Different predictors of memory scanning with unidimensional and digit stimuli

Distribution of visual attention over space.

“…He then sampled colors from this explicitly defined space to form various memory subsets. His results are the two-dimensional analog of those of Levy et al (1980). Subjects were (a) faster in classifying test colors as members of a memory subset when that subset formed a compact rather than a dispersed region in the underlying color space and (b) faster in classifying test colors as nonmembers of a memory subset when those test colors were further from the memory subjects.…”

“…More recent studies of the classification of stimuli differing in color (Fish, 1981;Levy, Goldberg, & Schmid, 1980) or of symbols differing in meaning (e.g., DeRosa & Morin, 1970; also see Hutchinson & Lockhead, 1977;Shepard, Kilpatric, & Cunningham, 1975) have shown that spatial proximity operates in such abstract or semantic spaces to determine response times as well as errors. Thus Levy et al (1980) presented memory subsets of gray Munsell chips that were drawn from a total of ten chips differing in equal steps along the continuum of lightness only. They then presented a probe chip of a certain lightness and the subject was to classify that lightness as having been or not having been among the values represented in the memory set.…”

“…Even when discrete symbols are used as stimuli, there is evidence that perceived relations of proximity among the referents of the symbols in the designated subset influence performance. In the case of digits, for example, subjects' discriminative reactions are generally faster if the digits in the memory set are consecutive (DeRosa & Morin, 1970) or all even (Rosenbaum, 1974), and subjects are slower to classify a digit as not being a member of the memory set the closer this digit is in magnitude to a member of the set (DeRosa & Morin, 1970;Levy et al, 1980;Morin, DeRosa, & Stultz, 1967). These findings suggest that discrete, symbolic stimuli are internally anchored to positions in an underlying, more continuous semantic space so that, for example, the representation of a digit implicitly entails its meaning and, hence, its semantic relatedness to other digits (Shepard et al, 1975).…”

“…The amount of excitation produced by a probe in a neighboring detector should be inversely related to the distance between the probe and the detector. There are reasons to suppose that the function relating measured discrimination time to distance between the objects to be discriminated (e.g., Henmon, 1906;Moyer & Landauer, 1967) or between a test object and the discrimination boundary (e.g., DeRosa & Morin, 1970;Fish, 1981;Levy et al, 1980) becomes asymptotically flat for large distances and, indeed, may approximate a reciprocal (hyperbolic) function (e.g., Curtis, Paulos, & Rule, 1973;Shepard et al, 1975;Shepard, Note 1). However, as a simpler first approximation, we suppose here that at least over a suitably limited range, the theoretically postulated excitation in detector i can be taken as falling off roughly linearly with its Euclidean distance, 'd\., from the probed square: e\ = bad.…”

How a cognitive psychologist came to seek universal laws

“…In the case of categorization learning and performance, generalization and discrimination time both fall off according to the same sorts of (exponential and reciprocal) functions of distance of a stimulus from the category boundary in the representational space (Shepard,198Ia). (Concerning dependence of reaction time on distance from category boundary, see, e.g., De Rosa & Morin, 1970;Levy, Goldberg, & Schmid, 1980. Concerning generalization in categorization, see Nosofsky, 1987Nosofsky, , 1989Nosofsky, , 1992Shepard & Chang, 1963;Shepard & Kannappan, 1991;Tenenbaum & Griffiths, 200 I a, 200 Ib.…”