Elsevier

Evolution and Human Behavior

Volume 29, Issue 5, September 2008, Pages 299-304
Evolution and Human Behavior

Original Articles
Evolved navigation theory and the environmental vertical illusion

https://doi.org/10.1016/j.evolhumbehav.2008.03.001Get rights and content

Abstract

This study outlines a previously unknown, large illusory component to one of the most common psychological experiences. Evolved navigation theory (ENT) suggests that perceptual and navigational mechanisms reflect navigational costs over evolution. Vertical surfaces pose a distinct cost of falling not present in horizontal navigation. However, horizontal surfaces sometimes form retinally vertical images and researchers often assume that retinal image determines distance perception. We tested ENT-derived predictions suggesting that observers would overestimate surface lengths based on environmental, not retinal, verticality. Participants drastically overestimated environmentally vertical surfaces only and did so at a magnitude related to surface length. These results replicate across multiple settings and methods and are supported by previous studies. Although researchers often assume that selection pushes perceptual mechanisms toward objective accuracy, this study suggests that genetic fitness can sometimes benefit from systematic illusions.

Introduction

It is often assumed that perceptual mechanisms function identically across most environmental features, especially basic visual perception of line or surface orientation, length, and location. However, evolved navigation theory (ENT) is a research approach that predicts specializations in the perception or navigation of environmental features that are reliably associated with navigational costs over evolutionary time (Jackson, 2005, Jackson & Cormack, 2007a). ENT is primarily an application of signal detection theory to navigation costs over evolutionary time as a means to predict the type and magnitude of specific navigational adaptations. A more general approach of looking at how decision consequences pose selection across domains has been summarized by Haselton and Buss (2000) as error management theory.

ENT focuses on understanding the selective forces particular to navigation as a means to predict unknown locomotor, visual, and other navigational adaptations. For example, vertical and inclined surfaces present a navigation risk of falling that is unmatched in horizontal surfaces. Falls of a few meters produce serious injuries that would be exceedingly rare when navigating a few horizontal meters. Vertical navigation likely posed a strong selective factor in the evolution of terrestrial organisms, certainly those with arboreal ancestors, such as humans. Thus, an implication under ENT is that some perceptual or navigational mechanisms might be specialized to lessen the risks of falling.

One possible method of establishing navigational route preferences is by exaggerating distance perceived from costlier routes because organisms prefer nearer navigational goals (Somervill & Somervill, 1977). A process that exaggerated perceived vertical surface length could thereby decrease vertical navigation in the presence of less costly alternatives. This would result in vertical surfaces perceived as longer than equidistant horizontal surfaces, which would decrease vertical navigation frequency and its associated falling costs. This simple mechanism could dynamically weight navigational decisions in real time across all surfaces by exaggerating perceived length based upon initial surface length and orientation. Such a mechanism captures both navigation difficulty and a primary predictor of falling costs: distance from the ground. This would also flexibly weight navigational decisions by cost without outright prevention of vertical navigation, which is beneficial for organisms such as humans, who have derived important benefits from vertical navigation over evolutionary time.

A previous investigation of ENT-derived predictions demonstrated that participants unknowingly overestimated vertical surface length at a magnitude corresponding to the potential falling risk (Jackson & Cormack, 2007a). However, many surfaces cast images that are oriented vertically with respect to the observer's head or eyes (i.e., egocentrically vertical) — including surfaces with trivial falling risk. For example, looking down at the distance from one's feet to a distant point ahead on horizontal ground casts an egocentrically vertical image on the retinae, yet poses negligible falling risk. Although researchers commonly assume that the retinal image largely determines perceived distance, it obviously does not predict falling costs accurately enough to decrease them.

The feature that does predict falling cost is environmental, or exocentric, verticality: the extent to which a surface parallels the direction of primary gravitational force. In order for the vertical overestimation derived from ENT to result in appropriate falling cost avoidance, it should exaggerate exocentrically vertical surface length, with little regard for egocentric verticality.

We addressed this question in the current study by comparing real-world distance estimates across effectively equal egocentric images that nonetheless had different exocentric orientations corresponding to very different falling costs. We predicted from ENT that participants would overestimate distance only from exocentrically vertical surfaces because such surfaces posed distinct falling costs over evolutionary time.

We also varied stimulus length in the current study, which we predicted could affect the hypothesized overestimation in one of two ways. First, participants might overestimate by a constant percentage of the stimulus length (i.e., Weber's law) because such a simple algorithm might provide sufficient falling cost avoidance. Alternatively, participants might overestimate by an ever-greater magnitude as stimulus length increases because longer vertical surfaces at these distances pose both greater likelihood and overall cost of falling.

Section snippets

Methods

Thirty-eight introductory psychology participants reporting normal (20:20) or corrected-to-normal vision estimated distances in an outdoor testing site. Fig. 1 schematically illustrates participants' distance estimates.

Results

Participants overestimated exocentrically vertical distances only and did so to a large, increasing degree as stimulus length increased (see Fig. 2). Distances that appear vertical on the retinae were only overestimated if they were vertical in the environment, with longer distances overestimated by greater magnitudes. Participants underestimated exocentrically horizontal distances to a slight degree that paralleled accuracy.

All six estimates significantly differ from both accuracy [the least

Discussion

As predicted from ENT, participants overestimated exocentrically vertical distances and did not overestimate exocentrically horizontal distances — despite the fact that actual distance was equal across both orientations and that egocentric orientation and image size were highly similar. We title this the environmental vertical illusion.

We found the environmental vertical illusion at a very large magnitude for a previously unknown psychological process that likely occurs constantly throughout

Acknowledgments

We thank Mike Domjan for research materials; Jenée James Jackson, Tasha Beretvas, Clarke Burnham, Randy Diehl, and Joe Horn for suggestions; Zach Keeton and Germaine Ho for collecting data; and reviewers for comments. We also acknowledge direction on citations by editor Martie Haselton.

References (26)

  • I.P. Howard et al.

    Human spatial orientation

    (1966)
  • R.E. Jackson

    Falling towards a theory of the vertical–horizontal illusion

  • R.E. Jackson et al.

    Evolved navigation theory and the descent illusion

    Perception and Psychophysics

    (2007)
  • Cited by (34)

    • When humans can fly: Imprecise vertical encoding in human 3D spatial navigation

      2022, Behavioural Brain Research
      Citation Excerpt :

      Such anisotropy in spatial memories is consistent with the pattern of the vertical-horizontal illusions (i.e., overestimation of the length of a vertical line compared with the same length of a horizontal line) in visual perception [25]. Such anisotropy in spatial memories could be due to humans’ larger horizontal visual field than their vertical visual field [23], higher risks of falling in the vertical dimension than horizontal dimension [16], or organizing memories of a multilevel building into regions according to horizontal levels so that the distance between two locations across regions (i.e., levels) appears to be longer than the same distance between two locations within a region (level) [4,13,24]. However, the above anisotropy in 3D spatial memories (i.e., human participants overestimated vertical distance but underestimated horizontal distance) only indicates differences in encoding vertical and horizontal information; it does not necessarily indicate a horizontal advantage (i.e., more precise coding horizontally) in 3D spatial memories.

    • Evolved navigation theory and the plateau illusion

      2013, Cognition
      Citation Excerpt :

      The away estimates were shorter than both the actual stimulus and observers’ perceptions of a surface without falling risks (i.e. ground). Underestimation in order to increase safe navigation extends directly from the ENT concept of overestimation in order to decrease risky navigation outlined in previous literature (Jackson, 2005, 2009, 2013; Jackson & Cormack, 2006, 2007, 2008, 2010; Jackson & Willey, 2011). Although the visual setting of Experiment 1 did not generate a clear underestimation, we observed the expected underestimation across all expected distances in Experiment 2, plausibly generated by the presence of clear navigation risks.

    • Evolved navigation theory and horizontal visual illusions

      2011, Cognition
      Citation Excerpt :

      Current participants did not overestimate surfaces without falling risk and ranged from 9% overestimation at low falling risks to 47% at high falling risks. This is nearly the exact range found for the corresponding situations in vertical surfaces (Jackson & Cormack, 2008). Environments (and individuals) exhibiting even greater falling risk or avoidance are the exact situations in which overestimation magnitude exceeds the range found here (Jackson, 2009; Jackson & Cormack, 2007).

    View all citing articles on Scopus
    View full text