Please use this identifier to cite or link to this item: http://arks.princeton.edu/ark:/88435/dsp015q47rr20n
 Title: Learning to Induce Firing Patterns: Closed-loop Stimulation as a Stepping Stone Towards Adaptive Cognitive Prosthetic Authors: Steinhardt, Cynthia Advisors: Buschman, Timothy Department: Independent Concentration Class Year: 2016 Abstract: Over the course of history, prosthetics have progressed from an aesthetic choice to a tool capable of partial or fully allowing recovery of formerly lost abilities. Neuroengineering progress has allowed these improvements to be internalized for treatment of formerly life threatening and debilitating conditions, such as Parkinson’s disease and epilepsy. While in the past, permanently damaging treatments, such as resection of related brain tissue was the primary form of treatment, neuroengineering has provided alternatives in the form of implantable devices for closed-loop stimulation that read outputs from the brain through a computer, detect, intercept, and override harmful signals between brain regions. This study focuses on the creation and initial testing of an adaptive algorithm and interface to be used in a brain-computer interface that could overcome a lesion to a neural circuit in the brain, as well as the behavioral apparatus on which to perform further in vivo testing of an algorithm. The first specific aim of this experiment was to compare the performance of algorithms that searched for possible input patterns. We tested algorithms that explored the space of possible inputs by 1) random walk (baseline), 2) sequential exploration of connections of a subset of neurons (by Sequential and Taylor series based algorithm) 3) and exploration of various frequencies of activation patterns (discrete cosine transform, principle component analysis, and addition of Gaussian based algorithms). Relative performances of these algorithms on networks of increasing size and complexity of connectivity were tested to assess relative usefulness of algorithms overall and for specific types of networks (e.g. networks with distributed and localist connectivity). 2 The second aim was to set up a stimulating and recording system for electrophysiology and optogenetics. The system was initially tested on the somatosensory attention system in mouse cortex. The extent to which firing could be induced and captured in terms of frequency and amplitude of response across a population was measured. The ability to target cell-types related to attention and stimulate those cells optogenetically was also measured. Optogenetically-induced firing was then compared to recorded activation in response to external stimulation-induced responses to test for the ability to induce synchrony that emulates synchrony induced by attending stimuli. Tests were performed in an apparatus built for an awake behaving mouse which proved to be useful for awake, behavioral testing for hundreds of trials at a time. Finally, the cortical damage induced by several weeks of stimulating and recording was also assessed. Together these tests were used to confirm the constructed apparatus would be useful for in vivo testing of the adaptive algorithm and measure the level spatial and temporal specificity with which optogenetics and electrophysiology could be used in a BCI to overcome impairments induce by lesions to a neural circuit in the brain. Extent: 137 pages URI: http://arks.princeton.edu/ark:/88435/dsp015q47rr20n Type of Material: Princeton University Senior Theses Language: en_US Appears in Collections: Independent Concentration, 1972-2020

Files in This Item:
File SizeFormat