12.07.2015 Views

Dynamical Systems in Neuroscience:

Dynamical Systems in Neuroscience:

Dynamical Systems in Neuroscience:

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

158 Conductance-Based Models5.2.2 Equivalent potentialsInspired by the reduction idea of Kr<strong>in</strong>skii and Kokoz (1973), Kepler et al. (1992) suggesteda systematic method of reduc<strong>in</strong>g the complexity of conductance-based Hodgk<strong>in</strong>-Huxley-type modelsC ˙V = I − I(V, x 1 , . . . , x n )ẋ i = (m i,∞ (V ) − x i )/τ i (V ) , i = 1, . . . , n ,where x 1 , . . . , x n is a set of gat<strong>in</strong>g variables. The goal is to f<strong>in</strong>d certa<strong>in</strong> patterns orcomb<strong>in</strong>ations of the gat<strong>in</strong>g variables that can be lumped to reduce the dimension ofthe system. For example, we want to comb<strong>in</strong>e all resonant variables operat<strong>in</strong>g on asimilar time scale <strong>in</strong>to a “master” recovery variable, then do the same for amplify<strong>in</strong>gvariables.Let us convert each variable x i (t) to the equivalent potential v i (t) that satisfiesx i = m i,∞ (v i ) .In other words, the equivalent potential is the voltage which, <strong>in</strong> a voltage clamp, wouldgive the value x i when the model is at an equilibrium. Apply<strong>in</strong>g the cha<strong>in</strong> rule tov i = m −1i,∞ (x i), we express the model above <strong>in</strong> terms of equivalent potentials:C ˙V = I − I(V, m 1,∞ (v 1 ), . . . , m n,∞ (v n )) ,˙v i = (m i,∞ (V ) − m i,∞ (v i ))/(τ i (V ) m ′ i,∞(v i )) .S<strong>in</strong>ce the Boltzmann functions m i,∞ (V ) are <strong>in</strong>vertible, the denom<strong>in</strong>ators do not vanish.No approximations have been made yet; the new model is entirely equivalent to theorig<strong>in</strong>al one, it is just expressed <strong>in</strong> a different coord<strong>in</strong>ate system. The new coord<strong>in</strong>ates,however, expose many patterns among the equivalent voltage variables that were notobvious <strong>in</strong> the orig<strong>in</strong>al, gat<strong>in</strong>g coord<strong>in</strong>ate system.Kepler et al. (1992) developed an algorithm that substitutes resonant and amplify<strong>in</strong>gvariables by their weighted averages. The weights are found us<strong>in</strong>g Lagrangemultipliers and strictly local criteria aimed at preserv<strong>in</strong>g the bifurcation structure ofthe model. There is also a set of tests that <strong>in</strong>forms the user when the method is likelyto fail. The method results <strong>in</strong> a lower-dimensional system that is easier to simulate,visualize, and understand.5.2.3 Nullcl<strong>in</strong>es and I-V record<strong>in</strong>gsWe saw that the form and the position of nullcl<strong>in</strong>es provided important <strong>in</strong>formationabout the neuron dynamics, i.e., the number of equilibria, their stability, the existenceof limit cycle attractors, etc. The same <strong>in</strong>formation, <strong>in</strong> pr<strong>in</strong>ciple, can also be obta<strong>in</strong>edfrom the analysis of the neuronal current-voltage (I-V) relations. This is not a co<strong>in</strong>cidence,s<strong>in</strong>ce there is a profound relationship between nullcl<strong>in</strong>es and experimentallymeasured I-V curves.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!