Mental simulation of routes during navigation involves adaptive temporal compression

Abstract

Mental simulation is a hallmark feature of human cognition, allowing features from memories to be flexibly used during prospection. While past studies demonstrate the preservation of real-world features such as size and distance during mental simulation, their temporal dynamics remains unknown. Here, we compare mental simulations to navigation of routes in a large-scale spatial environment to test the hypothesis that such simulations are temporally compressed in an adaptive manner. Our results show that simulations occurred at 2.39× the speed it took to navigate a route, increasing in compression (3.57×) for slower movement speeds. Participant self-reports of vividness and spatial coherence of simulations also correlated strongly with simulation duration, providing an important link between subjective experiences of simulated events and how spatial representations are combined during prospection. These findings suggest that simulation of spatial events involve adaptive temporal mechanisms, mediated partly by the fidelity of memories used to generate the simulation.

Keywords: Episodic memory; Hippocampus; Imagination; Prospection; Recollection; Virtual reality.

Fig. 1

Overview of task design and city composition. (A) Outline of the path passively viewed from the first-person perspective during the exposure phase with the location of the five target landmarks. (B) Street level view of the encoding phase showing non-landmark buildings and the facades of the five target landmarks. Note that the green arrow pointing to the target landmark was only displayed on screen if more than 90 seconds elapsed during the route. (C) Task order for a single trial of the simulation phase. Two second interstimulus intervals (ISI) showing a white fixation cross were inserted between each task component (PostSQ: post simulation questionnaire; PostNQ: post navigation questionnaire).

Fig. 2

Comparison of simulation and navigation performance between movement speed conditions (slow: 4 virtual m/s, medium: 6 virtual m/s, fast: 8 virtual m/s). Error bars represent 95% confidence intervals. (A) All groups showed statistically significant differences in subsequent route navigation time. (B) Simulation times in the slow and medium groups differed from the fast group, but not between the slow and medium groups. (C) The slow movement group had a statistically faster compression rate than the medium and fast groups. Note that p-values use the Bonferroni correction for multiple comparisons.

Fig 3

Regression plots showing the relationship between mean centered simulation times and the predictor variables for each movement condition. (A) Path time for the slow movement condition (b = 0.17, t(191) = 4.78, p < 0.001). (B) Simulation confidence (see Table 1) for the slow condition (b = −0.86, t(191) = −2.02, p = 0.04). (C) Path time for the medium movement condition (b = 0.19, t(207) = 4.34, p < 0.001). (D) Vividness for the medium movement condition (b = −1.12, t(207) = −2.94, p = 0.004). (E) Path time for the fast movement condition (b = 0.14, t(191) = 3.34, p = 0.001). (F) Spatial coherence (see Table 1) for the fast movement condition (b = −0.88, t(191) = −3.06, p = 0.003). Shaded blue area represents 95% confidence interval.

Figure 4

Partial correlation plots showing the relationship between excess path time and mean centered simulation time for each movement condition and all conditions combined.