Aerosol concentrations: effect on planetary temperatures.

independent variables yielded statistically significant effects. The interaction between masking condition and number of target segments was also significant (P < .01); this interaction reflects the decreased detectability of the multisegmented targets and their decreased susceptibility to masking. The reliability of the backward enhancement effect was assessed by simple t-tests, comparing the mean ratios of masking to nonmasking thresholds with the hypothetical ratio of 1.00 for the 16-segment target under both adaptingfield conditions and for the 8-segment target under the light adapting-field condition only. Enhancement would be indicated where these ratios had values significantly less than 1.00. While the mean ratios associated with the 16-segment targets were both statistically significant (P < .02), the ratio associated with the 8-segment target was not (P = .10). As a further test of the reliability of the enhancement effect for the 16-segment target, we combined data from the present study for the light adapting-field condition with that from the preceding one (9). Of the 12 subjects, 10 had ratios less than 1.00, a distribution which departs significantly from chance as assessed by a simple sign test (P < .05). We are by now fairly confident in the reality of backward enhancement. How might it be interpreted? We favor at this point the following explanation. Any target generates both excitatory and inhibitory effects. Some targets, such as those with many internal contours, generate more inhibition than others, perhaps through the mechanism uncovered by Hubel and Weisel (10) in their electrophysiological investigations of retinal receptive fields. For such targets, the masking stimulus serves as a disinhibitor (analogous to the function of the second mask in the case of target recovery). That is, the mask acts primarily to reduce, or transform, the target-generated inhibition, leaving the excitatory component dominant and thereby yielding "backward enhancement." For simple targets, which generate little inhibition (relative to excitation), the mask serves primarily to reduce or transform target-generated excitation, thereby yielding the conventional backward masking effect. WILLIAM N. DEMBER MARY STEFL Department of Psychology, University of Cincinnati, Cincinnati, Ohio 45221