Donald M. Johnson

SERIAL ANALYSIS OF THINKING

As the other papers in this monograph demonstrate, there are many ways to study thinking. I n large part, it seems that we are trying to understand what happens during those brief episodes of concentrated intellectual activity that occur when some kind of problem interrupts the routine activities of ordinary life. I t is not altogether necessary, of course, to analyze such a problem-solving episode. There is much to be learned by treating it as an undifferentiated whole and using over-all measures of outcome, such as type of solution and time required, for comparison with antecedent variables. Attempts a t analysis usually promise some additional information not obtainable from these conventional measures and should be evaluated in respect to this promise. The most familiar analysis in the past was that by John Dewey (1910), who analyzed reflective thought into five steps, preparatory to explaining to schoolteachers how thought may be studied. Then Graham Wallas (1926) published a similar analysis reducing thought to four stages, of which “preparation” and “incubation” are perhaps the best known today. In later research on thought processes investigators have requested the thinkers to talk as they worked and then tried to classify the verbal output in categories of functional significance. Whereas the early writers seemed to assume that the real thought processes would be revealed by the introspections of an astute psychologist, current methodology views such research as attempts to demonstrate the consistency and utility of a coding system (Johnson, 1955). At first the experimenter simply watched the movements of his subjects and reported that he observed trial and error, insight, or perseveration. More recently the rules for identifying a response sequence have been written in advance, and the frequency of such a sequence compared with chance frequencies or with frequencies under other conditions. It is not common in experimental psychology today to attempt a complete description of problem-solving activities; rather the experiments are usually planned so that only those patterns of activity pertinent to a particular hypothesis are observed, and such patterns are not called thought processes because this research stems from a different tradition. However, in a broader sense the aim is the same, that is, to describe problem-solving activities in functional units that are larger than the single response and smaller than the whole problem-solving episode. A sophisticated example of this approach is the study of concept attainment by Bruner et al. (1956) in which the subjects selected cards one after another in an attempt to discover the concept that the experimenter considered to be correct. From the mathematics of the case and the limitations of human memory, the investigators were able to describe certain ideal strategies, such as focussing and scanning, and then to observe how frequently such strategies were followed under different conditions and with what success. The appara-

Experimental Analysis of the Reliability of Observations of Social Behavior

(1952). Experimental Analysis of the Reliability of Observations of Social Behavior. The Journal of Social Psychology: Vol. 35, No. 2, pp. 151-160.

Increasing Originality on Essay Examinations in Psychology.

Most teachers would agree that education for original thinking is worthwhile. Two experiments reported here show how it may be accomplished within the class setting.

Word frequency and verbal comparisons

The importance of word frequency in verbal performance is usually attributed to its effect on the evocation of words, but it may also have an effect on the comparison of words. One experiment involved simple comparisons of 5-letter groups and the other involved evaluation of 3-letter clues to 5-letter anagrams. In both experiments higher scores were made on high-frequency lists than low. Verbal comparison can be added to the list of verbal operations influenced by word frequency.

Concept acquisition as a function of frequency of occurrence in verbal context

The experiment investigated concept acquisition as a function of number of occurrences of an unknown word in the context of a short story. Each of the nine stories was presented in one of six conditions (0, 1, 2, 4, 8, 12 occurrences of the word) to 120 college students. Five-item multiple-choice tests constructed to test concept acquisition showed adequate reliability, and for each word improvement was significant at the.01 level. Hence the learning of concepts from verbal context can be studied systematically, and the improvement, as in other learning experiments, appears to be a decreasing function of practice.