Abstract submitted for presentation at the
1998 meeting of the Society for Neuroscience
What role do visual and cognitive factors play
in sequence learning and wayfinding? To answer such questions, we had
human subjects navigate multiple-T junction mazes in a computer rendered,
virtual 3D environment. Over repeated encounters with the same maze
layout, subjects' responses at choice points became faster and less
error-prone. To explore the representations being learned in this task, we
created mazes that vary along three dimensions:
To perform this
task successfully, subjects must overcome massive associative interference
generated when only two different responses (left or right turns) are
mappped onto many, identical choice points. Distinctive visual landmarks
help to differentiate choice points, thereby reducing associative
interference. Optic flow may play a similar role by allowing subjects to
construct a cognitive map that can be used to distinguish otherwise
identical choice points.
In addition to
behavioral studies on normal subjects, we report electrophysiological
results from epileptic patients who had subdural electrodes implanted as
part of a pre-surgical evaluation. ERPs, time locked to responses at maze
decision points, were obtained at multiple electrode sites across the
surface of the medial temporal lobe and near the hippocampus.
For mazes exceeding the
span of immediate memory, learning is significantly facilitated by either
optic flow or distinctive landmarks.