Brightness Sensation and the Neural Coding of Light Intensity
Robert B. Barlow Jr. Institute for Sensory Research Syracuse University
We can see stars of the 8th magnitude and distinguish objects in bright sunlight: an operating range of more than 10 log units of light intensity. We can detect with exquisite sensitivity small changes in illumination throughout this enormous range ( Hecht, Peskin, & Patt, 1938), but can we judge the absolute level of illumination itself? That is, can the visual system perform as a photometer?
More than 40 years ago Spring and Stiles ( 1948) found that our pupil reflex was graded over at least 8 log units of light intensity. Their measurements were later extended by Alpern and Ohba ( 1972). Using the psychophysical technique of magnitude estimation, Stevens ( 1953) found that observers could scale the brightness of test flashes over about 4 log units of light intensity. Ten years later, Stevens and Stevens ( 1963) showed that the range of dark-adapted observers was at least 7 log units. Mansfield ( 1973), Barlow and Verrillo ( 1976), and Bolanowski ( 1987) confirmed and extended these earlier observations.
Studies of circadian rhythms provide further evidence for the photometric performance of a mammalian visual system. Phase shifts of the circadian rhythm in locomotor activity in hamsters are graded with light intensity over a 4-log unit range ( Takahashi, DeCoursey, Bauman, & Menaker, 1984). A similar intensity -- response relationship exists for the suppression of melatonin in hamsters ( Brainard, Richardson, King, Matthews , & Reiter, 1983). These responses are mediated by information transmitted from the retina to the circadian pacemaker in the suprachias-