Complex animal behaviors are supported by computations made across large ensembles of neurons distributed among multiple brain regions. Our current understanding of how individual neurons and the circuits they are a part of represent and process information has depended upon the ability to observe the millisecond timescale dynamics of these neuronal networks. The multi-electrode array has been the principle tool utilized for the isolation of large numbers of simultaneously recorded neurons, though current approaches have required a tradeoff among resolution, spatial coverage, longevity, and stability. Alongside the recordings that these arrays are capable of producing, existing methods of assigning times and labels to the continuously sampled extracellular voltage trace (a process termed spike sorting) has required extensive manual input.

To address the shortcomings in the currently available tools, we developed a novel spike sorting software suite, MountainSort, and a new polymer probe-based, modular recording platform. In this thesis I describe, validate, and demonstrate the utility of these new tools in their ability to: (1) cluster neural events, (2) stratify the qualities of these clusters to identify putative single-units, (3) outperform other spike sorting methodologies, (4) record from hundreds of neurons distributed across multiple regions simultaneously, (5) record single-units for months, and (6) record from the same single-units for over a week. I propose that together, these tools enable the study of previously inaccessible questions.