Levels of Processing and Their Effect on Information Retention
Memory and Self Assessment 2 Abstract One of humanities’ greatest strength is our capacity to learn, although not all methods of learning were created equal. If humanity is to continue to grow as a species our methods of learning must continue to grow and improve. This experiment illustrates how different methods of encoding can affect how information is retained. Nine hundred and ninety nine participants were presented with 26 pairs of words under both semantic and phonemic conditions and then tested to see how many they could recall. Participants were also asked to assess themselves prior to the experiment.
It was found that most people overestimated their abilities and did not take into account the method of encoding. Memory and Self Assessment 3 Levels of Processing and their effect on Information Retention Do we perceive our ability to retain information accurately? Significant time and effort has been invested into researching memory and the effect the depth of processing has on the ability to recall words. Research has shown that semantic encoding has resulted in significantly better word retention than such as phonemic encoding (Craik & Tulving 1975).
When asked to estimate their ability to perform in tasks involving memory, the majority of people will not take the method of encoding into account and will estimate inaccurately (Dunning, Heath & Suls 2004). The reason that semantic encoding has resulted in a better word retention than phonemic is due to the variation in the depth of processing. Sensory interpretations such as the sound or appearance of a word are processed at shallower levels and produce only shortterm recall. These levels are involved in phonemic encoding, for example grouping words together depending on how they sound, such as TRAIN and SPAIN.
Deeper levels of processing concern the meaning of the word and result in a more long-term recollection. These levels are involved in semantic encoding, grouping words together depending on their meaning, CAT and DOG for example. The reason people estimate their ability to recall inaccurately could be one of a many. It could be that people feel the task simple and overestimate their ability or they could think the task difficult, when in reality it isn’t, and underestimate their abilities. Another reason could be that people aren’t aware of the theory behind the task and are unable to make an informed decision (Dunning, et al. 004) This experiment aims to demonstrate whether semantic or phonemic encoding yields a higher recall, and to investigate how accurate people are at estimating their abilities. The primary hypothesis of this experiment is that participants in the semantic condition, the deeper level of processing, will result in a higher percentage of words recalled. The secondary hypothesis is that people aren’t aware of how the encoding process affects their ability to recall the words and so the estimations for both conditions will be the same.
The tertiary hypothesis is that people will overestimate their abilities and so the predictions for both conditions will be higher than the given value. Memory and Self Assessment 4 Method Participants There were 992 first year psychology students that participated in the experiment. The age and gender of the students was not recorded. 518 students participated in the phonemic condition and 474 students participated in the semantic condition. Materials The experiment involved the use of a web browser on a computer; participant’s input was done with keyboard and mouse. 6 word pairs were used and six test word pairs. Procedure Participants were randomly assigned to either the phonemic or semantic condition. They were told to go to a computer open the web browser to the experiment, enter their name and select the condition they had been assigned. Participants were instructed in the experiment’s procedure and then asked to estimate the percentage of words they would be able to recall. They were then presented with a trial of six practice word pairs, presented one at a time for 30 seconds.
Whilst the word pairs were visible, the participants were asked whether they agreed or disagreed, if the words rhymed or were in the same semantic category. After the participants selected they were given feedback about whether the answer was correct or incorrect. The participants then began encoding. During encoding participants were presented with 26 word pairs, one at a time for 30 seconds each. Participants then had to select whether they agreed or disagreed for each word pair. No feedback was given during encoding. Participants then began the recall phase.
During the recall phase one word from each of the previous pairs was presented on the screen, all 26 pairs were presented one at a time, participants were asked to enter the partner word. During this phase there was no time limit and participants were not penalised for incorrect spelling. After this phase a result page opened, with a table indicating scores for each phase. Results The results illustrate that for both (Table 1) the phonemic and (Table 2) semantic conditions the mean actual recall score was lower than the mean estimated recall score.
Participants in the semantic condition had a higher mean recall than participants in the phonemic condition. The Cohen’s d for this was 0. 61, which is a medium effect. The estimated recall for both conditions was similar, the participants in the phonemic condition having a slightly high mean, and Cohen’s d for this was 0. 09 showing this was a very small effect. Memory and Self Assessment 5 Table three shows the combined results of both conditions, it shows that the estimated recall is greater than the actual recall. The Cohen’s d for this effect was 1. 4 a very large effect. Table 1 Statistics for the Phonemic encoding condition Table 2 Statistics for the Semantic encoding condition Table 3 Statistics for both conditions combined Discussion It was found that participants in the semantic condition remembered on average 10% more word pairs than participants in the phonemic condition. This supports the hypothesis that deeper levels of processing lead to greater ability to retain information. It would then be a logical conclusion to say that semantic encoding leaves a more long-term memory.
Since there was an almost negligible difference in the mean estimated recall between the two conditions, Cohen’s d supports this. It is reasonable to say that people were unaware of how the different levels of encoding would affect their ability to recall the word pairs. This supports the secondary hypothesis, which states there will be no difference in the estimates between the two conditions. The tertiary hypothesis is also supported; it was noted that people estimated their recall to be on average 28% higher than their actual recall. Which shows that people will Memory and Self Assessment 6 verestimate their ability to recall word pair. This could be due to the lack of information given in the task, for example a greater emphasis was placed on the first phase of the experiment. Alternatively people could have perceived the task as easy. This study has lead to similar findings such as those in Lewandosky and Hockley (1987). This study does differ in a few key aspects though, such as the introduction of a recall estimate. It is quite a sobering fact to know that people have such little knowledge of how different encoding conditions can greatly affect one’s ability to retain information.
This experiment has only acted as introduction into the field, however further research into this area can lead to better teaching and studying methods. The experiment had a few limitations, the venue for the test for example, participants were not isolated and could talk during the experiment, and also there was a significant amount of background noise and other distractions. Participants were not well supervised during the experiment; it would have possible for some participants to cheat by recording the word pairs.
To prevent this future experiments should isolate participants during the testing process and supervise them more vigilantly. I would also suggest that participants should be asked to estimate the percentage of words they can recall between phase 1 and phase 2, this should eliminate bias since some participants could have spent ore time trying to commit the words to memory knowing that they would have to recall them later. Another point of interest would be to look at the correlation between word pairs remembered and whether the participant agreed or disagreed with the particular word pair.