Gamma synchronization and spike timing.
Technische Universität, Darmstadt
[Ph.D. Thesis], (2013)
Available under Creative Commons Attribution Non-commercial No Derivatives, 2.5.
Download (9MB) | Preview
|Item Type:||Ph.D. Thesis|
|Title:||Gamma synchronization and spike timing|
Neural synchrony is a curious phenomenon of the brain and has been studied now for decades. In this thesis, I made further steps in investigating the functions of neural synchrony using two approaches: analysis of the spike trains recorded from cat primary visual cortex, and theoretical analysis of simulated spike trains. In the first study, I investigated autocorrelations of the spike trains in response to moving gratings. Essentially, the autocorrelation is a second order statistic of spike times. The majority of the observed autocorrelations were oscillatory with an average period in the range of beta/gamma band (~ 25 Hz). Interestingly, the oscillation frequencies were modulated by the direction of the moving gratings. This modulation effects fluctuated in the range of ~ 3 Hz. Those revealed the potential coding capabilities of oscillation frequencies in the spike trains. In the second study, I contributed to the development of scaled-correlation analysis. Scaled correlation analysis is a method that enables the isolation of autocorrelation and cross-correlation histograms of fast signals components. I produced an analytical proof that scaled correlation implicitly attenuates the slow-frequency components in a similar way as a high-pass filter. In the third study, I investigated the relation between statistics of spike times on one side and the rules for obtaining spike time dependent plasticity (STDP) on the other side. It is well established that STDP is characterized by a time range of ~100 ms within which the spike time dynamics can lead to plastic modification of synapses. Results of previous studies in our group (Schneider et al. 2006, Havenith et al., 2011) and those of König et al (1995) showed that the difference of the first order statistic of spike times, i.e. the relative spike times change with the stimulus property. These relative spike times do not exceed ~ 15 ms. Therefore, the range of relative spikes time falls into the range of STDP. Using a random walk model of the STDP and assuming Poisson spike trains, I derived the first two moments of the synaptic weights evolving stochastically over time. It became clear that the average synaptic weight at the equilibrium depends most heavily on two factors, the level of the synchrony between input and output spikes, and the relative spike times. We obtained the same results for an additive and a multiplicative STDP model. In addition to the theoretical work, I estimated the degree to which spikes recorded in vivo were able to modify the putative synapses which may exist between the recorded neurons. The spikes were fed to an experimentally established and computer-simulated STDP model, known as the suppression model (Froemke and Dan, 2002). Results for suppression model were comparable to those obtained from the theoretical analysis, despite the marked difference between the STDP models used in the two analysis (suppression model vs. additive model) and between the spike trains (patterned in the gamma-band vs. Poisson distributed). I could show that this is due to the fact that the prediction of the suppression model converges to the additive models of STDP when the rate of spikes is sufficiently low. In conclusion, the present work shows that the frequency of oscillatory patterns in spike trains can be used potentially to code stimulus-related information, these phenomena can be studied efficiently using scaled-correlation even if spike trains are polluted with slow-frequency components, and the time-delays in the resulting synchrony between spike trains can significantly affect the connectivity of the underlying network via STDP.
|Place of Publication:||Darmstadt|
|Classification DDC:||500 Naturwissenschaften und Mathematik > 500 Naturwissenschaften
500 Naturwissenschaften und Mathematik > 570 Biowissenschaften, Biologie
|Divisions:||10 Department of Biology
10 Department of Biology > Computational Biology and Simulation
10 Department of Biology > Systems Neurophysiology
|Date Deposited:||31 May 2013 07:41|
|Last Modified:||31 May 2013 07:41|
|Referees:||Galuske, Prof. Ralf and Laube, Prof. Bodo|
|Refereed:||24 April 2013|