CoChat

Not seeing the forest for the trees: combination of path integration and landmark cues in human virtual navigation

Open access

Published version of a preprint

The DOI matches its Crossref record; 56 of the 56 references it cites were found in a registry or an index.

Abstract

Introduction: In order to successfully move from place to place, our brain often combines sensory inputs from various sources by dynamically weighting spatial cues according to their reliability and relevance for a given task. Two of the most important cues in navigation are the spatial arrangement of landmarks in the environment, and the continuous path integration of travelled distances and changes in direction. Several studies have shown that Bayesian integration of cues provides a good explanation for navigation in environments dominated by small numbers of easily identifiable landmarks. However, it remains largely unclear how cues are combined in more complex environments. Methods: To investigate how humans process and combine landmarks and path integration in complex environments, we conducted a series of triangle completion experiments in virtual reality, in which we varied the number of landmarks from an open steppe to a dense forest, thus going beyond the spatially simple environments that have been studied in the past. We analysed spatial behaviour at both the population and individual level with linear regression models and developed a computational model, based on maximum likelihood estimation (MLE), to infer the underlying combination of cues. Results: Overall homing performance was optimal in an environment containing three landmarks arranged around the goal location. With more than three landmarks, individual differences between participants in the use of cues are striking. For some, the addition of landmarks does not worsen their performance, whereas for others it seems to impair their use of landmark information. Discussion: It appears that navigation success in complex environments depends on the ability to identify the correct clearing around the goal location, suggesting that some participants may not be able to see the forest for the trees.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3389/fnbeh.2024.1399716

Cites

56 of 56 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Triangle completion task and environmental complexities across conditions. (A) Participants are guided away from a goal location along two sides of a triangle. The goal location is indicated in our VR environment by a campfire, and waypoints by piles of firewood, that participants are asked to collect because their campfire went out. Waypoint locations are not changed and participants in all repetitions and conditions walk the same outbound path and are asked to return to their goal location, the campfire, at the end (example trajectory). At trial start the goal location is approached from coordinates [0, −10] (red arrow). Environmental complexity is varied by altering the number of available objects between zero, one, two, three, ten, and 99. The conditions with higher numbers of objects contain also the objects from conditions with a lower number of objects. Scale is given in virtual meters (vm), that correspond to actual meters in the VR. (B–D) Show screenshots of the first-person view from the VR for conditions with zero, three, and 99 objects, respectively, at the second waypoint in the direction of the goal location. UI elements shown in the screenshots from top to bottom are: (1) a red timeout bar, indicating the time left for the trial; (2) an “I'm Home!” button, to log in the final position; (3) the written instruction “Press ‘space' to show/hide mouse cursor” below; (4) and a purple bar indicating the current trial within the session.
Figure 1 Triangle completion task and environmental complexities across conditions. (A) Participants are guided away from a goal location along two sides of a triangle. The goal location is indicated in our VR environment by a campfire, and waypoints by piles of firewood, that participants are asked to collect because their campfire went out. Waypoint locations are not changed and participants in all repetitions and conditions walk the same outbound path and are asked to return to their goal location, the campfire, at the end (example trajectory). At trial start the goal location is approached from coordinates [0, −10] (red arrow). Environmental complexity is varied by altering the number of available objects between zero, one, two, three, ten, and 99. The conditions with higher numbers of objects contain also the objects from conditions with a lower number of objects. Scale is given in virtual meters (vm), that correspond to actual meters in the VR. (B–D) Show screenshots of the first-person view from the VR for conditions with zero, three, and 99 objects, respectively, at the second waypoint in the direction of the goal location. UI elements shown in the screenshots from top to bottom are: (1) a red timeout bar, indicating the time left for the trial; (2) an “I'm Home!” button, to log in the final position; (3) the written instruction “Press ‘space' to show/hide mouse cursor” below; (4) and a purple bar indicating the current trial within the session.CC BY · PubMed Central
Figure 2 Homing performance of two example participants depends on number of objects. For both participants performance increases with up to three objects. For one participant (A), performance decreases with ten or 99 objects, and for another it stays high (B). Trajectory endpoints of single trials are depicted by dots. Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Outbound and homing trajectories are indicated by light grey dotted lines. Marker size does not match actual tree dimensions. Cutout does not show all objects, please see Figure 1 for a more zoomed-out view of the environment.
Figure 2 Homing performance of two example participants depends on number of objects. For both participants performance increases with up to three objects. For one participant (A), performance decreases with ten or 99 objects, and for another it stays high (B). Trajectory endpoints of single trials are depicted by dots. Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Outbound and homing trajectories are indicated by light grey dotted lines. Marker size does not match actual tree dimensions. Cutout does not show all objects, please see Figure 1 for a more zoomed-out view of the environment.CC BY · PubMed Central
Figure 3 Homing performance at population level increases in environments with up to three objects, and then decreases up to 99 objects. Trajectory endpoints of single trials are depicted by dots. Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Marker size does not match actual tree dimensions. Cutout does not show all objects, please see Figure 1 for a more zoomed-out view of the environment.
Figure 3 Homing performance at population level increases in environments with up to three objects, and then decreases up to 99 objects. Trajectory endpoints of single trials are depicted by dots. Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Marker size does not match actual tree dimensions. Cutout does not show all objects, please see Figure 1 for a more zoomed-out view of the environment.CC BY · PubMed Central
Figure 4 Performance accuracy (top) and precision (bottom) on population level are highest with three objects. Boxplots are based on median position error in virtual metres (vm) (top) or standard deviation (SD) (bottom) of N = 23 single participants with n = 24 repetitions within respective conditions (see dots). Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Boxplot notches indicate 95% confidence intervals around the median. Green and purple horizontal bars at the bottom indicate low and high clutter effects (see Section 3.3 for details).
Figure 4 Performance accuracy (top) and precision (bottom) on population level are highest with three objects. Boxplots are based on median position error in virtual metres (vm) (top) or standard deviation (SD) (bottom) of N = 23 single participants with n = 24 repetitions within respective conditions (see dots). Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Boxplot notches indicate 95% confidence intervals around the median. Green and purple horizontal bars at the bottom indicate low and high clutter effects (see Section 3.3 for details).CC BY · PubMed Central
Figure 5 Quadratic fits on the log-transformed position error (top) or log-transformed position error standard deviation (SD) (bottom) with ordinal condition show a local minimum (red vertical line) close to the three object condition. Fits include data from N = 23 participants with n = 24 repetitions per condition. Ordinal conditions from zero to five correspond to conditions with zero, one, two, three, ten, and 99 objects respectively. Quadratic regressions fit the data significantly better than a linear regression and the local minimum at three objects is validated in a linear regression model with forward difference coding (see Table 1). Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Green and purple horizontal bars at the bottom indicate low and high clutter effects (see Section 3.3 for details).
Figure 5 Quadratic fits on the log-transformed position error (top) or log-transformed position error standard deviation (SD) (bottom) with ordinal condition show a local minimum (red vertical line) close to the three object condition. Fits include data from N = 23 participants with n = 24 repetitions per condition. Ordinal conditions from zero to five correspond to conditions with zero, one, two, three, ten, and 99 objects respectively. Quadratic regressions fit the data significantly better than a linear regression and the local minimum at three objects is validated in a linear regression model with forward difference coding (see Table 1). Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Green and purple horizontal bars at the bottom indicate low and high clutter effects (see Section 3.3 for details).CC BY · PubMed Central
Figure 6 Performance improvement in accuracy and precision of position error from zero to three objects and performance impairment from three to 99 objects. Green and purple curves indicate kernel-density estimations fitted to the difference in median position error (left) or position error standard deviation (SD) (right) between zero and three objects (green), or three and 99 objects (purple), respectively, of N = 23 single participants with n = 24 repetitions within each condition (see dots). Boxplot notches indicate 95% confidence intervals around the median. All but one participant performed better with three than with zero objects. 11 participants performed significantly worse, and 3 significantly better with 99 objects, while 10 others remained equally good (see Section 3.4 for details). Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right).
Figure 6 Performance improvement in accuracy and precision of position error from zero to three objects and performance impairment from three to 99 objects. Green and purple curves indicate kernel-density estimations fitted to the difference in median position error (left) or position error standard deviation (SD) (right) between zero and three objects (green), or three and 99 objects (purple), respectively, of N = 23 single participants with n = 24 repetitions within each condition (see dots). Boxplot notches indicate 95% confidence intervals around the median. All but one participant performed better with three than with zero objects. 11 participants performed significantly worse, and 3 significantly better with 99 objects, while 10 others remained equally good (see Section 3.4 for details). Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right).CC BY · PubMed Central
Figure 7 Position error precision and accuracy are significantly correlated in all conditions. There are large inter-individual differences with example participants (A–F) respectively performing accurately and imprecisely (A, E), inaccurately and precisely (B, C), inaccurately and imprecisely (D), or accurately and precisely (F). Data points indicate median position error and its standard deviation (SD) for each of N = 23 participants with n = 24 repetitions per condition. Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Grey areas indicate 95% confidence intervals. Marker size does not match actual tree dimensions.
Figure 7 Position error precision and accuracy are significantly correlated in all conditions. There are large inter-individual differences with example participants (A–F) respectively performing accurately and imprecisely (A, E), inaccurately and precisely (B, C), inaccurately and imprecisely (D), or accurately and precisely (F). Data points indicate median position error and its standard deviation (SD) for each of N = 23 participants with n = 24 repetitions per condition. Colours marking individual participants are matched across all figures, based on a given participant's median position error in the 99-object condition (refer to Figure 4 top right). Grey areas indicate 95% confidence intervals. Marker size does not match actual tree dimensions.CC BY · PubMed Central
Figure 8 Overview of MLE modelling results. Panels show participant goal estimates (black dots) and landmark positions (green triangles) for different experimental conditions, overlayed on top of maximum likelihood model predictions for the respective condition. Heatmap colour signifies the predicted likelihood of targeting a given area, with darker colours indicating greater likelihood. Marker size does not match actual tree dimensions. (A) Introduces the ring-shaped distance estimate model, based on fitting a normal distribution to the distance component of the goal estimates. Heatmap shows the overall best model fit (maximum likelihood estimate) for this model type for the pooled data of all N = 23 participants for the zero object condition. (B) Introduces the fan-shaped direction estimate model, based on fitting a von Mises (circular normal) distribution to the directional component of the goal estimates. Heatmap shows the overall best model fit (maximum likelihood estimate) for this model type for the pooled data of all participants for the zero object condition. (C) shows the prediction resulting from an integration of both model components (ring×fan) for the same data. (D–F) Show the resulting prediction from an integration model combining a full PI model with an additional ring model for each object fitted at 10 vm distance (ring×fan×∏ring). (G, H) Show the resulting prediction from a model combining a full PI model with an alternation of ring models for each object fitted at 10 vm distance (ring×fan×∑ring).
Figure 8 Overview of MLE modelling results. Panels show participant goal estimates (black dots) and landmark positions (green triangles) for different experimental conditions, overlayed on top of maximum likelihood model predictions for the respective condition. Heatmap colour signifies the predicted likelihood of targeting a given area, with darker colours indicating greater likelihood. Marker size does not match actual tree dimensions. (A) Introduces the ring-shaped distance estimate model, based on fitting a normal distribution to the distance component of the goal estimates. Heatmap shows the overall best model fit (maximum likelihood estimate) for this model type for the pooled data of all N = 23 participants for the zero object condition. (B) Introduces the fan-shaped direction estimate model, based on fitting a von Mises (circular normal) distribution to the directional component of the goal estimates. Heatmap shows the overall best model fit (maximum likelihood estimate) for this model type for the pooled data of all participants for the zero object condition. (C) shows the prediction resulting from an integration of both model components (ring×fan) for the same data. (D–F) Show the resulting prediction from an integration model combining a full PI model with an additional ring model for each object fitted at 10 vm distance (ring×fan×∏ring). (G, H) Show the resulting prediction from a model combining a full PI model with an alternation of ring models for each object fitted at 10 vm distance (ring×fan×∑ring).CC BY · PubMed Central
Figure 9 Maximum likelihood parameter estimates for the path integration model. Top row shows maximum likelihood estimates with bootstrapped 95% confidence intervals for the parameters of the distance models fitted on the pooled data of all N = 23 participants (μ gives the average expected distance, σ the standard deviation of expected distances), bottom row shows parameters for the direction models (μ gives average expected direction, κ is the concentration parameter of the underlying von Mises distribution and has been transformed to circular standard deviation for easier readability). See Equations 1, 2 for a full description of the model functions. Fitted parameters are shown for each experimental condition separately, with overlapping 95% confidence intervals indicating that parameter estimates are similar between conditions. Dashed red lines indicate geometric ground truth where applicable.
Figure 9 Maximum likelihood parameter estimates for the path integration model. Top row shows maximum likelihood estimates with bootstrapped 95% confidence intervals for the parameters of the distance models fitted on the pooled data of all N = 23 participants (μ gives the average expected distance, σ the standard deviation of expected distances), bottom row shows parameters for the direction models (μ gives average expected direction, κ is the concentration parameter of the underlying von Mises distribution and has been transformed to circular standard deviation for easier readability). See Equations 1, 2 for a full description of the model functions. Fitted parameters are shown for each experimental condition separately, with overlapping 95% confidence intervals indicating that parameter estimates are similar between conditions. Dashed red lines indicate geometric ground truth where applicable.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

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

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 2

  1. Müller M. (2023). Finding back home: from vector models to virtual forests. PhD thesis, Bielefeld University.
  2. Van Rossum G. Drake F. L. (2009). Python 3 Reference Manual. Scotts Valley, CA: CreateSpace.

Cited by

See all 9

Metadata

Authors
Scherer, J., Müller, M. M., Unterbrink, P., Meier, S., Egelhaaf, M., Bertrand, O. J. N., & Boeddeker, N.
Affiliations
Bielefeld University (DE)
Corresponding author
Scherer, Jonas
Venue
Frontiers in Behavioral Neuroscience
Published
2024-05-21
Volume
18
Pages
1399716-1399716
ISSN
1662-5153 (ISSN-L)
OpenAlex
W4398174155
Cited by
8
References
56
Citations by year
2025: 6 · 2024: 2
FWCI
0.86
Type
journal article
Language
en
Open access
Locations
  • Frontiers in Behavioral Neuroscience (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • Publikationen an der Universität Bielefeld (Universität Bielefeld) (submittedVersion, cc-by)
  • Publikationen an der Universität Bielefeld (Universität Bielefeld) (submittedVersion)
Funders
Deutsche Forschungsgemeinschaft (460373158)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Spatial Cognition and Navigation
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Memory and Neural Mechanisms
  • Physical Sciences › Computer Science › Human-Computer Interaction › Virtual Reality Applications and Impacts
Keywords
path integration, landmark cues, human navigation, cue combination, Bayesian integration, homing, virtual reality navigation, individual differences, complex environments, maximum likelihood estimation

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.

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

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.