Carl P. Duncan

Retrieval of low-frequency words from mixed lists

In two experiments Ss were presented either a list of high- or a list of low-frequency words (unmixed lists), or lists containing equal numbers of words from the two frequency levels (mixed lists). In recall there was a significant interaction between level of frequency and type of list: In unmixed lists high-frequency words were better recalled; in mixed lists low-frequency words were better recalled.

Anagram solving as a function of bigram frequency

Summary In three experiments on anagram solving, the variables were high vs. low total bigram frequency both of the anagrams and the solution words. In two of the experiments, spacing between letters of the anagrams was varied. Subjects were run in groups with all anagrams presented at once in two experiments and individually with one anagram at a time in the third experiment. The time allowed for solving was varied from experiment to experiment. With the exception that more anagrams were solved as more time was alowed, no main variable had any consistent effect across experiments. The consistent finding over all three experiments was a highly significant interaction between bigram frequency total of anagram and word. Anagrams with low totals were solved more easily for words with low than for words with high totals. Anagrams with high totals were solved more easily for words with high than with low totals. Certain interpretations of this interaction were discussed.

ATTEMPTS TO INFLUENCE PERFORMANCE ON AN INSIGHT PROBLEM

Within an S-R approach to problem solving, an insight problem may be defined as one in which solution responses are low in the hierarchies of responses elicited by problem stimuli, i.e., have initially low probability of occurrence (Maltzman, 1955). As will be seen, such problems seem to be very little influenced by prior training techniques where only nonspecific transfer is possible, i.e., where there is no specific stimulus or response similarity between the training procedure and the transfer task (problem). [Specific transfer is another matter (see an example in Saugstad, 1957).] The present paper reports a number of experiments in which three types of nonspecific training were used prior to a particular insight problem.

Thinking of a word under different retrieval constraints

Subjects were given, as stimuli, either letters or bigrams and asked to think of a word beginning with those stimuli. Words emitted were to be 3, 4, or 5 letters in length. Each of the six groups contained 100 Ss. Subjects were given 60 sec per stimulus to produce a word. Response measures were the number of words emitted, latency of response, and language frequency of emitted words. It was found that responses were fewer, latencies were longer, and word frequencies were lower for bigram than for letter stimuli, and for longer vs. shorter word lengths. It was pointed out that the effective pool of acceptable words is smaller for bigram than for letter stimuli and is probably smaller for longer than for shorter words. The findings with all response measures were interpreted leager terms of a greater difficulty in thinking of words as pool size is reduced.

Effect of word frequency on thinking of a word

Three experiments were performed to determine the effect of word frequency on thinking of a word. In all experiments Ss were given certain characteristics of words, such as the first letter, the last letter, the number of letters, or the class of the object named by the word, and were asked to think of a word fitting the given characteristics. In all experiments, more words of higher frequency were emitted than were words of lower frequency. In general, the tendency to emit words of higher relative frequencies held throughout the whole range of absolute frequencies. Not all high-frequency words that could have been emitted were emitted. The findings were interpreted in terms of a combination of the spew hypothesis, and of incomplete sampling from a population of words.

Mediation in verbal concept learning

Summary In a study of mediation in verbal concept learning, words that are members of known classes, and words that could not readily be classed with other task words, were each used both as relevant and as irrelevant stimuli. These four conditions were studied in three experiments that varied in the number of relevant and irrelevant words presented. The results were the same, on the whole, in all three experiments, both in terms of trials to learn, and with an error measure assumed to be sensitive to mediation. Among relevant words, class words were more rapidly learned, and showed more evidence of mediation, than nonclass words. Among irrelevant words, there was no difference, as a main effect, between class vs. nonclass words. There was a significant interaction such that when class words were relevant, performance was usually better with nonclass irrelevant words, but when nonclass words were relevant, performance was better with irrelevant class words. It was suggested that there was evidence for mediation of response when class words were positively reinforced, and perhaps for mediation of nonresponse when class words were negatively reinforced.

Storage and retrieval of low-frequency words*

Retneval of words as a function of their language frequency was studied by having Ss attempt to recogruze the words, recall the words after one presentation, or produce (think of) the words from their initial bigrams. It was found that one reason many low-frequency words could not be thought of (often necessary in anagram and other problem-solving tasks) was because they were not stored by S. as measured by failure to recognize them as words. Those low-frequency words that were stored were more difficult to retrieve than high-frequency words, both in production and in recall. High-frequency words did not exhibit failure of storage. but showed considerable difficulty in retrieval. both in recall and in production.

PROBABILITY VS. LATENCY OF SOLUTION OF AN INSIGHT PROBLEM

In a study of anagram solving, Mayzner and Tresselt (1958) plotted curves showing probability (cumulative frequency) of solution of anagrams as a function of log time (latency) of solution. For each of eight different conditions, the empirical points were rather well fitted by a straight line. The authors then used these linear functions in a commendable attempt to infer something about problem-solving precesses (see later). Mayzner and Tresselts paper suggested testing whether log probability plots would also be linear for another problem. From a recent study (Duncan, 1961) there was available the solution times of an insight problem (pendulum solution of Maier two-string problem) for a large number of Ss. Since the independent variables had little or no effect in any of the several experiments, the data of all Ss were used to plot Fig. 1. In preparing Fig. 1 a number of arbitrary decisions were made. First, the base N used to compute the cumulative percentage was the total number of Ss ( 5 2 8 men, 560 women) who attempred ro solve the problem. Since about 94% (498) of the men and about 72% (405 ) of the women solved the problem in the 15 min. allowed, the plotted points rise only to these percentages, thereby correctly representing probability of occurrence of solution under the particular experimental conditions used. Second, the log times, not the raw times, were grouped. The step-interval was .1 log unit, and the intervals were equally spaced on the baseline; these procedures avoid the clustering of points at the high end of the curve often seen in log probability plots. Third, the straight lines were drawn only by visual inspecrion. The straight lines in Fig. 1 may not be, although it is hard to judge, as good fits of the data as are the lines drawn by Mayzner and Tresselr for their data (their Fig. 1, p. 378). There are some differences between their graph and the present one. Each of Mayzner and Tresselts curves was based on 100 solution times obtained by taking five times from each of 20 Ss, thus combining latencies within Ss and latencies between Ss, whereas the present Fig. 1 is based on a single latency from each of a large number of Ss (the data were grouped in both studies). Perhaps within Ss latencies are more homogeneous than .completely independent latencies. Also, Mayzner and Tresselts curves

Recognition of names of eminent psychologists.

Faculty members, graduate students, undergraduate majors, and introductory psychology students checked those names they recognized in the list of 228 deceased psychologists, rated for eminence, provided by Annin, Boring, and Watson. Mean percentage recognition was less than 50% for the 128 American psychologists, and less than 25% for the 100 foreign psychologists, by the faculty subjects. The other three groups of subjects gave even lower recognition scores. Recognition was probably also influenced by recency; median year of death of the American psychologists was 1955, of the foreign psychologists, 1943. High recognition (defined as recognition by 80% or more of the faculty group) was achieved by only 34 psychologists, almost all of them American. These highly recognized psychologists also had high eminence ratings, but there was an equal number of psychologists with high eminence ratings that were poorly recognized.

Syntactical encoding in short-term memory

Subjects learned and recalled lists of nonsense words presented without sentence structure, or the same words embedded in key syntactical positions in active, passive, or interrogative voice sentences, for three trials. On Trial 4, all S s learned and recalled nonsense words presented in an active voice snetence. In all conditions (active, passive, interrogative, word list) recall decreased, PI (proactive inhibition) increased, from Trial 1 to Trial 3. On Trial 4, the group continuing with active voice showed further PI. The word list group showed a high degree of recovery from PI. The passive and interrogative groups showed intermediate amounts of recovery from PI. The recovery from PI when sentence structure was introduced, or was changed from one voice to another, is taken to indicate use of syntax as an encoding cue for recall.