|Ph.D Thesis||Department of Electrical Engineering|
|Supervisors:||Prof. Meir Ron|
|Prof. Marom Shimon|
|Full Thesis text|
Biological systems of practically any level of organization exhibit complex dynamics, reflecting the multitude of underlying processes and interactions involved in their make-up. Such systems are composed of large heterogeneous populations of non-linear and stochastic elements, and are constantly modified by their environment. In the face of this overwhelming complexity, it is extremely difficult to identify the state variables that are relevant to the functionality of the system, and to access them experimentally. New theoretical approaches and experimental paradigms are needed in order to meet this challenge; the present thesis describes my efforts in this direction.
The main contribution of the research presented in this thesis is the development of the Response Clamp, a closed-loop experimental technique for the study of neural systems. A pre-defined feature of the system's output is "clamped" by implementing a control circuit that measures this feature, compares it to a chosen preset value, and corrects errors by changing stimulation parameters. This is achieved using a real-time design that implements a Proportional-Integral-Derivative (PID) controller. This procedure enables tight control over the functionality of interest, as well as exposure of the internal state variables which are relevant to this functionality.
The effectiveness of the Response Clamp methodology is demonstrated at two different levels of organization. First, the method is used to gain direct, continuous access to the threshold dynamics of single neurons embedded in a network developing in-vitro, and to study the interactions between this dynamics and the synchronous activity in the network. Second, the method was used to study cognitive performance in a simple perception-reaction task in human subjects, and to explore the sources and possible roles of the dynamic instability exhibited in such tasks.
In addition to the Response Clamp, the thesis presents a formal theoretical framework that aids in the analysis of complex, structured, heterogeneous systems. This approach is applied in the development of a simple mechanistic model for selective adaptation, a macroscopic phenomenon in which the system adapts to one source of stimulation while preserving and even enhancing its sensitivity to other sources. The model leads to several predictions that were corroborated in computer simulations and in large-scale networks of biological neurons in-vitro. Remote from the Response Clamp framework as this section is, it shares the common objective of extracting state variables that expose the relations between phenomena of excitability at different levels of organization.