The purpose of this study was to investigate the Pollyanna Principle and whether it possessed the capacity to overshadow the phenomenon of mood congruence on a recognition task.  It was hypothesized that participants would score higher on the recognition task for positive affective words in comparison to recognition scores for affectively neutral and negative words; this difference should be should be apparent for all participants no matter which condition they were in.  If the Pollyanna Principle possesses the capacity to overshadow mood congruence, it was hypothesized that there would be nonsignificant differences present when mean recognition scores for the three word lists were broken down by mood induction conditions. 

The results for the current study partially supported the Pollyanna Principle and first hypothesis.  There were no significant differences found between the recognition scores for the affectively positive, neutral, and negative word lists when scores were looked at according to treatment condition.  Similar results were found by Hartig and colleagues (1999) in an attempt to demonstrate mood congruent recall for words describing positive and negative emotions.  The researchers were unable to display a mood congruent effect on a free recall task with participants in both the positive and negative mood conditions recalling similar amounts of positive and negative words.

  When the three mood treatment conditions of elation, depression, and neutrality were discarded, significant differences were apparent for the recognition scores for the three word lists.  Participants recognized more positive and negative words than neutral words.  There was not an obvious processing dominance for affectively positive words over negative words.  While the results of the current study supported the notion of a processing dominance for positive words over neutral words, the idea of a processing dominance of positive words over negative words remained unsubstantiated. 

Research by Benjafield and Adams-Webber (1976) offered an interesting explanation for the present study’s inability to support a processing dominance for positive words over negative words.  The golden section hypothesis, that is investigated in their research, suggested that while individuals make efforts to see things positively, the negative side is not ignored.  People take steps in an effort to strike a balance between the positive and negative aspects of life.  If Benjafield and Adams-Webber are accurate in their hypothesis of human perception and cognition, their research possesses a possible explanation for the results of the current study.

            Research by Isen and Shalker (1982) investigated mood congruence, while acknowledging the fact that positive moods do not possess the capacity to overshadow the negative characteristics of particular stimuli.  Results supported the idea that individuals pay attention to both the positive and negative aspects of stimuli.

            The findings of the current study contradict the research of Forgas and Bower (1987) that indicated participants made more errors in both recall and recognition when material opposed the current mood state.  The present study did not find such a difference in recognition scores when mood was taken into consideration.

            Research by Matlin and Stang (1978) indicated two factors that were important in Pollyanna congruent recall, and which may also explain why a strong Pollyanna effect was not found in the present study.  The investigation of research on recall tasks identified the factors that were important in selective recall; the accuracy of generalizing the factors to a recognition task is questionable.  The two factors identified were the length of delay prior to recall and the criterion of learning.

            There was only a short delay between the original presentation of the word lists and the subsequent recognition task.  The participants reading the instructions for the recognition task caused the delay.  The length of delay depended upon the reading speed of the participants and was not controlled for.  Research has shown that the longer the delay before the recall task, and possibly the recognition task, the greater the Pollyanna effect displayed (Matlin & Stang, 1978).

            There was not a high criterion of learning present in the current study.  The participants were told that they would be asked to recognize the words seen in a later portion of the study, but a high level of perfection was not demanded.  According to research, when perfection is requested by the experimenter the incidence of selective recall congruent with the Pollyanna Principle will increase (Matlin & Stang, 1978).

A second hypothesis of the current research was the possible existence of a Pollyannic personality as a component of optimism.  It was supposed that if such a personality existed, an individual who was high in optimism might remain unaffected by the mood induction procedure that he or she experienced.  Also, a strong relationship would be found between optimism and the recognition score for affectively positive words.

            The results of the current study showed that a significant relationship existed between optimism and the pre-test ST-DACL.  There was not, however, a significant relationship between optimism and the post-test ST-DACL or the score on the recognition task for affectively positive words.  This evidence does not support the idea of an over-all Pollyannic personality.  It also contradicts the research of Matlin and Gawron (1979) that showed various measures of Pollyannaism to be moderately correlated with optimism and self-report happiness.

            Not only was there a significant relationship between optimism and the pre-test ST-DACL, but there was also a significant relationship between pessimism and the pre-test ST-DACL in the present study.  These findings, combined with the nonsignificant relationship between optimism-pessimism levels and mood congruent/Pollyanna effect, support the research by Mayer and colleagues (1992).  While their research supported the mood congruent effect, the correlation between optimism-pessimism levels and mood was noted.

This study contained not only an elation treatment condition and a depression treatment condition, but also a neutral condition.  The neutral condition served as the control condition which ensured that if differences in scores on the recognition task were caused by the treatment conditions, it would be readily apparent.  Some researchers did not use a neutral condition as a built in control in their research (Hartig et al., 1999).  Other researchers only investigated depressed and neutral moods in the search for the mood congruency (Varner & Ellis, 1998).  This present study possessed a distinct advantage over some of the previous research done in the area by looking at the recognition scores for the three conditions of elation, depression, and neutrality. 

            Another advantage the present study had over past research was the manner in which it contained the three affectively different word lists.  The use of neutral words along with affectively positive and negative words served as a built in control.  The presence of the three different word lists allowed the experimenter to assess whether a difference in recognition scores was truly attributable to affective word type.  Hartig and colleagues (1999) only used positive and negative words in a free-recall task while researching the phenomenon of mood congruent recall.  In research done by Varner and Ellis (1998) on mood congruent memory, participants were only presented with negative words and words that related to writing a paper.

            While the current study did possess some advantages over past research, it also was limited by some factors.  Some of the confounding variables that may have reduced the study’s internal validity and influenced the outcome of the study include: history, testing, selection, diffusion of treatment, and sequence effects.

            History, a potential threat to internal validity, is caused by events that trigger a change in the dependent measure that is not caused by the manipulation of the independent variable.  Participants in the study were tested during what may have been a very emotional time for some.  The Boston Red Sox were in the World Series, and a majority of the participants in the study were tested on nights on which games were played.  Since mood was the independent variable in the present study, factors other than the mood induction procedure used may have affected the moods of the participants. 

            The fact that the World Series occurred during the testing phase of this study is only of importance because the participants were from a small liberal arts college in Manchester, New Hampshire.  Having ‘their’ team in the series may have been a confounding variable.  Also, many individuals who signed up for the experiment did not come on nights that there were games.  There may have been a difference between those who came and participated in the study and those who did not.

            Another potential effect of history could have been the presidential election that occurred on one of the nights that participants were tested.  With the state of New Hampshire, where the college was located, being a battleground state, there were intense emotions experienced by some.

            The effect of history on mood was controlled for some degree by the administration of ST-DACL twice during the study.  With mood assessed before and after the occurrence of the mood induction procedure, the experimenter was able to determine whether the mood induction procedure was able to influence the moods of the participants. 

            The problem with using the same form of a test twice in a study is a potential threat to internal validity known as testing.  The effects of testing occur when a participant’s score on a given measure if affected by taking the same test more than once.  While using the ST-DACL Form B twice in the present study had the potential to confound the results, the ability to determine whether the mood induction procedure significantly changed mood was deemed to be of greater importance.

            Selection is another potential confounding variable that occurs when the participants in the various conditions of a study are not equivalent before the actual study begins.  One way that the present study controlled for selection was by randomly assigning participants to one of the three conditions.  Absolute random selection was not a possibility in this study.  Often when participants did not arrive at the allotted time, other students present in the Saint Anselm College Psychology Department or the Weiler Computer Lab were drafted for the study. 

Levels of optimism, pessimism, and mood were determined to be possible factors that may have caused the participants in the group to be different before the study began.  For this reason, the Optimism-Pessimism Scale and the ST-DACL (Form B) were given before the mood induction procedure.  The results showed that the groups were equivalent on these measures before mood was manipulated.

            For the present study, there remained the threat of diffusion of treatment in which information that is exchanged between participants could potentially affect the outcome of the study.  The experimenter, requesting that participants not discuss the study with their peers until after the entire study was complete, controlled for potential diffusion of treatment. 

There was only one point in the study that the participants had the opportunity to engage in conversation with the other participants.   This opportunity  occurred after the mood induction procedure and the second administration of the ST-DACL when the participants and experimenter went to the Weiler Computer Lab.  The experimenter discouraged conversation by a normal paced walk to the lab that was often single file.  Although the participants were not explicitly instructed not to engage in conversation, the tone of the walk did not invite conversation.

            The confounding variable of sequence effects is caused when the performance of a participant is affected by earlier conditions of a study.  The participants underwent two separate recognition tasks, with each task containing affectively positive, neutral, and negative words.  The SuperLab program used controlled for possible sequencing effects by randomizing the presentation of the individual word lists within each recognition task.  The order of the recognition task was not controlled for.  The participants always experienced the recognition task with lists one and two before the recognition task with lists three and four.

            The Velten (1968) mood induction procedure was used in this research.  The current study did not, however, utilize the built in control groups for demand characteristics which were typically used with Velten.  The limited pool of participants for the study prevented this control from being employed.  It was also deemed by the experimenter that research in support of the actual mood inducing power of the Velten did not indicate a need for such a control to be present.  The effects of the mood inductions were assessed using the ST-DACL twice during the course of the study.

            Participants in the current study were not given a third ST-DACL after the recognition task.  While this might have been a useful addition to the study, using the same test three times could have been a potentially large confound.  Future research should potentially investigate mood after the mood congruent/Pollyanna tasks are completed. 

             There was limited control exercised over the word lists used in the SuperLab phase of the present study.  The experimenter exercised control over the affective ratings of the word lists.  Control was not exercised over what part of speech the words were, the number of syllables in the words, the actual length and number of letters in the words, and whether the words were also in the Form B of the ST-DACL.  Future research should seek to exercise control over these variables.

            The greatest amount of control practical was exercised over the environment in which the participants were tested.  All testing occurred during the evening hours.  A majority of the testing occurred in the conference room of the Saint Anselm College Psychology Department.  The only individuals present during testing were the participants and the experimenter. 

However, the computer recognition task took place in the Weiler Computer Lab.  The recognition task had to be performed in the Weiler lab because the computers in the lab were the only ones with the program necessary to run SuperLab.  The experimenter was unable to control for the amount of people in the lab during testing.  Often there were only a small number of people using the computers in the lab, yet there still existed the possibility that the other people present distracted the participants in the study.  This possibility of distraction was a possible confounding variable that should be controlled for in future research.

            Potential threats to the construct validity of the present experiment were hypothesis guessing and the expectancies of the experimenter.  Hypothesis guessing was controlled for by the participants being asked on the questionnaire, at the end of the study, whether they felt they were able to guess the hypothesis of the experimenter at any point during the study.  If the participant believed that they knew what the hypothesis was, a space was supplied for them to provide his or her guess.  While a number of participants believed they knew the hypothesis of the experimenter, none of the guesses were accurate.  Expectancies of the experimenter were controlled by providing the same treatment to all participants no matter what the condition and also by reviewing the results in an unbiased manner.

            The present study possessed limited external validity.  The results from the current study can only be generalized to a small population.  Since the research was conducted using college students from a small liberal arts school in the northeastern section of the United States, the results cannot extent beyond this particular population.  The results gathered from this research have implications for potential future research in the area of moods and memory.

            The present study possesses many implications for future research.  Although the actual study possessed a variety of limitations, it was able to set the basis for future research in the area.  While the SuperLab program used was only able to conduct a recognition tasks, other aspects of the program could potentially be used in future research on the Pollyanna Principle.  The SuperLab program possesses the capacity to record the reaction time for each stimulus presented, thus allowing for the speed of response to be investigated in terms of both condition and word type in a recognition task.  Other possible uses of the SuperLab program include a Pollyanna spew task similar to the one in research by Hummer and colleagues (1992) or a computer task similar to the done by Osgood and Hoosain (1983) on the processing time of affectively positive and negative words.

            While the present study was unable to display a clear-cut Pollyanna effect, it presented evidence against the mood congruence effect.  The results of the present study serve as a basis for future research in the area.  Research on the Pollyanna Principle and the manner in which memory is influenced by the affective nature of a stimulus and/or mood is important and has implications for a variety of realms.  The manner in which people process material and what affective varieties of information are processed more efficiently is of utmost importance in a classroom situation and also in everyday life.