintenance of internal representation of spatial relations within an environment. In humans, however, findings have been mixed regarding sleep's contribution to spatial memory and navigation, which may be due to task designs or outcome measurements. We developed the Minecraft Memory and Navigation (MMN) task for the purpose of disentangling how spatial memory accuracy and navigation change over time, and to study sleep's independent contributions to each. In the MMN task, participants learned the locations of objects through free exploration of an open field computerized environment. At test, they were teleported to random positions around the environment and required to navigate to the remembered location of each object. In study 1, we developed and validated four unique MMN environments with the goal of equating baseline learning and immediate test performance. A total of 86 participants were administered the training phases and immediate test. Participants' baseline performance was equivalent across all four environments, supporting the use of the MMN task. In study 2, 29 participants were trained, tested immediately, and again 12 h later after a period of sleep or wake. We found that the metric accuracy of object locations, i.e., spatial memory, was maintained over a night of sleep, while after wake, metric accuracy declined. In contrast, spatial navigation improved over both sleep and wake delays. Our findings support the role of sleep in retaining the precise spatial relationships within a cognitive map; however, they do not support a specific role of sleep in navigation.

Keywords: sleep; spatial memory; spatial navigation.

Fig. 1.

Minecraft Memory and Navigation task. (A) Bird’s eye view of all environments. Participants experienced environments in first person. (B) Pink flags marked the object locations during both free exploration trainings. (C) During testing. Objects were cued for recall with written and visually aided instructions. Hearts on the screen refer to Minecraft’s lives. All players played within a single life and these were irrelevant to the task. (D) MMN task timeline. Participants were provided task instructions and administered a practice environment. In the task, participants completed two consecutive free-exploration trainings, in which they were to find all 12 objects. After trainings, participants were administered an immediate cued recall test and then a second, delay test. Study 1 evaluated only trainings and test 1, performance. (E) Study 2 timeline. Participants completed the two consecutive free-exploration trainings and immediate test. Twelve hours later, after a night of sleep or day of wake, participants completed test 2.

Fig. 2.

Minecraft environment training and test results. (A) Training time. There were no significant differences between training 1 and training 2 in the amount of time spent in the free-exploration phases. (B) Number of objects found. No significant differences between environments were found in the number of objects found across training sessions. (C) Test time. Participants spent equivalent time across the four environments in the testing phase. (D) Distance from object location. We found no differences between immediate spatial memory performance across environments.

Fig. 3.

Minecraft sleep-dependent memory training session and test results. (A) Exploration time. Participants found and opened the 12 chests significantly faster in training 2 than in training 1. (B) Number of chests opened. No significant differences between conditions were found in the number of chests opened across trainings. Participants were required to open each chest at least once to identify and learn each unique object to proceed to the next phase of the task. Alternatively, there was a 10-min training session limit. (C) Baseline performance. Left side of graph: Participants spent the same amount of time in test 1 regardless of condition. Right side of graph: Participants placed the objects at equivalent distances in test 1, regardless of condition. (D) Distance from location. We found a significant two-way interaction between condition and test timing. Participants maintained their location accuracy after a night of sleep compared while their accuracy significantly declined after a comparable time awake. (E) Performance difference over time is plotted (the two-way interaction). In this graph, dotted bars represent training, solid bars represent test. In the Top row, within condition, lighter colors represent training 1 and darker represent training 2. In the second row, within condition, lighter colors represent test 1 while darker colors represent test 2. Asterisks denote main effect of training (Fig. 1A) and test (Fig. 1E) p’s <0.05.

Fig. 4.

Navigation analyses. An example path is illustrated in orange with the navigation analyses explained. 1) Spatial memory accuracy was measured as the Euclidian distance between the true location (B) of the object and the participants’ placement of the object at test (C). 2) Path length was the total number of steps between the start location (A) and participant’s placement of the object (C). 3) Header direction was the average angle (three example angles presented at D) in which the Minecraft character was directed at each path step between the start location (A) and the true object location (B) until the participant placed the object (C). Cumulative header direction was the summation of the error (all D example arrows added at each path step) with lower numbers representing more direct paths. 4) Initial orientation angle (F). We averaged the angle oriented for the first five steps (angle of arrows in blue) for each object trial as a proxy for an allocentric sense of direction. 5) Approach angle (G). We calculated the angle with which participants approached the object at training and compared it to the approach angle at test. Participants had a 45° view of the world at all times, thus within this range was considered the same viewpoint. 6) Search time within proximity (H) was the total number of steps taken within concentric circles of radii at 10, 20, and 30 blocks near the true object location.

Fig. 5.

Global navigation analyses. Across all navigation analyses, we found a significant difference of test session, with improvement at test 2 compared to test 1, but no effect of condition. Plotted are (A) path length, which is the average number of steps between the start location and the participant’s object placement. (B) Header direction is the average angle at each step between the start location and the true object location. (C) The cumulative error is the summation of each angle between the start and true location. This takes into account when participants have turned around and are headed in the incorrect direction, with higher scores representing greater error. (D) Participant’s search time percentage within proximity of the true location is plotted for each radii of 10, 20, and 30 blocks away from the true location at each test. Within conditions, all light colors represent test 1 and darker colors represent test 2. Asterisks denote main effect of test p’s <0.05.

Fig. 6.

Path length and spatial memory analysis at the object-by-object level. In neither condition did object memory and path length correlate (A–D). Here we show the nonsignificant relationships for both conditions between the difference in memory performance from test 1 to test 2 and the difference in path length from test 1 to test 2.

Fig. 7.

Header direction and spatial memory analysis at the object-by-object level. Header direction is the angle moving at the start of test trial with 0 indicating the best fit line from starting position to true object location. Header direction correlated positively with path length for both conditions at each test, demonstrating that participants with better orientation for the true object location took fewer steps prior to placement.