Identifying response features suitable for building a factor model.

2017-04-06T18:36:20Z (GMT) by Michael K. J. Sproule Maurice J. Chacron
<p>18 measures were chosen to capture the variation observed across the entire population of ELL pyramidal cell neurons in response to a communication signal (i.e. “small chirp”) occurring at different phases (0 90 180 and 270 degrees) of a continuous beat cycle. Measures are described and their origin within the complete stimulus waveform are indicated. Beat cycles (1–2) precede chirp onset while beat cycles (2–4) proceed the 100ms chirp window following chirp onset. The “data type” refers to the distinct stages of preprocessing from which the 18 measures originate (spike times, cycle histogram or PSTH). The total number of measures considered “Number per neuron”, and of those, the ones that are normally distributed are tallied. The identities of the stimuli (i.e. chirp phase or beat) from which the normally distributed measures belong are indicated. Collinear relationships among this subset of measures are identified. From the collinear pairs of measures identified one was randomly chosen for removal to yield the final features used in CFA classification which are similarly tallied.</p>