Date of Award

Spring 5-15-2020

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Doctor of Philosophy (PhD)

Degree Type



A compelling vision for the future of neuroscience is the ability to sense neural activity throughout the bulk of the brain with exquisite resolution. Popular visions usually include intricate electrode technology intruding into the neuropil, meandering along nerve tracts, and sensing the whole brain. These popular visions stem from the belief that we must always have an outsider’s perspective of neural activity. According to this belief the closest thing neuroscientists can achieve to an insider’s perspective is to shadow every neuron (or almost every neuron) with an electrical or optical recording device. Yet, the brain naturally has an expansive sensor network. The brain already aggregates and organizes neural activity according to computational function. The brain does this through the operation of single neurons, which have arrays of many dendrites to process inputs arriving from far and wide. These processed inputs are concentrated at the soma of the neuron where they drive rich dynamics, and where the neuron translates these inputs into outputs. One of the most venerable methods in neuroscience, the patch-clamp intracellular recording technique, can record these rich input driven dynamics. Neuroscience has xi long held the goal of patching into the full network dynamics with patch-clamp, but it is difficult to reconstruct network dynamics information. Fortunately, the neural criticality hypothesis provides a justification for expecting to find network dynamics information, and the modern field of nonlinear dynamics provides tools for reconstructing full dynamics from scant information. The neural criticality hypothesis is the idea that the brain can exhibit phase transitions, but tunes itself to sit at a point (called a “critical point”) between two phases where the most dynamical complexity arises. One of the key phenomena of critical systems is “scale-freeness” which is widely observed in the brain. One implication of Scale-freeness is that some statistics are always the same whether observed at very small scales or very large. For critical phenomena scale-freeness is both extensive and precise, if scale-freeness is limited in a system then it suggests that system is not a critical system. We adopt condensed matter physics’ rigorous standards for experimentally identifying critical systems. We show that we can meet these standards with long intracellular recordings. We also show that our findings agree with large scale population recordings. After establishing this proof-of-concept, we then use new methods for modeling nonlinear dynamical systems to extract small details about visual stimulus from short intracellular recordings. These details were too small to be reliably detected in the output of neurons. We use models of nonlinear dynamics because of their relationship to a neural coding paradigm: Attractor network theory. Thus, we also have novel evidence of dynamical attractor based neural code in primary visual cortex. Therefore, we have advanced both the neural criticality hypothesis and the attractor network theory of neural coding while demonstrating that we can patch into in-situ neural communication networks and get information that previously required electrode arrays or other population recording methods.


English (en)

Chair and Committee

Ralf Wessel

Committee Members

Anders Carlsson, Zohar Nussinov, Baranidharan Raman, Mikhail Tikhonov,