The recognition of single randomly generated, "meaningless" patterns has been studied during the past few years by a number of investigators (Attneave, 1957; Crook, Gray, Hanson, & Weisz, 1959; Vanderplas & Garvin, 1959b). Several experiments have further examined the abilities of human 5s in recognizing such patterns over certain simple variations, such as random noise (Crook et al., 1959; Hillix, 1960),contour noise (Fitts & Leonard, 1957), and systematically continued transformations (LaBerge & Lawrence, 1957). But no experiments have been reported in which 5s were asked to learn sets of variants of a pattern through experience with individual examples of these sets. Yet this is the typical procedure for computer "pattern recognition" simulation programs (Selfridge & Neisser, 1960; Uhr, 1962). It also presents a good experimental paradigm for the learning of patterns by human beings. Given N patterns and n variants of each pattern, 5 is presented with each of the nN particular pattern instances, asked to give it the correct name of the N names, and then told the correct name. This, in more systematic form, is what the child experiences during the natural experiments of his dayby-day learning. That is, the child is given particular instances of such things as "chair," "dog," the letter "A," or his mother's face; and he learns to build up general concepts, or pattern classes. This sort of experiment, then, should serve as a convenient method for studying concept formation and perceptual learning of patterns in human 5s. It should also give a convenient framework within which to examine the predictive fit of pattern recognition simulations by computer programs, when they purport to be embodiments of theoretical models of human form perception. The use of meaningless patterns for stimulus sets reduces the wealth of information that the human can bring to bear. It also allows for control and systematic variation of the parameters of the stimulus sets. The main experiment examined the learning of pattern sets as a function of (a) interrelations between patterns, (&) practice, and (c) individual differences between 5s (including the simulated model as an 5). Subsidiary experiments examined the effects of different (a) durations of stimulus presentation, (&) size of pattern set, and (c) complexity of pattern. In addition, three form perception
[1]
Oliver G. Selfridge,et al.
Pattern recognition by machine
,
1960
.
[2]
Leonard Uhr,et al.
A pattern recognition program that generates, evaluates, and adjusts its own operators
,
1961,
IRE-AIEE-ACM '61 (Western).
[3]
F. Attneave,et al.
Transfer of experience with a class-schema to identification-learning of patterns and shapes.
,
1957,
Journal of experimental psychology.
[4]
Hillix Wa.
Visual pattern identification as a function of fill and distortion.
,
1960
.
[5]
J. A. Leonard,et al.
Stimulus correlates of visual pattern recognition: a probability approach.
,
1956,
Journal of experimental psychology.
[6]
Wilbur H. Highleyman,et al.
A Note on Optimum Pattern Recognition Systems
,
1961,
IRE Trans. Electron. Comput..
[7]
Two Methods for Generating Matrices of Forms of Graded Similarity
,
1957
.
[8]
J. M. Vanderplas,et al.
Complexity, association value, and practice as factors in shape recognition following paired associates training.
,
1959,
Journal of experimental psychology.
[9]
L UHR,et al.
"Pattern recognition" computers as models for form perception.
,
1963
.
[10]
J M VANDERPLAS,et al.
The association value of random shapes.
,
1959,
Journal of experimental psychology.
[11]
Leonard Uhr,et al.
A Computer Simulation of Pattern Perception and Concept Formation
,
1962
.