Analysis of spontaneous and sensory-driven activity in ferret
V1.

J. Zhao1,2, G. Szirtes1, M. Eisele1, J. Fiser3,4, C. Chiu4,
M. Weliky4, and K.D. Miller1.  1Center for Theoretical Neuroscience
and 2Dept. of Physics, Columbia University, NYC NY; 3Dept. of
Psychology, Brandeis U., Waltham MA; 4Center for Visual Science,
U. Rochester, NY.

We analyze multiunit recordings from linear arrays of 16 electrodes
spanning 3 or 9 mm in awake ferret V1, as in Fiser et al. Nature
431:573 (2004).  Recordings were made at ages ranging from 29 to 168
days postnatal.  Fiser et al. 2004 found that activity from P30 to P90
was dominated by similar activity patterns whether in dark or when
stimulated by white noise or a natural movie. They showed that
temporal correlations on a single electrode were long at early ages
but became progressively shorter, while spatial correlations at a
single time were short-ranged at early ages but became long-range at
later ages.  Correspondingly, activity patterns became dominated by
bursts spanning all electrodes.

We find the principal components of simultaneous activity across the
electrodes.  At later ages, most of the variance is in the first
component, which is uniform across electrodes (each electrode deviates
by the same number of standard deviations from its mean activity).
This component's autocorrelation shows some tendency to oscillate,
with a bump of power in the range 10-17Hz.  This temporal structure is
quite similar for dark and movie stimuli.  However, for noise stimuli,
particularly at ages >= P120, very long-lasting oscillatory
autocorrelation at 11-12 Hz is seen. This may represent alpha
activity, which has been argued to represent an "idle" or "disengaged"
state, suggesting the awake animal may disengage from the noise
stimulus.  More generally, this dominant first component seems likely
to represent a global state rather than specific visual input.
Subtracting off the principal component, the remaining activity shows
correlations that are much more localized in space and time. Power in
the remaining activity seems to fall off as a power of spatial
frequency, suggesting that it might have no characteristic spatial
scale.  Supported by RO1-EY13595 (KDM) and RO1-EY 012494 (MW) from the
NEI.