Neuromodulation can induce reliable switches and tuning of behaviour output via signaling pathways that use a handful of molecular species such as hormones, monoamines and neuropeptides. The circuit mechanisms that enable these switches remain mysterious, but are fundamental to brain function. We are studying how individual neurons and small networks of neurons self-organize to express an appropriate combination of ion channels and receptors so that reliable modulation can occur. (See Marder, O’Leary & Shruti 2014 for a further outline of this conceptual problem).
Our ongoing theoretical work (in collaboration with Guillaume Drion, Alessio Franci and Eve Marder) has shown how activity-dependent feedback control of ion channels and receptors can permit a neuron or circuit to respond to a modulator that has multiple molecular targets in a way that is robust to variability in the cell and circuit properties.
This work will help us understand how coordinated switches in behavior can occur as a result of neuromodulation across the entire nervous system and how this signalling can fail, or lead to pathologies. Examples of behavioral switches include motor output (running vs walking vs swimming) as well as changes in excitability and synaptic integration that occur in the nervous system during different behavioral states such as anxiety, courtship and attentional arousal. We are very interested in finding experimental collaborators with interests in how modulation works at the cell, circuit and behavioral level!
How does nervous system robustness trade-off with flexibility and performance?
Neurons and neural circuits need to regulate their properties in a way that compensates fluctuations in activity and the environment (O’Leary & Marder 2016), but does added robustness imply diminished flexibility in the kinds of output they can produce? Similarly, do robustness and flexibility trade-off with metabolic efficiency or information transmission? These questions are extremely important for understanding the constraints that shape how nervous systems work, but are relatively unexplored at present because most work tends to focus on the consequences of optimizing a single property (such as information capacity or energy efficiency).
We are investigating relationships between robustness, flexibility and performance using ideas from control engineering. Preliminary results (O’Leary et al 2014) show how homeostatic regulation mechanisms that exist in neurons can provide robustness to ‘normal’ physiological perturbations (such as changes in network activity) but can also cause pathological sensitivity to ‘unexpected’ perturbations, such as loss of specific ion channel genes.
In addition to providing a theoretical framework for basic questions about fault-tolerance and robust function in the nervous system, we want to apply these concepts and models to understand pathologies that are observed in real life and that may be explicable in terms of aberrant compensation, including epilepsy and addiction.
Flexible & robust nervous system function from reconfiguring networks
When we recognize a sound, initiate a limb movement or navigate a familiar environment, we trigger a pattern of neural activity that faithfully represents events, sensations and actions. Remarkably, such neural representations are maintained in spite of the fact that the neural circuits in our brains are continually being modified by experience, environmental perturbations and turnover of cellular signaling components. More remarkably, very recent experiments show that patterns of neuronal activity and connectivity that support stable percepts and associations can themselves continually reconfigure over time, to the extent that spiking activity representing a fixed feature of the environment moves from one set of neurons to another (Lutcke et al 2013). These findings represent a significant obstacle to understanding how brain circuitry gives rise to behaviour and challenge existing theories of brain function, which assume that stable circuit output is reflected in stable underlying connectivity, physiology and activity patterns.
This project will explore novel applications of Control Theory to develop a theory of reconfiguring neural circuits based on biological principles. This is split into three specific objectives:
Objective #1: Interpret and analyze known neural plasticity mechanisms at the single neuron and microcircuit level using control theory; identify how mechanisms interact to support continual parameter reconfiguration and stable function.
Objective #2: Characterize functional properties of reconfigurable neural circuits that perform specific tasks, especially robustness and performance tradeoffs, using the underlying mechanisms (Objective #1) as building blocks.
Objective #3: In close collaboration with experimentalists, relate conceptual, computational and formal mathematical models of reconfiguring neural circuits (from Objectives #1 & #2) to real-world biological experiments, generate predictions and design new experiments.