Imagery and paired-associate recognition

Imagery and paired-associate recognition GLEN A. RASER and WA YNE H. BARTZ. Iowa State Unil'ersitl'. Ames. Iowa . Three groups of Ss (N = 60) given Imagery. Word._ or :Vo instructions were presented 40 paired-associates once and tested in a recognition procedure. The pairs were composed of nouns and/or their line drawings. Imagery instructions resulted in higher, word instruction lower, performance thall did the control instructions. Line drawings facilitated performance as response terms, but not as stimulus terms. The results were taken as support for a differentiation effect interpretation of imagery in paired-associate learning. Studies of paired-associate learning of nouns differing in tenns of rated image-evoking capacities have shown high imagery nouns to have greater effects among stimulus tenns than among response tenns (Paivio, Yuille, & Smythe, 1966; Paivio & Madigan, 1968). This effect is contrary to that found for stimulus materials rated on other measures (e.g., meaningfulness) and has led Paivio (1965) to offer the "conceptual peg" hypothesis. This hypothesis states that high imagery stimulus tenns easily evoke a sensory image which is then related to the response tenn and serves as a cue for retrieval of the response tenn. Thus concrete nouns are superior to abstract nouns (Paivio, 1965), and pictures, which provide images directly, have been shown to be superior to their verbal labels (Paivio & Yarmey, 19(6). Dominowski & Gadlin (1961) have suggested that, contrary to the "conceptual peg" hypothesis, imagery provides stimulus differentiation which, in turn, reduces intralist interference. It is possible to speculate that in the typical recall procedure, where stimulus tenns are presented as S recalls the response teons, imagery could provide a basis for stimulus differentiation. If this is the case, a recognition procedure might provide a basis for response d(fferentiation and the reported superiority of imagery for stimulus terms might not occur. This study was designed to investigate the role of imagery in paired-associate recognition. Differences in the imagery of stimulus and response tenns of pairs was established by using words and their drawings (Paivio & Yarmey, 1966) in all four possible pair combinations. To further investigate the role of imagery, Ss were instructed to use imagery during learning. As a comparison, other Ss were instructed to use the characteristics of words and others were not given specific associative instructions. SUBJECTS The ISO Ss were students enrolled in psychology courses at Iowa State University. Fifteen were assigned in a random fashion to each of 12 treatment combinations (3 Instructions by 4 List Types) amt were run in a group. STIMULUS MATERIALS The- initial item pool consisted of 160 monosyllabic nouns selected from Thorndike & Lorge (I944} which could be represcnted by a simple ·line drawing. Each word and its line drawing was separately reproduced in black ink on a 3 x 5 card. [""fty basic word pairs were then randomly selected and 35 mm nq~ative slid.:s were made of each of the four pair combinations of words and pictures tWord-Word, Word-Picture, Picture-Word, and Pichlrl'-Picturc). Slid.:s for the recognition task were prepared by photographing the stimulus term alone for eadi pair. The response member of each pair was photograpned in a set of seven like items. Thc·six incorrl'l"t aUernatives included two response terms and two stimuJus tcrms fmm other pairs and two items randomly seleckd from the n'm.tining pool uf items. A complete set of the test slides was prepared lIsing the printed words as was one using the linc drawings. Ps}'chon. Sci .• 1968. \ 01. 12 IS} INSTRUCTIONS One-third of the Ss were instmcted to find a unique mental image to relate members of a pair. Another third were instmcted to find a word or letter cue to aid in relating the pairs. Both uf these groups were given examples of the type o~ strategy they were to use. The remaining third were simply instructed to learn or relate the pairs. PROCEDURE A Kodak ROO Carousel slide projector wa, used to present the stimulus materials. After the group's particular instructions, the 40 pairs were presented for 8 sec each only once. (A pilot study had shown that 15 sec presentation time resulted in extremely high levels of recognition.) The test trial followed and consisted of presentation of the stimulus term of the pair for 2 sec and the response alternatives for 10 sec_ A -different random order was used in presenting the pairs during the test trial. S was to select and record his response on a data sheet during the 10 sec response slide interval. Ss were then asked to briefly describe, in writing, the mental image, word or letter cue, or any other method they used in learning each pair. RESULTS The recognition results are shown in· Fig. I. The Instructional conditions differed significantly, F(2,168) = 15.26 (p < .01). Tests among these means showed that Imagery instructions resulted in significantly higher performance than did either of the other two (p< _01); the No instructions condition resulted in higher perfonnance than did Word iRStructions (p < .05). The effects of list types was also significant, F(3,168) = 7.04 (p < .(1). Tests among these means were conducted by contrasting performance for list types as a function of the types of stimulus materials. used as the stimulus and response tenns. These tests showed that the degree of imagery was a factor among response terms, F(I,168)= 1-9.82 (p< .01); pairs with Pictures as response terms were better recognized than were those with Words. Imagery was not found to be a significant factor among stimulus terms as pairs with Words and those with Pictures as stimulus terms were recognized equally well (F < I). Figure 2 shows the types of errors made on each of tire List Types in terms of per cent total errors within List Type groups. The incorrect alternatives in the recognition task had been either response items, stimulus items, or words not in the lists. Stimulus and response errors were made in about the same proportion in all but the Word-Picture groups which showed a much higher incidence of selecting response term alternatives. Groups with Words as response terms made more extra-list errors as compared to groups with Pictures as response terms. The distribution of 40 o 0--0 IMAGE INSTR. _WORD IN&TR. .. -1-NO INSTR_ POp .... P P'-W LIST T'fPE w-w Fig. I. Mean correct recognitions as a function of List Type and Instructions. 385 70 ~ RESPONSI:. ERRORS -STIMULUS ~-o() EXTRA-LIST " . - .. NO RESPONSES '" ~60 II:: ::;50 ..J ~40 ...o ... 30 z .o-------() t: 20 II:: ~'O ---0--, ..... ... --- ___ ... _____ _----A 01--P~-=P------~W~-~P.-----~P~-~W~----~W~-WU- LIST TYPE Fig. 2. Per cent of total errors for each type of error as a function of List Type. types of errors among the Instructional conditions showed no large or consistent differences nor was there any in (...truncated)