The spatial scale of contrast adaptation in the retina

When visual contrast changes, neurons in the retina adapt by changing their sensitivity and temporal window of stimulus integration to better match the range of incoming signals. What happens, however, when contrast only changes in some parts of a scene, for example, when a high-contrast object appears in front of a low-contrast background? By analyzing the activity of retinal neurons under spatially local changes in contrast, we found that these neurons adapt in a global fashion, adjusting how they process visual stimuli even at locations where visual contrast did not change.

HFSP Career Development Award holder Tim Gollisch and colleagues
authored on Thu, 07 March 2013

Our visual system has to constantly adjust its processing characteristics to changing stimulus conditions, such as mean light level and contrast. These adaptation processes start in the retina, the first stage of visual processing in vertebrates. When contrast increases, retinal ganglion cells decrease their sensitivity and shorten the temporal window over which they integrate incoming signals. In the natural environment, however, contrast is not necessarily homogeneous over space. Individual objects may appear at high contrast within a low-contrast environment, such as the rear lights of the car ahead of you amidst the foggy landscape along a country road. This raises the question whether the adaptive phenomena are confined to the spatial location of an appearing high-contrast object or whether other regions of visual space are also affected. This has important consequences for visual processing because it determines whether high-contrast objects suppress the representation of low-contrast objects and affects the representation of moving objects.

To investigate the spatial scale of contrast adaptation, we recorded the spiking activity from retinal ganglion cells in isolated salamander retina under visual stimulation that included locally confined changes in contrast. Each ganglion cell is sensitive to light from an extended region of space – its spatial receptive field – and this receptive field may contain more than a single high-contrast or low-contrast object. We therefore analyzed whether the processing of visual stimuli at a fixed contrast is affected when visual contrast changes in some other region of the receptive field.

Figure: Determining the spatial scope of contrast adaptation in retinal ganglion cells. The visual stimulus is subdivided into several subfields, in which flickering light intensity is shown. For some of these subfields, the intensity range is suddenly enlarged, corresponding to an increase in contrast. Temporal windows of stimulus integration and sensitivity are analyzed through a linear–nonlinear cascade model by computing linear filters and nonlinearities for subfields where contrast changes (top). Both model components change in a similar fashion for subfields with and without contrast changes, indicating the global nature of contrast adaptation in these cells.

Indeed, we found that ganglion cells process local visual signals differently, depending on the visual contrast at other locations within the cells’ receptive fields. In fact, both the sensitivity and the temporal window of stimulus integration change, and the changes are very similar to those observed for locations where contrast really does change. This shows that contrast adaptation occurs mostly in a global fashion, affecting the receptive field in its entirety rather than being restricted to local subfields where contrast changes actually occur. From a mechanistic point of view, this is surprising because previous investigation had suggested synaptic dynamics at the inputs into ganglion cells as the origin of contrast adaptation, which could mediate local adaptation effects. Using computational models, however, we found that nonlinear signal processing at these synapses lead to complex, counterintuitive phenomena, which seem to lie at the heart of a variety of observed contrast adaptation characteristics.

Reference

Local and Global Contrast Adaptation in Retinal Ganglion Cells, Mona M. Garvert and Tim Gollisch (2013). Neuron 77915-928, doi: 10.1016/j.neuron.2012.12.030.

Neuron link

Pubmed link