A Theory of Visual Thresholds

• Published in: Nature (London) 1958-01, Vol.181 (4621), p.1488-1488
• Main Authors: GREGORY, R. L., CANE, VIOLET R.
• Format: Article
• Language: English
• Subjects: Humanities and Social Sciences; letter; multidisciplinary; Science; Science (multidisciplinary)
• ISSN: 0028-0836; 1476-4687
• DOI: 10.1038/1811488a0

IN our first communication 1 on this subject we did not consider the effect of changing shape of the distribution of the neural impulse-rate, r , with consequent change in variance, as the photic intensity, I , decreases. We have, however, considered this in later papers 2,3 . Marriott's prediction from our original formulation that Δ I would become negative for stimulus conditions giving a sufficiently small percentage of positive responses depends on the distribution being normal. It is quite clear, however, that the distribution must change at low intensities, for a negative pulse-rate is impossible, and so it must become asymmetrical for low values of r . Now when we consider threshold conditions for various areas of stimulation, the pulserate for a given fibre must rise as the area, A , is reduced, if the threshold is to be maintained. But since fewer fibres are stimulated, the total number of impulses in unit time might not rise. We are inclined to think that for human central vision, at least, detection is based on the average pulse-rate. One of our reasons for holding this view is that in detecting a signal, Δ I , against a background, I , increases in the area of the background produce a lower threshold for discriminating the Δ I field 4 . It should be noted that this result does not hold for short flashes of Δ I . In this case (and most threshold determinations are made using short flashes) the background area has no effect, suggesting that the discrimination for short flashes of the signal, as opposed to continuous viewing, is a matter of detecting changes over time rather than over space. The experimental finding, that increase in background area, for continuous viewing, produces a reduction in the signal intensity (Δ I ) required, suggests to us that no mechanism such as retinal facilitation will account for areal summation. It looks as though perception involves selecting relevant areas in the stimulus field. For example, if a narrow black ring is added to the background in the differential threshold situation, so that the signal lies at its centre, the threshold rises, and is the same as though the part of the background lying outside the ring were not there 4,5 .