Document Type



This study examined whether humans would learn to solve simple problems by applying an efficient rule, that is, concept teaming. Twenty-four college students were shown a series of five letter combinations created from the first three letters of the alphabet (A, B, and C), presented on a computer terminal (e.g., ACAAA). Pressing computer keyboard keys changed the letters of the alphabet. When all five letters were identical (e.g., AAAAA), the computer informed the subject that they were correct. The most efficient solution to each problem (the rule) was to change only letters that did not match the majority of the letters presented. Forty-five percent of the subjects failed to apply an efficient rule. During each of the 36 trials, regardless of the letters presented at trial onset, they made all the letters A (or B or C) and reported that their strategy was necessary in order to complete the problems. Two hundred and nineteen other students were employed in ten variations on the basic paradigm. Across conditions, between 20%-85% of the students failed to respond according to an efficient rule. The results are discussed in terms of response stereotypy and superstitious response chains that prevent the emergence of efficient concept learning. Some implications for animal research and educational philosophy are also addressed.