Abstract
We present a new model of insect antennal lobe in the form of integro-differential equation with short-range inhibition. The learning in the model modifies the odor-dependent input by adding a term that is proportional to the firing rates of the network in the pre-learning steady state. We study the modification of odor-induced spatial patterns (steady states) by combination of odors (binary mixtures) and learning. We show that this type of learning applied to the inhibitory network "increases the contrast" of the network's spatial activity patterns. We identify pattern modifications that could underlie insect behavioral phenomena.
Original language | English (US) |
---|---|
Pages (from-to) | 1041-1047 |
Number of pages | 7 |
Journal | Neurocomputing |
Volume | 58-60 |
DOIs | |
State | Published - Jun 2004 |
Externally published | Yes |
Keywords
- Antennal lobe
- Integro-differential equations
- Olfaction
- Plasticity
ASJC Scopus subject areas
- Computer Science Applications
- Cognitive Neuroscience
- Artificial Intelligence