Abstract
Synfire chains are feed-forward networks with random connections between layers. Theory and experiments suggest that activity in such a structure propagates from layer to layer as a synchronous volley of spikes. Activity in synfire chains may represent mental items or percepts categorically and can code for the evolution of a process in time. Synfire chains produce precise spatiotemporal firing patterns composed of a sequence of synchronized firing in a series of small groups of neurons. They can also learn to recognize such spatiotemporal patterns. This can be used to dynamically bind elementary mental items into complex representations. Using current techniques, echoes of such precise spatiotemporal patterns can be (and have been) detected.
Original language | English |
---|---|
Title of host publication | Encyclopedia of Neuroscience |
Publisher | Elsevier Ltd. |
Pages | 829-832 |
Number of pages | 4 |
ISBN (Print) | 9780080450469 |
DOIs | |
State | Published - 2009 |
Externally published | Yes |
Keywords
- Binding of features
- Compositionality
- Cortical activity
- Neural network
- Synchronicity