Abstract :
[en] This dissertation explores the mechanisms behind reliable neuromodulation in neurons exhibiting significant variability in ion channel properties. Neuromodulation, driven by agents such as dopamine, serotonin, and histamine, dynamically influences neuronal activity and network behavior by targeting ion channels via metabotropic receptor pathways. These complex signaling cascades must overcome intrinsic neuronal heterogeneity, posing challenges to understanding their consistent effects. This work combines computational neuroscience, dynamical systems theory, and control engineering to uncover how neuromodulation achieves reliability in such variable systems. The research is divided into three major contributions:
- Understanding Neuronal Degeneracy Mechanisms: Using dimensionality reduction techniques on conductance-based models, two primary mechanisms of neuronal degeneracy are identified. The first, homogeneous scaling, involves uniform proportional adjustments in ion channel conductances, while the second, variability in conductance ratios, reflects cryptic variability in dynamic properties that manifest under perturbations. These mechanisms often compete or align depending on the neuromodulatory state, affecting conductance correlations. This analysis highlights how neuromodulation follows indirect but neuron-type-specific trajectories in conductance space, ensuring robust effects despite variability.
- Development of a GPCR-Based Neuromodulation Controller: A novel control system inspired by G-protein coupled receptor signaling pathways is introduced. The controller translates neuromodulatory signals into real-time adjustments of ion channel densities using dynamic input conductances as control variables. This adaptive system models neuronal excitability as a feedback control loop, simplifying the complexity of neuromodulation. Simulations demonstrate its ability to maintain consistent neuronal activity patterns across a population of neurons with widely varying ion channel properties.
- Interaction Between Neuromodulation and Homeostasis: The synergy between neuromodulation and homeostatic regulation is explored. Neuromodulation selectively adjusts specific ion channel correlations, while homeostasis broadly scales all conductances to maintain intracellular calcium levels. Modeling their interaction reveals how these complementary mechanisms maintain robust neuronal activity. The combined approach mitigates potential failures that would arise from either mechanism in isolation.
Additionally, the principles of neuromodulation are extended to robotics, where a neuromodulatory network is used to dynamically reconfigure locomotion patterns in a quadruped robot. This system enables seamless transitions between rhythmic gaits, such as trotting and galloping, by modulating specific neuromodulatory neurons. The robustness and adaptability of this approach offer potential applications in neuromorphic engineering and neural network control. By integrating experimental and computational insights with control theory, this dissertation provides a comprehensive framework for understanding how neuromodulation achieves reliability in the face of neuronal variability. It advances the theoretical foundation for both neuroscience and bio-inspired engineering applications.