Method
Participants
Fifteen undergraduates from the University of California, Los Angeles participated in this experiment. There were 8 men and 7 women, age ranging from 19 to 30 years old. All 15 participants were enrolled in a research method class and they participated in the experiment as a part of a course requirement.
Design
A 2 x 2 (Type of sentence x image) within-subjects factorial design was used. The two independent variables were the type of sentence and the type of image with the respective to the italicized word in the sentence. The two levels of the type of sentence were humorous and nonhumorous sentences. The two levels of the type of image were matching or nonmatching images respective to the italicized words in their sentences. Matching images depict the italicized word in their corresponding sentences. Nonmatching images did not depict the italicized words. The dependent variable was the number of italicized words correctly recalled.
Materials
Sentences. Two sets of 20 sentences obtained from Schidmt (1994) were used. At the humorous sentence level, the 20 sentences had the mean rate for humor ranging from 3.08 to 4.0. At the nonhumorous sentence level, 20 other sentences had the mean rate for humor ranging from 1.25 to 2.86. The intensity of humor was determined by participants using a 5-point Likert-type scale where 5.0 was the highest humor rating. Using these two sets of sentences, four
lists were created, each with 20 different sentences. Random assignment was used to determine the placement of the sentences on the lists so that the serial positions of the sentences across the four lists are different from one list to the other. The words that will be used for recall were always italicized across the four lists.
Pictures. 20 Microsoft Word Clip Art images were used for the matching conditions. The images were obtained using the key words of the italicized words. For the nonmatching condition, 10 neutral key words (i.e., sun, flower) were used to obtain the 10 images from Clip Art. The nonmatching images were randomly assigned to the sentences. Each of the four lists has the same type and number of nonmatching images. Within every list, an image only appeared once. All images were 261 pixel by 159 pixel. All images were presented centered and below its respective sentence.
PowerPoint slides. Computers with Microsoft PowerPoint program were randomly assigned to show one of the four types of sentences to participants. Four different lists each with 20 sentences and 20 images were obtained. Each list has 5 humorous sentences with matching images, 5 humorous sentences with nonmatching images, 5 nonhumorous sentences with matching images, and 5 nonhumorous sentences with nonmatching images. I counterbalanced the lists to control for order effect so that the serial positions of the 20 sentences are random within each of the four lists. To control for specific item effect, I use a balance Latin square to ensure that each of the 20 italicized words from the two sets of sentences appear in each of the four experimental condition. PowerPoint slides were used to present all the instructions, lists of words and pictures, and the first five math problems.
Response sheet. A 8 ½ in. by 11 in. paper were used to write the italicized words that the participants recalled.
Distractor test. 10 arithmetic math problems were used as a distractor test. The first five math problems were presented in a PowerPoint slide and the next five math problems were presented in a 8 ½ in. by 11 in. piece of paper. The reason why the distractor task was divided into both a PowerPoint and regular written exercise was to reduce as much of the demand characteristics of the experiments as possible. Participants write the answer to all 10 math problems on the paper.
Procedures
Participants were randomly assigned to a computer with one of the four types of lists. The experimenter told the participants to follow the instructions on the packet of paper given to them on how to access the presentation lists and the instructions presented by the computer. The experimenters made sure that all participants got to their respective lists before telling them to begin. Participants all started together. The instructions on the computer directed the participants to look the following 20 slides and try to remember as much information as they can about the slide. Each slide was shown for 7 seconds, one after another, and a blank screen was shown for 2 seconds between each slide.
After all 20 slides were shown, a new slide informed the participants that they now have five minutes to solve both the five math problems presented in the slide and five math problems on the sheet next to the computer. The experimenter announced that time was up and told the participants to stop what they were doing and were to look back at the computer for further instructions. The final slide appeared on the screen and instructed the participants to write down as many italicized words as they can remember on the recall sheet. The recall test was an unexpected recall test because the participants did not expect that they need to remember the italicized words of the sentences rather than images or the content of the sentences. In the process of counting the number of italicized word correctly recalled, I used a lenient scoring criteria. If the main meaning of the correct italicized word was written if not the exact word, it would still be counted as a correctly recalled word. For example if the correct word was “cheap” but “cheaply” was written, it would be counted as one correctly recalled word.
Results
Figure 1 presents the average number of italicized words correctly recalled when they were presented in the form of humorous versus non-humorous sentences and as a function of the type of image. Looking at the pattern of results displayed in Figure 1, it appears that italicized words, in general, were better recalled when presented in the form of matching images than when presented in the form of non-matching images and when paired with humorous sentences rather than non-humorous sentences. The degree to which recall was affected by type of sentence, however, appears to depend on the form in which the items to be recalled were presented. More specifically, when sentences were paired with matching images, type of image appears to have had a large effect on their later recall; whereas, when sentences were paired with nonmatching images, type of image appears to have had little or no effect on their later recall.
To test these apparent effects, the data were analyzed using a two-way within-subjects ANOVA, which revealed no significant main effect of type of sentence, such that average correct recall was not significantly different when sentences were paired with matching images
(M = 1.03, SD = 0.88) than when paired with non-matching images (M = 0.97, SD = 0.62), regardless of the form in which they were paired, F(1, 14) = 1.66, MSE = 1.56, p = . 063. Also, no significant main effect of type of sentence was found, such that average correct recall was not significantly different when images were paired with humorous sentences (M = 0.97, SD= 0.93), than when paired with non-humorous sentences (M = 0.63, SD = 0.81), regardless of the way it was paired, F(1, 14) = 1.07, MSE = 1.56, p = 0.32. Additionally, there was no statistically significant interaction between type of image and type of sentence, as indicated in Figure 1,
F(1, 14) = 0.37, MSE = 0.73, p= 0.56. At the humorous sentence level, matching images have a mean of 1.27 and non-matching images have a mean of 0.67. At the non-humorous sentence level, matching images have a mean of 0.8 and non-matching images have a mean of 0.47.
Discussion
I was interested in the effects that humor and incongruence have on memory. In my experiment, humor was operationally defined as the type of sentence, humorous or nonhumorous, and incongruence was operationally defined as the type of image respective to the italicized word of the sentence, matching or nonmatching images. Participants were randomly assigned to one of the four PowerPoint lists slides. We operationally defined memory as the number of italicized words correctly recalled by the participants. I hypothesized that the more humorous articles are, the more you are likely to remember it. Specifically in my experiment, I predicted that italicized words in humorous sentences will have higher recall than words in nonhumorous sentences. I hypothesized that the more incongruent an article is relative to the message, the more likely that the information of the articles will be remembered. Specifically in my experiment, I predicted that italicized words that are paired with nonmatching images will have higher recall than those that are paired with matching images. I also hypothesized that there would be an interaction between humor and incongruence. Specifically, I predicted that at the nonhumorous sentences level, nonmatching pictures will better recalled than matching pictures but at the humorous sentence level nonmatching pictures will be recalled even more. Thus, the when the effects of incongruence and humor are combined together, their effects intensify so the level with humorous sentences and nonmatching images would have a highest number of word recall relative to any other level.
Statistical results of my experiment showed that there was no significant main effect of humor, no significant main effect of incongruence, and no significant main effect of interaction. Specifically, the amount of italicized words in humorous sentences were not recalled higher than those that were in nonhumorous sentences and words that are paired with nonmatching images were not recalled higher than those that were paired with matching images. Also, my prediction that there would be a significant interaction in which the condition where humor and incongruence are combined, humorous sentences with nonmatching images condition, there would be a highest number of word recalled was not found in my experiment.
The experimental findings in Schmidt (1994) experiment one found that humorous articles will be recalled more than nonhumorous articles. In my experiment I have tried to replicate these findings but my results did not reflect Schmidt’s results. Comparing the experimental methodologies between Schmidt’s and mines, I noticed some differences that might have explained for our dissimilar results. Schmidt’s operational definition for memory was the number of words in the sentences correctly recalled. My operational definition for memory was the number of italicized words correctly recalled. In my experiment, even though for almost every subject there were just a few number of italicized words correctly recalled, the total number of words recalled were much higher indicating that perhaps if I had used Schmidt’s operational definition for memory, our results might have matched. The presentation of the sentences in my experiment was silently shown through PowerPoint whereas participants in Schmidt’s experiment heard the sentences through prerecorded tones. Audio presentation of the materials might have elicited a significant main effect rather than visual presentation. Also, Schmidt’s participants saw the sentences for 10 seconds compared the 7 seconds that were given to my participants. The extra time given in Schmidt’s experiment might have given more time for his participants to process the humor to creating a significant main effect of humor. Also, the humor intensity of the sentences that were given to Schmidt’s participants were rated by people their own age and era. However, since I also used the same sentences but gave them to participants that are in a different era, 12 years later from when the time Schmidt’s sentences were compiled, the cohort effect is a problem in my experiment. It is possible that my participants did not find the sentences that were suppose to be funny to be actually funny because of the time gap, therefore, my results show that significant difference between the type to sentences.
Hastie and Kumar (1979) found that incongruent behaviors to that of a given personality trait were remembered higher than congruent or neutral behaviors. Since my hypothesis was also that incongruent materials are recalled higher than congruent materials, my experiment tried to replicate Hastie’s and Kumar’s findings. There were many procedural differences between our experiments that might have been the cause for the dissimilar results. Hastie and Kumar operationalized incongruence as discrepant behaviors to a given personality. For instance, participants were given a personality (e.g., intelligent) then a set of incongruent behaviors (e.g., made the same mistake twice). Participants also were given sets of congruent and neutral behaviors. The recall task was to write down as many behavioral descriptions that were presented as possible to measure memory. All the experimental materials were verbal articles whereas in my experiment, images were substantial elements of my operational definition of incongruence. I operationalized incongruence as nonmatching images to the respective italicized word. Also, participants in Hastie and Kumar’s experiment read the personality and behaviors descriptions out loud unlike the silent viewing of PowerPoint slides in my own experiment. The difference in the types of audio and visual stimuli between mines and Hastie’s and Kumar’s experiment might have attributed to our dissimilar results. Futhermore, recalling behaviors were probably easier because participants most likely have performed the behaviors that Hastie and Kumar presented creating personal meanings to the participants rather than the distinct words (e.g. hydrogen bomb, blotters) in my experiment that might have lesser meaning to my participants.
Houston, Childers, and Heckler (1987) found that incongruent picture and verbal descriptions in advertisement resulted in higher recall of the product’s information than congruent picture and verbal descriptions. My experiment resembled of that of Houston et al.’s in the respect that we both use pictorial and verbal articles to operationalize incongruence. However, the differences in the type of pictures and other methodologies might have been the reason why, unlike them, I was not to get a significant main effect of incongruence. The pictures that Houston et al. used were black and white and were pretested to show that the pictures adequately conveyed the product attributes. Whereas, I used only the available colored images from Microsoft Word ClipArt and had to use my own judgement in selecting pictures, the final pictures that were chosen might not have sufficiently reflected the italicized words (e.g., the word “life” was pictorial presented as a nurse with a newborn baby). Also, Houston et al. presented the images above the verbal descriptions whereas I did the opposite. Additionally, Houston et al.’s participants were given 15 seconds to view the advertisement instead of 7 seconds. The difference in orientation, type of picture, and time given between Houston et al.’s experiment and mines might have resulted in the dissimilar results.
The limitations that were unique to my experiment could also possibly have caused my results to not reflect prior research’s predictions. With only 15 participants available, I was forced to confine my experiment to a within-subjects design. This design might have stopped us from coming up with a significant main effect for incongruence because Houston et al. (1987) found a significant main effect by executing a between-subjects design. Also, because whenever a small sample size is used, the power of the experiment is substantially reduced and is harder to reject the null hypothesis which might have explained why I did not get any significant results. The results of a small sample size are also likely to only reflect a minority’s information even though the majority of the findings show that there should have been significant main effects and interaction. Furthermore, there was no random sampling because all my participants were psychology students with a research methods background. These characteristics no longer make naïve participants and could have influenced the results. Participants might have experienced fatique effects because of my experiment was not the first experiment they went through. Within the experiment, the humorous sentences might have cause a carry-over effect that could have prevented the participants to pay attention to the subsequent sentences. Also, the presentation of the PowerPoint slides might have been too fast for participants to read completely both the sentences and the images, causing them not to experience the effects of incongruence and humor, thus skewing the results. Finally, participants might have been confused whether to recall the image or the italicized words because the experimenter in my experiment did not verbally stress the recall directions, thus correct word recall might have been reduced.
Future research that would want to repeat my experiment should take into consideration the following suggestions to increase the possibility of obtaining significant results. To increase external validity, obtain a bigger and more diverse sample of participants rather confining participants to the group of psychology students that were in my experiment. To increase internal validity, reduce demand characteristics, fatique effect, and carry-over effect by performing a between subjects design like the design used in Houston et. al (1987). If type of sentence will be repeated as the operational definition for humor, instead of using Smidtch’s sentences, new lists of sentences with humor intensity rated by new participants to ensure that participants really think the sentences are funny or not. If the type of image will be repeated as the operation definition for humor, images that are meant to match words should undergo separate test to ensure their similarities to the words. For the operational definition of memory, researchers should contemplate adopting a free word recall where the main ideas or the any words in sentences can be recalled as oppose to restricting recall to italicized words of the sentence. Schmidt, Hastie and Kumar, and Houston et al. have used this type scoring and operation definition for memory and found significant results. Unlike my experiment, at least two experimenters should score the results and together reach an inter-rater agreement to avoid bias in scoring the data. Finally a double-blind experiment should be used where neither the participants nor experimenters can predict what condition the results are coming from.
My purpose of my experiment was the find the effects that humor and incongruence have on memory. I did not find any significant main effect or interaction between the two variables due to the limitations of my experiment. However, I am confident if future research were to replicate my research with my suggested modification in methodology, significant results will be obtained that will help us gain great insights on how these two stimuli influence memory. A good understanding of these processes will enable psychologists to construct experiments and to make predictions of ways to use these stimuli to enhance recall in the real world (e.g., education, advertisement, personal identity). Understanding the ways emotional stimuli affect retention of education materials can help the school system develop a more effective way to teach students important information. For businesses, understanding how memories are preserved can help advertisers create powerful advertisements to catch consumers. Personal memory is important for shaping our attitudes and perceptions the world so with a better knowledge of how memory are formed and sustain, people can live a more fulfilling life.
References
Hastie, R., Kumar, A. P. (1979). Person memory: Personality traits as organizing
principles in memory for behaviors. Journal of Personality and Social Psychology, 37,
25-38.
Houston, J. M., Childers, L. T., Heckler, E. S. (1987). Picture-word consistency
and the elaborative processing of advertisement. Journal of Marketing Research, 25,
359-369.
Schmidt, R. S. (1994). Effects of humor on sentence memory. Journal of Experimental
Psychology, 20, 953-967.
Figure Caption
Figure 1. Mean number of words correctly recalled as a function of type of sentence and type of image.