Abstract
When the head is rotated about an axis tilted from the vertical, in total darkness, compensatory eye movements, with a velocity approximately equal to that of the head, are generated and persist indefinitely. This study implements a neural network model which could be trained to produce this velocity estimation and to explicitly but naturally include the tilt angle normalization. A feedforward network was used, with connection weights determined by back-propagation. The input to the network is a pattern of otolith cell excitations as in the mathematical model. Its output is a single activation representing a velocity estimate.
| Original language | English |
|---|---|
| Pages (from-to) | 248 |
| Number of pages | 1 |
| Journal | Neural Networks |
| Volume | 1 |
| Issue number | 1 SUPPL |
| DOIs | |
| State | Published - 1988 |
| Externally published | Yes |
| Event | International Neural Network Society 1988 First Annual Meeting - Boston, MA, USA Duration: 6 Sep 1988 → 10 Sep 1988 |