Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

SUPPLEMENTARY MATERIALS Decoding Procedure. We applied a decoding procedure to further describe the

Decoding Procedure. We applied a decoding procedure to further describe the
information about the impending movement(s) that is encoded by the planning activity of the
whole population. A one-nearest-neighbor classification algorithm (Duda et al., 2001)
classified every given test trial by comparing the neural response of the whole population
during this test trial with the population responses during other trials. The decoding was done
for the PRR populations in each monkey separately. The datasets that were used for the
training of the algorithm as well as for the decoding consisted of 70 neurons for monkey C
and of 42 neurons for monkey Z. Since the number of trials that had been recorded in the
various neurons under each condition was not always the same, we included only the first five
recorded trials per condition in the dataset. Each of these five trials was then classified (as a
test trial) based on the similarity of its neural signature in the population and the activation
vectors in all other (four) trials used as a training set (`leave-one-out decoding' procedure).
This procedure was iterated 500 times. The test trial was assigned to the same class as the
training trials with the smallest Euclidian distance (i.e., its `nearest neighbor') within the n-
dimensional activity space (n corresponds to the number of cells). The decoding performance
was defined as the percentage of correctly classified test trials. We combined trials from
single- and double-reach blocks and used a general classification system to predict the
direction of the upcoming reach as well as the task type, i.e. whether the movement is going
to be a single- or double reach (`full decode'). For a subset of 24 cells in one monkey that had


Source: Andersen, Richard - Division of Biology, California Institute of Technology


Collections: Biology and Medicine