Investigation into different levels of processing

The following study was based on the “levels of processing theory” put forward by Craick and Lochart (Craick & Lockhart, 1972), which states that we are more likely to recall information when deeper levels of processing are used.

The purpose of the study was to investigate the relationship between levels of processing and the relative recall rates. The study examined three levels of processing words; phonemic, graphemic and semantic. The phonemic words were represented by rhyme, graphemic words were represented by capital letters and semantic were represented by adjectives. The study was conducted on n= 77, Thames Valley University Undergraduate Psychology students.

The first part of the study involved the display, with the aid of a projector, of a list of words to the participants. The words displayed were changed every 5 seconds, and each represented their respective categories; grapehmic, phonemic and semantic. For the autographic category, all the participants were asked to decide whether the words were written in capital letters and to either write yes or no. For the phonemic category, the participants were given a word and asked to write down another word that rhymed with the given word. Lastly, for the semantic category, the participants were asked to use adjectives to describe the given words. For the second part of the study, the participants were given a new blank piece of paper and asked to recall the original words that were presented to them without looking at the original words. The words from each category were calculated respectively in order to see which category scored the highest recall of words.

The results obtained from the experiment were significant with at P<0.001 level. A Bonferroni test was conducted, since a directional hypothesis was predicted. Additionally, a one way Anova test showed that the participants had better recall with semantic processing, where M=4.65) than with phonemic processing (M=2.88) and graphemic processing (M=1.43). Additionally, M=4.65) than with phonological processing (M=2.88) and graphemic processing (M=1.43), as shown in Figure 1; the F ratio = (2, 152) = 75.11, p<0.001.

Introduction:

It was first suggested in 1890 by William James, an American psychologist, that memory could be divided into two categories; Short Term Memory (STM) and Long Term Memory (LTM). In simplified terms, short-term memory was primary memory according to James, and was limited to more immediate memory and is therefore fragile. Secondary memory was more refined and therefore long-lasting (Eysenck, 2005 p.191).

Atkinson and Shiffrin (1968) later suggested three categories of memory storage being sensory store, short-term store and long-term store (Eysenck, 2005 p.189). The sensory store is based on stimulation of the senses, and was theorised to be the ‘entry point’ of information storage. Most of these recollections in sensory store were believed to last only for the duration of that stimulation. Subsequently, short-term memory, according to Broadbent (1958), arises when further concentration or attention was given to a particular stimulus, while in long-term memory information is ingrained and can be better recalled (Eysenck, 2005 p.191). These three stages were collectively known as the multi-store model.

In 1972, Craik and Lockhart developed the concept of levels of processing. This theory came about after the criticism of the aforementioned multi-store model, which focused on the structures involved in the STM and the LTM. What had been put forward was that these two categories of memory were stored separately in the mind, as they did not have the same fixedness or stability. Instead, the level of processing theory focuses on the processes which are involved in memory. At the core of Craik and Lockhart’s theory was the notion that it is the level or degree of processing that determines memory rather than the time elapsed, which minimised the distinction between STM and LTM. In place of time, memory, according to Craig and Lockhart, could be distinguished by infinite levels – from shallow to complex – by which the brain processes information (Craik and Lockhart, 1972). It is the depth of the processing that regulates memory, and processing is based on rehearsal.

They argued that rehearsal could be divided into maintenance rehearsal and elaborative rehearsal. Basically, maintenance rehearsal is the repetition of a stimulant such as a word to be stored in short-term memory. This does not automatically bring about long-term memory. However, in elaborative rehearsal, unlike the shallow processing in maintenance rehearsal, the brain processes the information on a deeper level. This form of rehearsal is semantic and requires more extensive thought and analysis as to the meaning. This increases long-term memory (Martin, G. et al., 2007 p.313)

Nevertheless, there have been criticisms of Craik and Lockhart’s theory on levels of processing. One of the main arguments criticising this theory is the fact that the degree of processing being used by a subject in an experiment cannot be determined. ( Eysenck, 2005 p. 208) Others such as Roediger and Gallo (2002) have argued that there is no distinction between intentional and incidental memory which is not addressed by levels of processing (Eysenck, 2005 p. 208). In addition, transfer-appropriate processing, suggested by Morris et al. (1977), asserted that processing based on relevance may produce different kinds of recall. (Eysenck, 2005 p. 209) Craik and Lockhart (2002) also later refined their theory and stated that shallow processing is not necessarily limited to short-term memory, nor is does deep processing automatically result in long-term memory (Eysenck, 2005 p. 210).

Hypotheses:

Experimental Hypothesis (H1): There will be a significant positive relationship between levels of deeper processing and the recall of words.

Null Hypothesis (H0): There will NOT be a significant positive relationship between levels of deeper processing and the recall of words.

Method:

Design:

The study consisted of a repeat measure ANOVA test which was not counter-balanced. All the participants took part in each of the conditions.

The study consisted of a singular Independent Variable (IV), which was an incidental learning task, and was divided into 3 levels of independent variables; semantic, graphemic and phonemic. The Dependent Variable (DV) was the number of words recalled.

Participants:

The total number of participants was n=77. The participants were all undergraduate psychology students at Thames Valley University. This was a mixed gender group, though a predominantly female, opportunist sample. All the participants were between the ages of 18 -50, with the majority being mature students. The participants were not requested to give personal information and participated as a partial fulfilment of the course requirement, however, they could pull out of the experiment at any time.

Material/ App:

The following items were used during this experiment:

A hand held stop watch which was used to time how long each the word was displayed.

Blank sheets of paper, used as an answer sheet.

Pens

Standardised instructions and briefing.

Stimulus Word list displayed on projector. There were a total of 30 randomly selected words displayed, with 10 words presenting each category.(Appendix 1) Each of the words were coupled with a question, which required either semantic, phonemic or autographic processing.

Procedure:

The participants were read out the standardised instructions, then were given a briefing. The participants were then given a blank sheet of paper to use as an answer sheet. The examiner then began the experiment. The first part of the study involved the examiner displaying a list of words to the participants. The words displayed were changed every 5 seconds and each represented their respective categories; phonemic, graphemic and semantic. For the graphemic category, the participants were asked to decide whether the words was written in capital letters, writing either yes or no. For the phonemic category the participants were given a word and asked to write down a rhyming word, e.g.: cart = mart. Lastly, for the semantic category, the participants were asked to use adjectives to describe the given words, e.g.: honey = sweet. All the results were to be compiled on one sheet of paper. For the second part of the study, the participants were given a new blank piece of paper and asked to recall the original words that were presented to them without looking at the original words. The results were then added up. The words from each category were calculated respectively in order to see which category scored the highest.

Results:

Due to the experiment being a repeat measure experiment, an ANOVA was performed to obtain results. The data was parametric and there were three conditions present. Furthermore, since a directional hypotheses was predicted, a Bonferroni post hoc test was used. The results are displayed in the following tables and figures.

Table 1: Descriptive Statistics for the recollection of words from all three conditions.

Conditions:

Mean

Standard Deviation

N

Capital (Graphemic)

1.43

1.34

77

Rhyme (Phonemic)

2.88

1.69

77

Adjective (Semantic)

4.65

1.94

77

Table 1 demonstrates that, according to the mean results, there is a significantly high difference between the graphemic and semantic words. This is in line with the predicted direction of the hypothesis.

A one way ANOVA test showed that the students had better recall with semantic processing (M=4.65) than with phonemic processing (M=2.88) and graphemic processing (M=1.43), The F ratio = (2, 152) = 75.11, p<0.001. Therefore, the ANOVA revealed there were significant differences between the levels of processing, however, it did not clearly indicate where these differences lay. To test the validity of the hypothesis, a post hoc test was used in order to see where the differences were, as the ANOVA test was unable to show this.

Figure 1: A line graph demonstrating the results of the study conducted, showing the relationship between deeper levels of processing and word recall.

Figure 1 demonstrates that the results of the study were clearly in line with the experimental hypothesis.

A Bonferroni Post-Hoc Test was conducted since a directional hypothesis was predicted. Furthermore regarding the extent of those differences, the results obtained showed that the significance is less than 0.05. The differences between the means of graphemic and semantic is 3.22, phonemic and semantic is 1.766, and graphemic and phonemic is 1.45. The significance is less than 0.05, which is highly significant showing the P value of P<0.001.

Moreover, the Post Hoc test shows that they are all significantly distinct from each other. These results suggest that the semantic level of processing provides the greater recall.

Discussion:

The purpose of the levels of processing experiment was to determine whether deeper levels of processing result in better recall. The experimental hypothesis (H1) is; There will be a significant positive relationship between levels of deeper processing and the recall of words. This experiment was accepted, thus rejecting the null hypotheses (H0), which stated that; There will NOT be a significant positive relationship between levels of deeper processing and the recall of words. The significance was clearly illustrated in figure 1. Moreover, the results from the one way ANOVA and Post Hoc tests confirmed the significance level at P<0.001. Following the descriptive statistics, the means and the standard deviations illustrated that there was a compelling difference between the graphemic and semantic conditions. Once again, the hypothesis was accepted and the null rejected.

An overview of the experiment indicates that the obtained results support the findings of Elias & Perfetti (1973) (Cox, 2001 ). Notably however, these results do not support Hyde & Jenkin’s (1973) study, which showed that participants of incidental tasks recalled the same as the intentional task participants (Eysenck, 2005). Further, the experiment fails to put into consideration the findings of Tyler, et al (1979), who suggested that retention of information is due to processing effort and not the depth of processing. On average, participants spend a longer time processing more difficult tasks given, which supports Tyler’s study, and it could therefore be assumed that the results obtained from the study are due to more effort being allocated to the given task (Craik & Lockhart, 1972).

Additionally, the study was not of a counter-balanced nature, which may have affected the participants recollection. The type of processing, with regard to effort and time taken to process, may also be influential. These variations could lead to deeper processing as more effort is used recalling, thus creating difficulty in determining which factors influenced the results. It is noteworthy that in level of processing experiments, it is often rather difficult to determine which level of processing is being used. This can be seen in a study by Hyde & Jenkins, 1973 where it was illustrated that the ability to recall a word, involved linking it to its meaning, however it was unclear why this was the case. Another objection was that the task of determining the part of speech to which a word belongs is a shallow processing task. On the contrary, other researchers assert that this involves semantic processing. Lastly, in the levels of processing theory, there is no concise explanation as to why deeper levels of processing results in better recall. According to Eysenck (Eysenck, 2005) the levels of processing theory describes rather than explains this occurrence.

Notably however, a number of improvements could have been made to the experiment. As previously mentioned, there were more female participants than male. The gender distribution should have been more even. Furthermore, the seating environment of the experiment was poorly set up. The participants were seated too closely to each other, creating the possibility to cheat by looking at a neighbour’s sheet. Additionally, the study did not put learning disabilities into consideration. For instance, people with dyslexia have phonological processing impairments with may alter the outcome of such an experiment. An additional improvement could include using a wider population and an independent group design, as the experiment was quite biased – only undergraduate psychology students took part in this repeat measure design. This could have given the subjects the advantage of practice, therefore making it easier for them to recall the words and making the results of the experiment less reliable.

On the whole, the experiment turned out to be an success as the findings were in correlation with Craik and Lockhart’s findings. As the results showed that deep processing, which involves attention to meaning and rehearsal will generally result in better recall.

References:

Cox, E (2001). Psychology for A level. Oxford University Press

Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behaviour, 11, 671-684.

Davey, G. et al. (2004), Complete Psychology. Hodder Headline: London .

Eysenck, M.W., (2005), Cognitive Psychology: A Student’s handbook. Psychology Press: New York.

Martin, G. et al. (2007), Psychology (3rd edn.) Pearson Education Limited: Essex.

Appendices: