Why does a longer trip feel farther, even when it isn't?
When people navigated a virtual environment, the duration of the journey selectively changed their estimates of distance between locations, but it did not change their memory for the order of landmarks or the overall accuracy of their sketched maps [3]. In other words, time spent moving acts like a thumb on the scale for one kind of spatial judgment—how far apart two places are—while leaving the broader layout of the map intact. This suggests the brain keeps at least two kinds of spatial knowledge somewhat separate: a metric sense of distance and a structural sense of how places connect [3].
That separation matters for everyday navigation. If you take a slow, winding route to a shop, you may overestimate how far away it is without necessarily misremembering where it sits relative to other landmarks. The finding is consistent with the idea that temporal experience is not simply a byproduct of moving through space but an input that gets integrated into spatial estimates, with the integration weighted depending on context [3].
The same timing cues matter more or less depending on your viewpoint and landmarks
The influence of time on spatial knowledge is not fixed. In the same virtual navigation study, an aerial, bird's-eye perspective improved people's accuracy at judging the overall layout only when landmarks were present; from a ground-level perspective, performance was more stable when landmarks were absent [3]. Landmark knowledge, perspective, the spatiotemporal properties of the journey, and a person's sense of direction each independently contributed to how accurate their cognitive map turned out to be [3].
This is best described as a context-dependent integration account: the brain flexibly weights temporal and spatial information depending on the structure of the environment and the format in which the space is represented [3]. A practical implication is that there is no single recipe for a good mental map. Whether timing, landmarks, or viewpoint dominates depends on what is available and how the space is being learned.
A true distance map emerges gradually, and timing helps build space-like maps for sequences
In humans learning a large virtual town over four sessions, patterns of brain activity in the posterior hippocampus and entorhinal cortex came to reflect the actual straight-line distances between locations, and this only rose significantly above chance by the final session [1]. The strength of this distance coding predicted how well individuals took shortcuts in a later, unfamiliar wayfinding test, while people who failed to develop it leaned on routes they had already traveled [1]. This supports the view that a genuine metric map develops gradually with experience and is specifically useful for flexible route planning [1].
Time also appears to be a building block for maps of non-spatial sequences. When people learned statistical sequences, a model that treated the sequence as a latent space—items as nodes and transition probabilities as edges—could capture their behavior by adjusting a single parameter for temporal discounting, meaning how much recent transitions are weighted over older ones [2]. Neural activity recorded during the task reflected these estimated latent spaces mostly in the temporal lobe, including areas involved in spatial navigation, and learning typically occurred within the first 500 trials, modulated by the structure of the space and each person's discounting [2]. The authors frame this as evidence linking sequence learning to a broader, domain-general cognitive map [2].
Engineers build maps the same way: experience plus timing
The same principle shows up in navigation systems designed to mimic spatial cognition. One robotic framework models grid cells and place cells to rapidly integrate self-motion cues, uses visual cues for relocation, and weaves these together through stored experiences that express the topology of the environment [5]. A predictive map, built with a successor representation learned through reinforcement learning, then lets the robot plan sequences of experience toward a goal and replan the shortest path when the environment changes [5].
An earlier vision-and-cognitive-map system similarly represents the environment as a graph and spreads activation along its arcs to find a goal, explicitly combining spatial navigation with temporal memory [4]. These engineered examples are useful because they make the integration concrete: a map is not just a picture of space but a structure that supports ordered, timed movement through it.
Sources used in this answer
Hippocampal Involvement in Spatial Navigation and Cognitive Map Formation
In humans learning a virtual town, posterior hippocampal-entorhinal activity came to reflect true Euclidean distances by the final session, and the strength of this coding predicted shortcut-taking efficiency, while anterior hippocampal activation increased with familiarity over time.
Neurophysiological Evidence for Cognitive Map Formation during Sequence Learning
During sequence learning, neural activity in the temporal lobe reflected latent-space estimates derived from a temporal-discounting model, with learning typically occurring within the first 500 trials and modulated by the latent space and individual discounting.
Spatiotemporal properties in cognitive map formation: context-dependent effects of landmarks and perspective
Navigation duration selectively biased metric distance judgments without affecting landmark sequence memory or sketch-map accuracy, and the effects of timing were modulated by landmark availability and navigational perspective.
Navigation and planning in an unknown environment using vision and a cognitive map
A vision-based navigation and planning system represents the environment as a graph cognitive map and spreads activation along its arcs toward a goal, combining spatial navigation with temporal memory.
Robotic Navigation Based on Experiences and Predictive Map Inspired by Spatial Cognition
A robotic framework integrates self-motion cues and visual cues through stored experiences to build a predictive map, enabling fast path planning and replanning in dynamic environments.