View source for DivInE-model for MT neurons


= '''DivInE'''-model for MT neurons =
This model is also termed '''DivInE'''-model, since it describes the adaptive response properties of MT neurons by means of divisive normalization, for more detailed info see also [https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1009595 this paper]:

<math display="block">\tau_e \frac{dA_e(t)}{dt} = -A_e(t) + g_e\left( \frac{I(t)}{A_i(t)+\sigma} \right)</math>

<math display="block">\tau_i \frac{dA_i(t)}{dt} = -A_i(t) + g_i\left( I(t) \right)</math>

Here, <math display="inline">g_X</math> are gain functions for <math display="inline">x\in\{e,i\}</math> with <math display="inline">g_X(I) = m_X(I-\theta_X)</math> for <math display="inline">I>\theta_X</math>, and <math display="inline">0</math> otherwise, while <math display="inline">A_e</math> and <math display="inline">A_i</math> could be interpreted as internal activations. Default parameters: <math display="inline">\tau_e=10</math> ms, <math display="inline">\tau_i=40</math> ms, <math display="inline">\theta_{\{e,i\}}=0</math>, <math display="inline">m_e=m_i=1 nA^{-1}</math>, <math display="inline">I=1</math> nA, <math display="inline">\sigma=0.25</math>. From the activation <math display="inline">A_e</math>, an output rate can be derived via <math display="inline">r(t) = r_0 A_e(t)</math> with, let’s say, <math display="inline">r_0=100</math> Hz.