M.Sc Student | Chen Ronen |
---|---|

Subject | Dynamic Adaptation in Single Neuron Models |

Department | Department of Chemical Engineering |

Supervisor | Professor Naama Brenner |

Neural systems are known for their ability to adapt to a wide range of conditions. In particular sensory systems, which process signals and transmit information, adapt to the statistical properties of incoming signals. Recent experiments have shown that sensory neurons implement a gain control mechanism, where the input/output function is rescaled by a constant factor, proportional to the standard deviation of the incoming signal. An important question, which arises from these findings, is whether this adaptive behavior is a single cell or a network property; modeling provides an important tool in answering this question.

The goal of the current theoretic work was to study the response of the Hodgkin Huxley model neuron to various stimulus ensembles. We used computer simulations of the model neuron exposed to incoming signals. Signals were chosen from stationary distributions with different standard deviations and correlation times, and input/output relations were computed for each ensemble.

Our results show that only some of the effects of statistical adaptation appear in the Hodgkin-Huxley model, and these appear in a limited range of parameters of the incoming signal. Specifically, for rapidly varying signals a rescaling of the input/output function is found, however as the signal correlation time increases this effect disappears, in contrast to experimental results. Thus more complicated models are required to understand neuronal adaptation.

Effective degrees of freedom can be added to the Hodgkin-Huxley model as slow variations in the ionic conductances. In the quasistatic limit, we studied the dependence of the input/output relation on the maximal ionic conductances. We found that the global input/output function under a statistical stimulus ensemble depends on the ratio between sodium and potassium conductance, and not on their separate values.