[visionlist] microsaccades in natural vision

James A. Bednar jbednar at inf.ed.ac.uk
Wed Aug 8 17:09:19 GMT 2007


|  From: Francisco J. Flores
|  Date: Aug  8 10:56:34 2007 -0400
|  
|  Hi all. Does anyone knows if microsaccades are actually occuring
|  during spontaneous ocular fixations? I'm having a hard time trying
|  to find microsaccades in recordings of eye movements when the
|  subject is freely exploring a natural image.

There was a poster at the Society for Neuroscience in 2002 claiming
that microsaccades are rare when viewing natural images or natural
video (see abstract attached), but I'm not sure if this has made it
into an archival publication yet...

Jim


@InProceedings{gray:sfn02,
  booktitle    = "Society for Neuroscience Abstracts",
  year	       = 2002,
  publisher    = "Washington, DC: Society for Neuroscience",
  note	       = "Program no. 622.3",
  author       = "C. M. Gray and T. Mckeehan and B. A. Olshausen and
                  S. C. Yen",
  institution  = "Center for Computational Biology Montana State Univ",
  title	       = "Neuronal Dynamics in Macaque V1 During Free-Viewing
                  of Natural Images",
  abstract     = "We have developed a simple paradigm to study striate
                  cortical activity during conditions similar to those
                  occurring naturally. In two monkeys, we measured eye
                  position using a scleral search coil and monitored
                  neuronal activity in V1. The animals were trained to
                  alternate between a fixation task for calibrating
                  eye position and for mapping receptive fields using
                  sparse noise stimuli, and a free-viewing task in
                  which they were allowed to freely inspect natural
                  images consisting of monochromatic stationary
                  pictures and dynamic movies of natural
                  scenes. During free-viewing, saccadic eye movements
                  occurred at median rates of 3-5/sec and visual
                  fixations had median durations of ~200 ms. The
                  animals showed no evidence of microsaccades during
                  fixation episodes, but the eyes exhibited drifts in
                  position varying in magnitude (0.01-0.2 deg) and
                  velocity (0.1-1.0 deg/sec). Neuronal activity was
                  overwhelmingly sparse. When sampled across all
                  fixations (40 ms resolution), firing rates were zero
                  60-80% of the time and showed a steep decrease in
                  probability with increasing rate. When averaged
                  across all fixations, neuronal responses occurred at
                  latencies of 40-50 ms, peaked at ~80 ms and decayed
                  to baseline by 150-300 ms. Small clusters of 2-3
                  neurons recorded on the same electrode often showed
                  decorrelated responses. When one neuron was
                  responsive, adjacent neurons often showed either
                  inhibitory responses or no response at all. Together
                  these data support previous observations that
                  neuronal activity in V1 during free-viewing is
                  sparse and decorrelated.",
}


More information about the visionlist mailing list