Evidence for independent parallel channels in tachistoscopic perception

Redundancy of noise elements and signals in visual detection of letters

“…On the average, the greater the number of different noise letters present in a display, the greater will be the likelihood that an imperfectly perceived noise letter will activate a sufficiently large subset of the feature detectors associated with a target letter to yield a false detection. This decision problem would be the primary locus of the predicted effect of noise redundancy in the models of Gardner (1973) and Shiffrin and Geisler (1973), but would be subsidiary to the hypothesized perceptual *This research was supported in part by USPHS Grant GM16735 from the National Institute of General Medical Sciences. I wish to thank Edith Skaar for her substantial contribution to the data collection and processing and Richard M. Shiffrin and Ronald Kinchla for a number of valuable suggestions concerning the theoretical interpretations.…”

“…For the models of Estes (1972), Gardner (1973), andShiffrin andGeisler (1973), all of which assume that inputs from the different noise letters of a display affect the decision process independently, this supposition is correct only if all noise letters are equally confusable with the signal letters. It can be shown that, if the noise letters differ in confusability , these models imply a facilitative effect of noise redundancy upon performance.…”

“…On the basis of several current models (Estes, 1972;Gardner, 1973;Shiffrin & Geisler, 1973), it must be predicted that if other factors are adequately controlled, the two conditions will yield similar results. Therefore, in the detailed design and analysis of the experiment, a primary focus of attention must be on the power of the test that is provided.…”

“…Multiple instances of a signal convey redundant information, but also they introduce physical similarity. I have assumed that signal redundancy exerts its effects primarily at the decision level, additional instances of a signal letter simply providing additional opportunities for identification (Estes, 1972), and it seems that a similar interpretation would follow from the models of Gardner (1973) and Shiffrin and Geisler (1973). But another possibility is that multiple instances of a signal might be mutually facilitating at the perceptual level as a consequence of their physical similarity.…”

Good continuation affects visual detection

“…This is generally referred to as the display-size effect. Second, performance declines as the similarity between target and noise (nontarget) items in the array increases (Estes, 1972;Gardner, 1973;McIntyre, Fox, & Neale, 1970). Third, performance improves as similarity among noise items increases (Estes, 1974;McIntyre et al 1970;McLaughlin, Masterson, & Herrmann, 1972).…”