Purchase this article with an account.
A. J. Ahumada, Jr., N. J. Coletta, A. B. Watson; Spatial Integration and the Neural Transfer Function. Invest. Ophthalmol. Vis. Sci. 2010;51(13):5175.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The neural transfer function (NTF) can be defined as the filter needed after the optical transfer function to obtain the contrast sensitivity function. Some measurements of the NTF have been based on interferometrically generated gratings of constant size. Ahumada & Coletta (2009 OSA Fall Vis. Mtg.) found that estimates of speckle noise power rose at high frequencies and hypothesized the rise was the result of a decrease in the useful area of the signal at high frequencies. This possible contribution to the measured NTFs was described by Sekiguchi, Williams, & Brainard (1993b JOSA A). Here we quantify this qualitative argument.
Assuming the effective signal grating area is the region where the cone density supports sub-Nyquist reconstruction of the gratings, we used the cone density measurements of Curcio & Sloan (1992 Vis. Neurosci.) to predict the falloff in signal energy in the high spatial frequency region. We compare this with the falloff in grating sensitivity from (Williams, 1985 JOSA A) in the same region.
The Curcio and Sloan data predict that near 55 cpd, the signal energy will fall by about -0.55 dB per cycle. For the Williams (1985 JOSA A) observers DW and MD (with the grating diameter of 1.5 deg and the coherent background of 10%), the median slopes in that region were -0.63 and -0.51, respectively. Coletta and Sharma (1995 JOSA A) only collected comparable data at 32 and 64 cpd. The slopes for observers NC and VS were -0.35 and -0.22 dB/cycle.
These results support the hypothesis of Banks, Geisler, & Bennnett (1987 Vis. Res.) that the neural component of the NTF passes cone signals without further high frequency attenuation. However, Sekiguchi, et al. and He & MacLeod (1996 JOSA A) have found additional falloff when grating signals of only a few cycles are used. Also, Watson & Ahumada (2008 J. Vis.) found that some (but not all) observers in visual acuity tests needed an NTF low-pass filter component. Perhaps there are other mechanisms, such as the temporal integration mechanism, that do poorly when the stimulus is small.
This PDF is available to Subscribers Only