Chapter 8: Spatial vision
Fundamental functions (pp. 215—226)
- What are the perceptual advantages of using long wavelength light (red region of the spectrum) for darkroom photo processing?
- Why is it easier for the dark-adapted eye to detect faint stars when they are not looked at directly?
- What happens to the size of visual receptive fields during dark adaptation?
What you need to know
- Photopic and Scotopic Vision (pp. 215—216)
- Dark adaptation
- The Purkinje shift
- Spatial Contrast Sensitivity (pp. 217—221)
- Stimuli: contrast, spatial frequency, orientation, and phase
- Spatial contrast sensitivity function
- Origins: optical limitations and neural responses
- Temporal Contrast Sensitivity (pp. 221—222)
- Stimuli: contrast and temporal frequency
- Temporal contrast sensitivity function
- Origins: neural responses
- Spatiotemporal Sensitivity (pp. 223—226)
- Spatiotemporal contrast sensitivity functions
- Origins: receptive field organisation and parallel pathways
Photopic and Scotopic Vision
During dark adaptation, both rods and cones become more sensitive to light, but they do so at different rates. Cones show a slight, rapid (10 minute) increase in sensitivity relative to slower (35 minute), more extensive changes made by rods.
The dark adaptation function that describes the shift from photopic to scotopic vision has a characteristic scalloped shape, which reflects the different contributions from the two photoreceptor systems.
The peak spectral sensitivity for cones occurs at a higher wavelength than for rods (555 nm vs. 507 nm). During dark adaptation the Purkinje shift occurs: the wavelength that appears brightest shifts from 555 nm in photopic vision, to 507 nm in scotopic vision.
One advantage of the Purkinje shift is that long wavelength light (red) is visible to the cones but not the rods of the dark-adapted eye. Acuity can be enhanced without compromising the superior sensitivity of dark-adapted rods.
Spatial Contrast Sensitivity
In this image, spatial frequency increases from left to right, while contrast increases from top to bottom. The shape of the visible, lower part of the image gives an indication of our relative sensitivity to different spatial frequencies; medium frequencies are visible at the lowest contrast (higher up the image). Spatial sensitivity for stationary gratings is band-pass. This image is known as a Pelli-Robson Chart
Spatial features of an image describe changes in image intensity across space.
The luminous intensity of a sinusoidal luminance grating varies repeatedly across space according to a trigonometric sine function. The resulting bright/dark bars have the following spatial properties:
- Contrast—related to the difference between the maximum and minimum luminance intensities at the peaks and troughs of the grating.
- Spatial frequency—the number of luminance cycles the grating repeats in one degree of visual angle (cpd). Since visual angle changes with viewing distance, so too does spatial frequency.
- Orientation—the tilt angle of the bars in a clockwise direction from vertical.
- Phase—describes which part of the grating cycle is aligned to a fixed reference point. One grating cycle corresponds to 360°, so a grating that is one-quarter of the way through its cycle at the point of reference has a phase angle of 90°.
This demonstration allows you to study how a grating's four defining parameters affect its appearance. Use the sliders to vary frequency, contrast, orientation, and phase.
The spatial contrast sensitivity function (CSF) established by Campbell and Robson (1968; see FP p. 219) shows that the contrast required for detection of a grating (spatial contrast threshold) depends on its spatial frequency. Spatial contrast sensitivity peaks at 3 cpd (1 cpd for scotopic vision), and declines more rapidly at higher than at lower spatial frequencies. Frequencies higher than 40 cpd (8 cpd scotopic) are undetectable even at maximum contrast.
The shape of the contrast sensitivity function is determined by:
- Optical limitations of the eye—as light is transmitted to the retina, diffraction and aberration introduce image blur, reducing the ability to resolve the bars of a grating. The optical transfer function (OTF) (Fig. 8.4; see FP p. 220), plotting contrast ratio as a function of spatial frequency, shows that the eye attenuates most contrast at high spatial frequencies, and accounts for the upper frequency limit of the CSF.
- Properties of visual neurons—neurons with large receptive fields respond to low spatial frequencies, and small receptive fields respond to high spatial frequencies (Fig. 8.5; see FP p. 221). The shape of the CSF reflects the quantity and responsiveness of differently sized receptive fields.
- Contrast sensitivity declines at low and high spatial frequencies because fewer, less responsive neurons have large- and small-sized receptive fields.
- Peak contrast sensitivity shifts to lower spatial frequencies in scotopic vision because larger receptive fields are employed.
Temporal Contrast Sensitivity
Temporal features of an image describe changes in image intensity over time.
A light that alternates repeatedly over time, between bright and dark intensities, has the following temporal properties:
- Contrast—related to the difference between the maximum and minimum alternating intensities.
- Temporal frequency—describes the number of flicker cycles that repeat in one second (hertz). The sequence for one flicker cycle is one bright light, one dark light.
This demonstration allows you to study how a flickering spot's temporal frequency and contrast affect its appearance. Use the slider to vary both. Note that the period in seconds (indicated by slider position) corresponds to (1/temporal frequency) in hertz or cycles per second.
When temporal contrast sensitivity is measured at a range of flicker rates, the results can be plotted in a temporal contrast sensitivity function. The temporal CSF shows that the contrast required for detection of a flickering light depends on its temporal frequency. In photopic vision, temporal contrast sensitivity peaks at 8 Hz, and declines more rapidly at higher than at lower temporal frequencies. Frequencies higher than 50 Hz are undetectable even at maximum contrast.
The shape of the temporal contrast sensitivity function is determined by neural factors. The different temporal responses of rods and cones feed retinal ganglion cells, which begin complex temporal interactions between the LGN and cortex.
Spatiotemporal features of an image describe changes in image intensity across space and over time.
A grating that alternates repeatedly in luminous intensity, so that bright bars become dark as dark bars become bright, has both spatial and temporal periodicity. This can be shown in a space-time plot.
Spatiotemporal contrast sensitivity is measured with a flickering grating. The spatiotemporal contrast sensitivity function shows that the contrast required for detection of a flickering grating depends on the interaction between its spatial frequency and its temporal frequency. At low temporal frequencies, spatial contrast sensitivity is band-pass (most sensitive to a band of mid-range spatial frequencies). At high temporal frequencies, spatial contrast sensitivity is low-pass (most sensitive to low spatial frequencies).
This image is identical to the image in Demonstration 8.2, except that the grating contrast reverses rapidly to generate flicker. Notice that medium spatial frequencies (middle-right of the image) are much less visible when the pattern flickers (use the control to pause the animation); spatial frequency sensitivity becomes low-pass at high temporal frequencies.
There are two alternative neural explanations for the shape of the spatiotemporal contrast sensitivity function:
- Receptive field organisation—the balance between the excitatory and inhibitory influences of centre-surround receptive fields changes with temporal frequency. At low temporal frequencies, centre and surround have equal influence. The net response of the receptive field is the difference between the low-pass centre and the low-pass surround, which is a band-pass spatial frequency response (Fig. 8.9b; see FP p. 224). At high temporal frequencies, the centre has more influence than the surround. The difference between the dominant low-pass centre and the weak low-pass surround is now a net low-pass spatial frequency response. Derrington and Lennie (1984; see FP p. 225) reported that the spatiotemporal sensitivity of both parvo and magno primate LGN cells was consistent with this idea.
- Parallel pathways—two separate parallel visual pathways, with different spatiotemporal properties, contribute to early visual processing:
- A sustained channel, most sensitive to high spatial frequencies and low temporal frequencies, and
- A transient channel, most sensitive to low spatial frequencies and high temporal frequencies.
The spatiotemporal properties of sustained and transient channels, revealed by psychophysical experiments, are consistent with cell responses in the parvo and magno divisions of the visual pathway, respectively.
Similar to certain clinical conditions in humans, Merigan and Eskin (1986; see FP p. 225) have shown that selective damage to monkey parvo cells reduces sensitivity to gratings with a combination of high spatial and low temporal frequencies.
The dominant view is that parallel pathways provide the best account of spatiotemporal sensitivity, but it is likely that the visual system uses both cell divisions together, rather than switching between them.
So What Does This Mean?
The transfer from cone to rod vision during dark adaptation takes approximately 30 minutes and is accompanied by a shift in peak spectral sensitivity to shorter wavelengths.
Contrast sensitivity functions form the foundations of our knowledge about spatial and temporal visual representations. The spatial CSF reflects the optical limitations of the eye, and the quantity and responsiveness of differently sized receptive fields. The spatiotemporal CSF indicates that early visual processing involves separate channels, with spatiotemporal responses that are consistent with those of cells in the parvo and magno divisions of the visual pathway.