direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

Anwendungen

Ausgewählte Abstracts

Influence of recurrent excitation and inhibition on the response dynamics in a network model of orientation tuning
Zitatschlüssel wimm07
Autor K. Wimmer and R. Martin and M. Stimberg and K. Obermayer
Buchtitel Proceedings of the 7th Meeting of the German Neuroscience Society / 31th Göttingen Neurobiology Conference 2007
Jahr 2007
Zusammenfassung The relative contribution of afferent, recurrent inhibitory and recurrent excitatory inputs and their relevance for cortical computations have long been a matter of controversy. Spike tuning to orientation in V1 has been simulated successfully in artificial neural networks with varying contributions of these components. Using intracellular measurements to constrain the model space, the most likely operating regime for a network model composed of Hodgkin-Huxley neurons in an artificial orientation map was previously shown to be recurrent with both, significant excitation and inhibition, and with parameters close to the unstable region of network operation [1, 2]. Here we extend the analysis from the steady state response of the network to its temporal dynamics. The analysis of the network response kernel for simulated reverse correlation input shows that feed forward dominated and balanced recurrent regimes closely track the afferent input, showing sharp onsets and offsets in their responses. Inhibition dominated networks show a shortened response while excitation dominated networks have a prolonged time course; both are incompatible with data from reverse correlation experiments. A high-firing-rate recurrent regime with strong excitation and moderate inhibition lacks the sharp offset but it is marked by critical slowing down near the phase transition to the unstable region. When considering the time course for neurons located either in the orientation domain or near a pin wheel center separately, only the balanced and feed forward regimes show location invariant responses to preferred orientation. Further analyses to constrain the operating point for cortical computations, such as an analysis of the response variability and the difference between the network's impulse response and the response kernel, estimated through ongoing stimulation, are also presented. In sum, the results for the balanced recurrent and for the feed forward regimes are best compatible with existing experimental results. Thus the evidence from the steady state simulations and from the temporal responses for different orientation map locations converges. It points towards a balanced regime with significant recurrent contributions that processes moderately tuned afferent input as the likely operating point of cortical networks processing sensory stimuli. [1] Mariño J et al., Nat Neurosci 8 (2005) [2] Wiesing P, et al., Soc Neurosci Abstr (2005)
Download Bibtex Eintrag

Zusatzinformationen / Extras

Direktzugang:

Schnellnavigation zur Seite über Nummerneingabe