Embedding of synchronous spike activity in cortical networks

The question whether information in the brain is coded in precise temporal patterns of action potentials or in time averaged firing rates still remains a matter of debate. The wealth of evidence for temporal interrelations in brain activity exhibits precisely timed spikes to be part of an important information concept, justifying a closer inspection of the underlying network mechanisms. As one such model, synfire chains are special feed-forward subnetworks that allow for stable modes of propagation of synchronous pulses of neuronal spike activity. These chains were previously investigated in isolation, with only a crudely simplified substrate of incoming background activity, with no feedback to and no spatial embedding in any kind of surrounding network. As a first result, this study introduces and characterizes diluted synfire chains without the statistically unrealistic original requirement of all-to-all inter chain connectivity. It is shown that the dynamics remain untouched, provided one applies an almost linear conversion rule with well controlled errors. In a second approach, it is demonstrated that the embedding of a chain within an anatomically inspired cortical network architecture requires that the chain pools are strongly localized, the pools themselves being close to each other. Taken together, the synfire chain is trapped in a cortical volume approximately corresponding to a cortical column. The same approach ensures that the existence of synfire chains is compatible with anatomical data. An investigation of synfire dynamics in such anatomically inspired networks reveals that that the pulse of synchronous spikes traveling along the chain strongly affects the network, to a point where it virtually dominates the dynamics. Depending on the embedding scheme, pulse propagation is possible, but much more volatile than in an isolated chain. Therefore, ideas to generalize the synfire definition are proposed in order to cope with this vulnerability.

Author dc:creator

• Hehl, Ulrich

Contributors dc:contributor

• Aertsen, Ad

• Synfirekette
• Einbettung
• Zeitcode
• Parallelrechner
• präzise raumzeitliche Muster
• synfire chain
• synchronous activity
• model based on quantitative anatomy
• spiking cortical neural network simulation on parallel computers

Repository record source_url

https://freidok.uni-freiburg.de/data/340

OAI identifier oai:identifier

oai:freidok.uni-freiburg.de:340