hide
Free keywords:
-
Abstract:
Purpose: Results from previous studies (e.g. Loomis et al., JEP, 1993) suggest that proprioceptive cues play a major role in human homing behaviour. We conducted triangle completion experiments in
virtual environments to measure homing performance based solely on visual cues.
Methods: Subjects were seated in the centre of a large half-cylindrical 180° projection screen and
steered smoothly through the simulated scene using mouse buttons. Experiments were conducted in two
environments: an extended volume filled with random blobs (inducing strong vection), and a
photorealistic town containing distinct landmarks. On each trial, subjects had to return to their
starting point after moving outwards along two prescribed segments (40m long, subtending a 30°..150°
horizontal angle) of an imaginary triangle. To exclude scene-matching as a homing strategy, the
simulated environment was modified to a different but similar one just before the subject started
the return movement.
Results: We found strong systematic errors in distance travelled but only small deviations in
turning angles. After some practice the variability (standard deviation) of the responses typically
dropped to roughly 10m for distance and 10 degree for turns (lower variability for town than for
blobs-scene). Omitting the scene modification before the return movement resulted in nearly perfect
performance, stressing the dominant role of piloting under natural conditions. Exchanging the mouse
interface for a more realistic bicycle interface, thus introducing proprioceptive cues for side-ways
tilt and pedal resistance, reduced the systematic error in rotation but also increased the overall
variability.
Conclusion: Path integration using optical information alone is sufficient for accurate homing.
