Psychophysics of spatial frequency selectivity Flashcards
What is the importance of Fourier analysis for vision?
Psychophysical evidence suggests that the brain performs a crude form of Fourier analysis, extracting useful spatial information be encoding luminance variations at each SF.
What was believed to explain the shape of the CSF until the 1960s?
That the CSF reflected a single detection mechanism (neural process) which processed all spatial changes in the brain, assumed to be broadly tuned to all SFs (responding better to some hence inverted U).
How did the brain’s use of Fourier analysis affect the interpretation of the CSF?
It makes more sense, as proposed by Campbell and Robson (1968), that the CSF actually reflects the combined activity of many independent mechanisms.
What does CSF stand for?
Contrast sensitivity function.
What are other terms for the independent mechanisms which detect spatial changes in the brain?
Neural processes, filters, detectors, or channels.
How do the independent mechanisms forming the CSF behave like filters?
They are each maximally sensitive to only a narrow range of SFs.
What do ‘low’ and ‘high’ SF filters encode?
Low = course variations - large objects and overall shape. High = fine structure - small objects and details.
What is a ‘square wave’ grating?
Like a sinusoidal grating, but luminance changes abruptly over space.
What’s so special about square wave gratings?
They can be Fourier synthesised by summing a set of sinusoidal gratings of increasing frequency (e.g. f, 3f, 5f…) and decreasing contrast.
When sinusoidal gratings, starting with frequency f, are combined, what is the difference in contrast between f and the final square wave grating?
The contrast of f is 1.27x greater.
What does the single mechanism CSF model predict about the difference in contrast detection thresholds for square wave and sinusoidal gratings?
That overall contrast is the crucial factor, so a sinusoidal and square wave grating of frequency f should have the same detection threshold.
What does the multiple filters CSF model predict about the difference in contrast detection thresholds for square wave and sinusoidal gratings?
Detection is possible when the most sensitive filter exceeds its threshold, and as the square wave grating contains a higher contrast sinusoidal grating it should require less overall pattern contrast.
What did Campbell and Robson (1968) find?
That contrast detection thresholds for square-wave gratings were typically 1.27x lower than for sinusoidal gratings - they were easier to detect.
Also at threshold, the gratings were perceptually indistinguishable. Therefore only the largest sinusoidal component of the square wave grating was visible.
What supporting evidence for the multiple filters model of CSF was found by Graham and Nachimas (1971)?
They used compound gratings to show that peaks-subtract phases (lower contrast) were equally detectable as a peaks-add phase for the same sinusoidal gratings.
Is spatial frequency adaptation selective?
- prolonged viewing of a high contrast pattern makes subsequent patterns harder to detect.
- according to the single mechanism model, this should affect all SFs, according to multiple filters only similar SFs should be affected.
- Blakemore and Campbell (1969) found it was selective.
