Abstract
Navigating through a visual maze relies on the strategic use of eye movements to select and identify the route. We studied navigation through novel and familiar mazes that were viewed from above and traversed by a mouse cursor.
Mazes (12x12 cells) of varying levels of complexity were randomly generated. The same maze was tested on two consecutive trials (training and test), with the test maze either: (1) identical to the training maze ("forward" condition), (2) identical, but with start and end locations reversed ("backward"), or (3) rotated 180 deg. Learning was found only in the forward condition when the identical sequence of actions was required on training and test trials. Travel times and saccade rates were reduced by 15% in test trials, regardless of maze complexity. There was no improvement in the backward condition, and there was a cost when the test maze was rotated (travel time increased 15%).
Eye movements revealed two modes of performance that almost never occurred concurrently: exploration, in which saccades were made to search for the correct path while the mouse was stationary, and guidance, in which saccades guided the mouse along the chosen path. Guidance episodes were highly stereotypical, with the eye executing sequences of saccades along the path, leading the mouse. Exploration was idiosyncratic. Some subjects explored extensively with saccades before beginning to move the mouse. Others alternated between episodes of exploration and guidance.
Examination of eye and arm movements during maze navigation show that people are able to devise and carry out complex, multi-faceted strategies that trade-off visual exploration against active motor performance. These strategies take into account available visual information, memory for the path, confidence that the correct path has been identified, the estimated cost in time of exploring with the eye or arm, and idiosyncratic tolerance for error.
Meeting abstract presented at VSS 2013