Improvement in human vision under bright light: grain or gain?

Abstract

SUMMARY1. The factor by which increment threshold changes with changing background intensity is less if the test flash is small than if it is large. This is commonly attributed to a reduction of the area over which visual signals are integrated as light adaptation increases.2. We propose and test an alternative hypothesis that the change in slope is the result of purely local processes: if it is assumed that increasing the background intensity increases the exponent of the local response function, but does not alter the extent of spatial integration, then the threshold of the small test flash will rise more slowly than the threshold of the large test flash simply because the small test flash is of a higher intensity than the large and therefore evokes a correspondingly greater local response.3. We measured small and large test field increment thresholds and dichoptic brightness matches as a function of background intensity.4. The log-log slopes of the small and large field increment threshold functions differed by not more than about 20 %, suggesting that even under the conventional interpretation of such data, the change of spatial integration is less than is usually supposed.5. The intensity of a large (2-3 deg) suprathreshold test field matched to a standard in the other eye varies with increasing background intensity with the same shallow slope as the small test (2-6 min) threshold versus intensity function; this is in agreement with the predictions of the local non-linearity hypothesis and suggests that there is no substantial change in spatial integration during light adaptation.

Spatial integration and sensitivity changes in the human rod visual system.

“…Similar shifts were found by Chen et al (1987;see their Fig. 8) in two of their four subjects and have been noted by Whittle & Challands (1969).…”

“…Traditionally, this is attributed to neural spatial reorganization somewhere in the retina; being most frequently related to the increasing prominence with light adaptation of the inhibitory surround in retinal ganglion cells (Barlow, 1958). However, Chen et al (1987) have shown for cone vision, first that there is only a small change in TVI slope caused by changes in test flash size (the slopes of their small and large field increment threshold functions differ by not more than 20%; on average 097 vs. 08) -so that independent of any interpretation the change of spatial integration is less than is usually supposed -and second, that the change in slope can be mostly accounted for by local adaptation changes that involve no alteration in spatial organization with background intensity.…”

“…As well as arguing that spatial reorganization is not an important cause of the change in gradient of the cone-detected increment threshold curve, Chen et al (1987) suggested an alternative explanation: that it is due to an increase in the slope of a strictly local adaptation-dependent non-linearity relating log stimulus intensity to response. The threshold intensity for detecting a small flash is (much) higher than that for detecting a large one.…”

“…Nevertheless, the change in the effective integration is relatively small. In fact, Chen, MacLeod & Stockman (1987) have calculated that the reduction in integration area of (Gaussian-shaped) receptive field centres by surround inhibition is probably no more than 19 % (009 log1o unit).…”

“…To decide between this local hypothesis and the reduction in spatial integration hypothesis, Chen et al (1987) compared small and large flash thresholds and large flash suprathreshold dichoptic brightness matches, all measured as a function of background intensity. The large flash brightness matching was performed at the same stimulus intensities at which the small flash reaches threshold so as to produce the same high local response -but with a large flash.…”