Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 43 of the 43 references it cites were found in a registry or an index; 12 of the 30 sources its claims cite were checked for support.

Cognitive Maps for a Non-Euclidean Environment: Path Integration and Spatial Memory on a Sphere

Open access
Kim, Misun, Doeller, Christian F. · 2024 · Psychological Science

Abstract

Humans build mental models of the world and utilize them for various cognitive tasks. The exact form of cognitive maps is not fully understood, especially for novel and complex environments beyond the flat Euclidean environment. To address this gap, we investigated path integration —a critical process underlying cognitive mapping—and spatial-memory capacity on the spherical (non-Euclidean) and planar (Euclidean) environments in young healthy adults ( N = 20) using immersive virtual reality. We observed a strong Euclidean bias during the path-integration task on the spherical surface, even among participants who possessed knowledge of non-Euclidean geometry. Notably, despite this bias, participants demonstrated reasonable navigation ability on the sphere. This observation and simulation suggest that humans navigate nonflat surfaces by constructing locally confined Euclidean maps and flexibly combining them. This insight sheds light on potential neural mechanisms and behavioral strategies for solving complex cognitive tasks.

Overview

CoChat Summary

TL;DR

In a preregistered immersive-VR study, 20 healthy young adults navigated matched spherical and planar environments, and their path integration on the sphere was strongly better explained by planar Euclidean geometry than by spherical geometry. Despite this Euclidean bias, participants showed reasonable object-location navigation on the sphere, which the authors interpret and simulate as flexible use of multiple local planar maps.

Key findings

  • In the triangle-completion task on the sphere, the planar-geometry model strongly outperformed the spherical model: approximated BF = 3.5e54 for turn angle and BF = 2.6e23 for distance; individual R² values were higher for the planar model than the spherical model for turn angle (0.67 ± 0.19 vs. 0.50 ± 0.18) and distance (0.36 ± 0.13 vs. 0.23 ± 0.15).
  • In the object-location memory test, long-distance trials on the sphere produced a large initial direction error: the long-minus-short difference was 38.5 ± 19.4° on the sphere versus 2.5 ± 4.8° on the plane (BF = 1.1e5, interaction; t(19) = 7.7, p < .001, Cohen’s d = 1.7).
  • Final position error was larger for long-distance trials on both surfaces, with no significant interaction: long-minus-short difference was 21.3 ± 18.3° on the sphere versus 15.8 ± 10.4° on the plane (BF = 0.8, t(19) = 1.3, p = .2, Cohen’s d = 0.3); mean position error for long trials on the sphere was 50.6 ± 20.4° versus chance = 90°.
  • Simulations indicated that route planning using multiple planar maps can approximate spherical navigation: optimally arranged maps yielded position and direction errors of less than 20° for distant targets, while suboptimal arrangements reached up to 60° direction error and 40° position error, still below chance.

Methods

  • Preregistered study with 20 healthy young adults (10 male, mean age = 26.4 years, SD = 4.8 years), tested over two days in immersive VR on an omnidirectional treadmill with a head-mounted display.
  • Participants completed free exploration, object-location memory training and test, and a triangle-completion path-integration task in tightly matched spherical and planar virtual environments containing 12 animal landmarks; sphere radius was 12.5 virtual units, and neighboring landmarks were separated by 63° geodesic distance.
  • Main analyses compared linear mixed models based on planar versus spherical geometry using BIC-derived Bayes factors, plus Bayesian t tests for object-location errors and simulations of route planning based on multiple planar maps.

Limitations

  • The paper states that vestibular and proprioceptive input was limited because participants walked on a flat treadmill surface, and it suggests future work with additional 3D vestibular input, microgravity, or a 3D motion simulator.
  • The paper states that the impact of extended experience in nonflat environments remains untested and would be valuable for understanding cognitive-map adaptability.
  • Evident from the text: the sample was small (N = 20), the best-fitting spherical radius at the single-subject level should be interpreted cautiously because of few data points and reduced model discriminability, and strategy measures relied partly on self-report.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1177/09567976241279291

Cites

43 of 43 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1. Path integration and cognitive maps for a spherical and planar surface. If people rely on a path-integration mechanism built on a flat surface (a), they will make a systematic error (overturn and overshoot) on the sphere. Red lines represent the outbound journey; green lines represent the inbound journey for planar geometry; and blue lines represent the inbound journey for spherical geometry. Spatial layout on the spherical surface (b) can be encoded in a global 3D map or multiple patches of planar maps.
Fig. 1. Path integration and cognitive maps for a spherical and planar surface. If people rely on a path-integration mechanism built on a flat surface (a), they will make a systematic error (overturn and overshoot) on the sphere. Red lines represent the outbound journey; green lines represent the inbound journey for planar geometry; and blue lines represent the inbound journey for spherical geometry. Spatial layout on the spherical surface (b) can be encoded in a global 3D map or multiple patches of planar maps.CC BY · PubMed Central
Fig. 2. Virtual environments. Bird’s-eye views of the sphere and planar environments are shown in (a) and (b), respectively. Both environments featured the same moon-like texture and a dark background. Twelve animal landmarks were positioned in a regular triangular lattice with the geodesic distance between the neighboring landmarks identical in the two environments. In (c) and (d) are example views from a first-person perspective during the object-location task on the sphere and planar surfaces. Visibility was matched in the two environments thanks to the fog. In (e) and (f) are shown example views of the corridor that participants walked along during the triangle-completion task. Of note, participants performed the tasks with a VR headset, and the field of view and depth perception were of higher quality than the current snapshots.
Fig. 2. Virtual environments. Bird’s-eye views of the sphere and planar environments are shown in (a) and (b), respectively. Both environments featured the same moon-like texture and a dark background. Twelve animal landmarks were positioned in a regular triangular lattice with the geodesic distance between the neighboring landmarks identical in the two environments. In (c) and (d) are example views from a first-person perspective during the object-location task on the sphere and planar surfaces. Visibility was matched in the two environments thanks to the fog. In (e) and (f) are shown example views of the corridor that participants walked along during the triangle-completion task. Of note, participants performed the tasks with a VR headset, and the field of view and depth perception were of higher quality than the current snapshots.CC BY · PubMed Central
Fig. 3. Triangle-completion task. Following an outbound journey consisting of two legs (a; red line), participants turned and moved straight to the invisible start location (black line). Participants performed the same task on the sphere and plane. If a path-integration system is based on planar geometry, participants were expected to follow the green line in both conditions. In contrast, participants would follow an optimal route on the sphere (blue line) if the path-integration system is adjusted for spherical geometry. All participants’ inbound paths on the sphere (b) are shown for three example triangle shapes. Gray lines show individual trials; the thick black line represents the mean path of all trials. In (c) the same is shown for the plane condition. The comparison of two linear mixed models in (d) shows that actual turn angle on the sphere was better explained by the ideal turn angle on the planar model. Individual participant data are colored together, and a small jitter is applied on the horizontal axis for visualization purposes. A bar graph shows the R2 of model fit for the plane and sphere model at the individual-participant level. Error bars show standard errors. In (e) the same is shown for the inbound distance.
Fig. 3. Triangle-completion task. Following an outbound journey consisting of two legs (a; red line), participants turned and moved straight to the invisible start location (black line). Participants performed the same task on the sphere and plane. If a path-integration system is based on planar geometry, participants were expected to follow the green line in both conditions. In contrast, participants would follow an optimal route on the sphere (blue line) if the path-integration system is adjusted for spherical geometry. All participants’ inbound paths on the sphere (b) are shown for three example triangle shapes. Gray lines show individual trials; the thick black line represents the mean path of all trials. In (c) the same is shown for the plane condition. The comparison of two linear mixed models in (d) shows that actual turn angle on the sphere was better explained by the ideal turn angle on the planar model. Individual participant data are colored together, and a small jitter is applied on the horizontal axis for visualization purposes. A bar graph shows the R2 of model fit for the plane and sphere model at the individual-participant level. Error bars show standard errors. In (e) the same is shown for the inbound distance.CC BY · PubMed Central
Fig. 4. Object-location memory tasks. Two types of error (a) were measured in this task, in which participants had to walk straight to an invisible target from various start locations on a sphere (top) and a plane (bottom). Direction error (θ) refers to the angle between the optimal route (red line) and the path taken by participants (blue line). Position error (—) refers to the geodesic distance between the true target location and the recalled location (black dotted line). In (b), example optimal paths (red lines) and all participants’ trajectories (blue lines) are shown when the start locations were either at a short or long distance from the target. If participants used multiple planar maps instead of a volumetric map, path planning would be particularly difficult for long trials on the sphere. Indeed, direction errors (c) were significantly larger for long trials on the sphere, and a significant interaction effect between the distance and environment type was observed. Position errors (c) were significantly larger for long-distance trials on both the plane and the sphere, without an interaction effect. Colored dots represent each participant’s mean error; error bars represent standard errors.
Fig. 4. Object-location memory tasks. Two types of error (a) were measured in this task, in which participants had to walk straight to an invisible target from various start locations on a sphere (top) and a plane (bottom). Direction error (θ) refers to the angle between the optimal route (red line) and the path taken by participants (blue line). Position error (—) refers to the geodesic distance between the true target location and the recalled location (black dotted line). In (b), example optimal paths (red lines) and all participants’ trajectories (blue lines) are shown when the start locations were either at a short or long distance from the target. If participants used multiple planar maps instead of a volumetric map, path planning would be particularly difficult for long trials on the sphere. Indeed, direction errors (c) were significantly larger for long trials on the sphere, and a significant interaction effect between the distance and environment type was observed. Position errors (c) were significantly larger for long-distance trials on both the plane and the sphere, without an interaction effect. Colored dots represent each participant’s mean error; error bars represent standard errors.CC BY · PubMed Central
Fig. 5. Path planning on the sphere based on multiple planar maps. Local triangular maps (a) are stitched together to find a route to the target. The maps can be aligned either optimally or suboptimally on a hypothetical plane. An example path is shown for a target at position 6 and a start location between positions 1, 2, and 3 (with a distance of 134° to the target). The purple line represents a geodesic line on the sphere; a red line represents a path based on an optimal arrangement of the planar map; and a blue line represents a path based on a suboptimal arrangement. Direction and position errors for trials with varying distance between the start and target are shown in (b), with red dots representing errors based on the optimal arrangement and blue dots representing errors based on the suboptimal arrangement. We simulated the suboptimal arrangement only for distances far enough (> 120°), as it is unlikely that people would recall suboptimal routes when the target is close to the start location.
Fig. 5. Path planning on the sphere based on multiple planar maps. Local triangular maps (a) are stitched together to find a route to the target. The maps can be aligned either optimally or suboptimally on a hypothetical plane. An example path is shown for a target at position 6 and a start location between positions 1, 2, and 3 (with a distance of 134° to the target). The purple line represents a geodesic line on the sphere; a red line represents a path based on an optimal arrangement of the planar map; and a blue line represents a path based on a suboptimal arrangement. Direction and position errors for trials with varying distance between the start and target are shown in (b), with red dots representing errors based on the optimal arrangement and blue dots representing errors based on the suboptimal arrangement. We simulated the suboptimal arrangement only for distances far enough (> 120°), as it is unlikely that people would recall suboptimal routes when the target is close to the start location.CC BY · PubMed Central

Verification

Claims the paper makes

8 Valid · 1 Not valid · 8 Not checked · 53 Own results and interpretations

30 more claims

Create a free account to see all 70 claims and to verify the 4 references still open.

See all claims — free account
  • Probably not a substantial claim. 18
  • It may state two separate claims as one. 23
  • The quote only partly states this claim. 3

Create a free account to see these claims.

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Consistentt(19) = 1.1reported p = .85 · recomputed p = .285

    “t(19) = 1.1, p = .85”

    The paper states a one-tailed test

  • Consistentt(19) = 7.7reported p < .001 · recomputed p < .001

    “t(19) = 7.7, p < .001”

  • Consistentt(19) = 1.3reported p = .2 · recomputed p = .209

    “t(19) = 1.3, p = .2”

3 of 3

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Self-check results appear here once an independent check has confirmed them.

Citations

Cites

Show all 40 references

Cited by

See all 9

Metadata

Authors
Kim, M., & Doeller, C. F.
Affiliations
  • Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
  • Max Planck Institute for Human Cognitive and Brain Sciences (DE)
Venue
Psychological Science
Published
2024-10-25
Volume
35
Issue
11
Pages
1217-1230
ISSN
  • 0956-7976 (ISSN-L)
  • 1467-9280
OpenAlex
W4403761054
Cited by
9
References
40
Citations by year
2026: 7 · 2025: 2
FWCI
0.97
Type
journal article
Language
en
Open access
Licence
Locations
  • Psychological Science (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Qucosa (Saxon State and University Library Dresden) (submittedVersion, cc-by)
  • MPG.PuRe (Max Planck Society) (submittedVersion, cc-by)
  • PsyArXiv (OSF Preprints) (submittedVersion, cc-by)
Funders
  • max-planck-gesellschaft
  • european research council (ERC-CoG GEOCOG 724836)
  • kavli foundation
  • Jebsen Foundation
  • helse midt-norge
  • norges forskningsråd (223262/F50; 197467/F50)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Spatial Cognition and Navigation
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Memory and Neural Mechanisms
  • Social Sciences › Geography, Planning and Development › Geography Education and Pedagogy
Keywords
cognitive maps, path integration, spatial memory, non-Euclidean geometry, mental models, spatial navigation, virtual reality
MeSH terms
Virtual Reality, Adult, Cognition, Female, Humans, Male, Space Perception, Young Adult, Spatial Memory, Spatial Navigation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Cognitive Maps for a Non-Euclidean Environment: Path Integration and Spatial Memory on a Sphere. https://app.cochat.ai/papers/10.1177/09567976241279291
The paper
Kim, M., & Doeller, C. F. (2024). Cognitive Maps for a Non-Euclidean Environment: Path Integration and Spatial Memory on a Sphere. Psychological Science. https://doi.org/10.1177/09567976241279291

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.