Sources of path integration error in young and aging humans

Matthias Stangl, Ingmar Kanitscheider, Martin Riemer, Ila Fiete, Thomas WolbersNature Communications 2020 · DOI

Abstract

Path integration plays a vital role in navigation: it enables the continuous tracking of one's position in space by integrating self-motion cues. Path integration abilities vary widely across individuals, and tend to deteriorate in old age. The specific causes of path integration errors, however, remain poorly characterized. Here, we combine tests of path integration performance in participants of different ages with an analysis based on the Langevin equation for diffusive dynamics, which allows us to decompose errors into distinct causes that can corrupt path integration computations. We show that, across age groups, the dominant error source is unbiased noise that accumulates with travel distance not elapsed time, suggesting that the noise originates in the velocity input rather than within the integrator. Age-related declines are primarily traced to a growth in this noise. These findings shed light on the contributors to path integration error and the mechanisms underlying age-related navigational deficits.

Introduction

Path integration—the ability to track one's position and orientation by integrating self-motion cues like vestibular signals, proprioception, and optic flow—is fundamental to human navigation. While it is well-documented that this "internal GPS" varies among individuals and declines with age, the specific computational mechanisms of failure have remained elusive. Is error caused by a "leaky" memory, systematic biases in speed estimation, or simply stochastic noise?

This study, published in Nature Communications, addresses these questions by applying a rigorous mathematical framework to human navigation data. By combining immersive virtual reality (VR) with stochastic modeling based on the Langevin equation, researchers decomposed path integration errors into specific sources. Their findings reveal that navigation errors are primarily driven by unbiased noise in velocity inputs that scales with distance, and that this noise is the primary driver of age-related navigational decline.

Methods

The researchers developed a novel experimental and analytical pipeline involving 62 healthy participants (30 young adults, mean age 22; 26 older adults, mean age 69).

Immersive VR Task

Participants wore head-mounted displays and walked along 10 different curved, sinuous paths in a large open space. To ensure consistent paths while allowing for internal tracking, participants were guided by a baton. At four stopping points per path, they were asked to estimate the distance and direction back to their starting point.

Experimental setup: A participant wearing a VR headset is guided along a path using a baton.

The experimental sessions were organized into blocks that included standardizing paths and breaks to maintain performance consistency.

Structure of an experimental block.

Langevin Modeling and Error Decomposition

Rather than measuring aggregate error, the authors modeled the navigator's internal state using a Langevin equation for diffusive dynamics. This approach treated path integration as a continuous-time stochastic process, allowing for the isolation of four potential error sources:

  1. Velocity Gain Bias: Systematic over- or underestimation of walking speed.
  2. Memory Leak: The decay of the integrated position estimate over time.
  3. Additive Bias: A constant directional "drift."
  4. Accumulating Noise: Stochastic fluctuations that grow as the participant moves.

Parameters were estimated for each participant using an Extended Kalman Filter (EKF) and maximum likelihood estimation, with model variants compared via the Bayesian Information Criterion (BIC).

Key Results

The study yielded several critical insights into the nature of human navigation errors:

  • Distance-Dependent Noise: A primary discovery was that noise accumulates with travel distance rather than elapsed time. This suggests that the noise originates in the velocity inputs (sensory/motor signals) rather than within the integrator itself (e.g., neuronal firing fluctuations).
  • Integrator Stability: The human path integrator proved remarkably stable. Memory leak and additive bias contributed less than 3% to total error, contradicting "leaky integrator" theories.
  • Dominant Error Sources: Accumulating unbiased noise accounted for 50–55% of the total squared error, while velocity gain bias accounted for 25–26%.
  • Aging and Noise: Older adults exhibited significantly higher absolute path integration errors. The model traced this decline specifically to a growth in the accumulating noise parameter ($p = 0.012$), rather than changes in velocity bias or memory leak.

Discussion

These findings suggest that the aging brain does not necessarily have a "broken" integrator but rather receives "fuzzier" velocity data. While the study utilized a "full-cue" environment, future research is needed to determine which specific sensory channel (vestibular, proprioceptive, or optic flow) is the primary source of this increased noise.

Beyond basic science, this modeling approach holds clinical promise. Navigational deficits are often among the first clinical symptoms of Alzheimer’s disease. Applying this Langevin-based decomposition could provide a powerful diagnostic tool for identifying early-stage cognitive impairment by pinpointing specific computational failures in spatial processing.

Summary updated

Abstract

Path integration plays a vital role in navigation: it enables the continuous tracking of one's position in space by integrating self-motion cues. Path integration abilities vary widely across individuals, and tend to deteriorate in old age. The specific causes of path integration errors, however, remain poorly characterized. Here, we combine tests of path integration performance in participants of different ages with an analysis based on the Langevin equation for diffusive dynamics, which allows us to decompose errors into distinct causes that can corrupt path integration computations. We show that, across age groups, the dominant error source is unbiased noise that accumulates with travel distance not elapsed time, suggesting that the noise originates in the velocity input rather than within the integrator. Age-related declines are primarily traced to a growth in this noise. These findings shed light on the contributors to path integration error and the mechanisms underlying age-related navigational deficits.
BibTeX
@inproceedings{stangl2020sources,
  author = {Matthias Stangl and Ingmar Kanitscheider and Martin Riemer and Ila Fiete and Thomas Wolbers},
  title = {{Sources of path integration error in young and aging humans}},
  booktitle = {Nature Communications},
  year = {2020},
  doi = {10.1038/s41467-020-15805-9},
}
Figure