Kirk M. Lunnen, Psychology
Understanding how different modes of informational presentation can affect memorability is of great importance from advertising to education and in other disciplines. In education, for example, it relates to important issues such as determining the most effective strategies for increasing students’ vocabularies. In advertising, it means the difference between product recognition (and consequent consumption) and product failure. The search for a method of presentation that can maximize this memorability has been the object of numerous researchers (Esser, Die, Seholm, & Pebley, 1979; Lippman & Shanahan, 1974; Bower, 1970; Paivio, 1969).
Several such studies have explored the possible differences between pictorial memory as opposed to textual (word) memory (Meyers-Levy, 1989; Esser eta!., 1980; Paivio & Csapo, 1969; Jenkins, Neale, & Deno, 1967). These studies and others indicate a superiority of recall for pictorial over textual stimuli. Other studies have evaluated recall for words in the context of a meaningful, simple sentence (Lippman eta!., 1973; Bower, 1970).
Of additional consideration are possible differences in the serial position effects (primacy and recency) between these alternate presentation styles. The primacy effect is defined as the tendency for the subject to disproportionately remember items at the beginning of a list in a serial learning task (Hall, 1989). Stein and Rain (1989) explain the primacy effect as the subject fixating on the beginning items which causes “retrospective confusion” on subsequent items (p. 137). The recency effect is similar to the primacy effect except the increased retention is for the items at the end of a list in a serial learning task. Many researchers posit that this phenomena is the result of the last few items in a serial recall task still being in the subject’s “working or active memory” (Paul & Whissell, 1992). A comprehensive analysis of these serial position effects is of great importance with regard to the memorability of novel stimuli, and (as it relates to these modes of presentation) is noticeably absent from the existent literature.
The purpose of this project was to evaluate all three of the above mentioned informational presentation styles and assess any group and gender differences that may exist. The researchers hypothesized that this study would replicate the previous findings showing the superiority of picture memorability over textual memorability. It was also hypothesized that pictorial memorability would prove significantly better than contextual memorability. In regard to the assessment by gender, it was expected that this experiment would prove analogous to Meyers-Levy (1989) who found that overall short term memory performance for males and females was equal.
The practical implications of this study include the following: 1) replication of previous results indicating pictorial superiority in memorability, 2) comparison of contextual presentation to pictorial and textual presentations, 3) replication of gender similarity in overall short term memory performance, and 4) analysis of interaction effects by group and by gender.
Method
Subjects
Participating in this project were 105 undergraduate introductory psychology students . This group consisted of 46 males and 59 females. The males average age was 19.75 with a standard deviation of 1.75. The females average age was 18.97 with a standard deviation of 1.40. The average overall educational level for the group was 13.49 years with a standard deviation of 0.99. Subjects volunteered to participate as an extra credit assignment in Introductory Psychology.
Materials
Twenty common items were selected from a popular children’s dictionary. This was done to ensure the identifiability of all the objects. Each item was carefully screened to delete any obvious associability with the other items. The items were then each formatted in three ways: in picture form, in word form, or in word form in the context of a simple descriptive sentence. The pictorial format consisted of simple black and white line drawings of the items. The textual format consisted of the item words printed in bold, uppercase letters. The contextual format consisted of simple descriptive sentences containing the item word oversized and italicized.
These sentences were formulated as per those used in Lippman & Shanahan’s (1973) study of pictorial facilitation. The three formats were then made into 35 mm slides with one slide for each item. Procedures
The subjects were randomly assigned to one of the three groups with as closely equalized gender representation as possible. The subjects were shown the items in a randomly selected order with 2.5 seconds viewing time for each item. Following the item presentation the subjects were required to complete a short demographic questionnaire. This was designed to minimize recency effects (Erdelyi & Becker, 1974). Free recall of the items for a period of 2.5 minutes followed the completion of the questionnaire. The recalled items were written by the subjects on an answer sheet provided.
Results
A two-way analysis of variance (ANOVA) showed no overall significant difference by gender (!..'(1,93) = 0.20, p > .05). However, a significant difference was found between groups (!..'(2,93) = 16.17, p < 0.01) with the scores on the contextual group being significantly lower than the scores on the other two groups. In addition, a significant interaction-by -group-and -gender-was discovered (-E(-2,93 )-=-3 .20,-p-< 0.05). Using Tukey’s Multiple Comparisons of Interaction Means the group and gender means were evaluated for significant differences. The approximate critical distance for Tukcy’s Multiple Comparison Procedure was found to be 2.87.
The multiple comparison procedure yielded the following information: I) the male contextual group scored significantly lower (p < 0.05) than the other two male groups and the female pictorial group, 2) the female contextual group scored significantly lower (p < 0.05) than the female pictorial group and the male textual group but not the female textual group.
In order to assess any serial position effects, the item list was divided into three sets: 1) First set was composed of the first six items; 2) Second set was composed of the seventh through the fourteenth items; and 3) Third set was composed of the last six items. Using a one-way AN OVA, the three sets where analyzed and found to be significantly different from one another (!..'(2,312) =32.19, p < 0.0001).
Discussion
Several studies have indicated that pictorial presentation of information promote superior memorability as compared to textual presentation (Meyers-Levy, 1989; Esser eta!., 1980; Bower, 1970; Paivio & Csapo, 1969; Jenkins eta!., 1967). These findings have been replicated on numerous other occasions (Paivio & Csapo, 1969). The fact that the results of this study seem to refute this notion, i.e. showing no significant difference between the pictorial and textual groups, appears to call into question the viability of this study. However, if the conclusions presented in Paivio & Csapo’s (1969) study as to why pictorial presentation is superior are considered, a possible reason for the disparity is discovered. They conclude that this superiority is explicable on the supposition that common-object pictures are stored both visually and in terms of the verbal labels they elicit and that memory for both will exceed memory for the verbal label alone (Paivio & Csapo, 1969).
However, in this same study it was found that the pictorial recall ability was adversely affected as the speed of item presentation was increased (Paivio & Csapo, 1969). They hypothesized that this increased exposure limited the pictorial coding to visual only, thus the subjects were unable to attach the verbal label as well (Paivio & Csapo, 1969). If one compares the time of individual item exposure in Paivio & Csapo’s (1969) fast and slow groups (0.1875 seconds and 0.5 seconds, respectively) to the 2.5 seconds of exposure for each item in the current experiment, the difference is quite dramatic. Following Paivio & Csapo’s (1969) line ofreasoning that the decreased time disallowed multiple coding, it is surmised that the relatively lengthy duration of exposure per item in this experiment was sufficient to allow for multiple coding of the both the pictorial presentation and the textual presentation. This resulted in statistically nonsignificant differences between the two groups.
In considering the significantly lower recall ability in the contextual group several possible conclusions are apparent. Bower (1970), among others, has shown the power of imagery as a memory enhancer. The simple sentences provided as the context for the items in this study were perhaps somewhat devoid of such imagery. In addition, and ironically, the very words included in the sentences to increase imagery content might in some cases proved more distracting and confusing than beneficial in the free recall process. This is evidenced by the fact that in multiple cases such extraneous words were written instead of the target words (e.g. “Captain” was written instead of the target item “Anchor”). Also, the increased number of words per item in the this group seemed to make multiple coding far more difficult in the time allotted.
As previously mentioned, the overall main effect by gender showed no significant difference between male and female performance. This corroborates the findings of Meyers-Levy (1989). However, the interaction effect showed several significant differences by gender. The differences between male and female scores in both the textual and contextual groups approached significance (p < 0.06). Interestingly enough, there is a reversal by gender between these two groups. This yields a significant difference between the male textual group score and male contextual group score while the female contextual group score stayed statistically comparable to the female textual group score. This finding corroborates Meyers-Levy’s (1989) conclusion that females are more able to activate a blending of hemispheric processing styles resulting in a more complete labeling of items in a contextual setting. This raises the question of different processing styles. The female subjects showed more consistency in recall ability between the textual and contextual groups while the males were more consistent between the textual and contextual groups. This leads to the conclusion that the female subject performance was less adversely affected by the limited ability for multiple encoding in the contextual group. Also interesting is the “reversal” by gender in score values between the textual and contextual gronps. Male subject performance indicated a great affinity for processing simple stimnli (e.g. a single word or picture) but showed great difficulty in the more “complicated” contextual gronp.
With regard to serial position differences, all three presentation formats conformed to typical serial memory performance-e.g. the primacy and recency effects were both established. Interestingly enough, the primacy effect was significantly greater than the recency effect. While this appears to contradict other stndies showing the latter as being stronger (Huang, 1985; Steiner & Rain, 1989; Paul & Whissell, 1992), the insertion of the demographic questionnaire (distracter) following item presentation corroborates the notion that “short presentation-recall intervals favor the recency effect while long presentation-recall intervals favor the primacy effect” (Korsnes & Gilinsky, 1993).
Additional research involving a retest of the same subjects with the same items in a different presentation order wonld increase the external validity of the results of this study. Until such an analysis is made these results are somewhat limited and should be viewed with some skepticism.