Related Experiment Videos
Adaptation of retinal processing to image contrast and spatial scale
S M Smirnakis1, M J Berry, D K Warland
1Department of Physics, Harvard University, Cambridge, Massachusetts 02138, USA.
Nature
|March 6, 1997
Summary
Neural circuits adapt to visual input variations. Retinal ganglion cells adjust to image contrast and spatial patterns, enhancing visual processing and potentially human contrast adaptation.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Vision Science
Background:
- Neurons face a dynamic range trade-off, limiting input encoding and gradation resolution.
- Light adaptation adjusts retinal sensitivity to mean light intensity but may not fully address complex visual scenes.
- Efficient visual encoding likely requires adapting to the full intensity distribution, not just the mean.
Purpose of the Study:
- To investigate if retinal ganglion cells adapt to image statistics beyond mean intensity, specifically contrast and spatial correlations.
- To characterize the timescale and magnitude of this adaptation.
- To identify the neural mechanisms underlying this adaptation within the retina.
Main Methods:
- Recording from retinal ganglion cells in response to controlled visual stimuli.
- Manipulating image contrast and spatial correlations while keeping mean intensity constant.
- Analyzing firing rate changes and their temporal dynamics.
Main Results:
- Retinal ganglion cells adapt to both image contrast and spatial correlations, even with constant mean intensity.
- Adaptation occurs over seconds, significantly slower than the immediate light response.
- Observed 2-5 fold changes in firing rate, mediated by at least two sites within the retinal network after photoreceptors.
Conclusions:
- Retinal processing exhibits significant plasticity, adapting to complex image statistics.
- This adaptation to contrast and spatial correlations may be crucial for efficient visual encoding.
- The findings suggest a mechanism contributing to human contrast adaptation in vision.