In this essay I aim to discuss the advantages and disadvantages of theories. I intend to look at theories on imagery, the hierarchical network, matrix, feature models, spreading activation and schemas.
There are examples of imagery theories from 500BC (Simonides' - Method Of Loci). This involves seeing objects in a series of rooms to aid recall. The narrative story uses a similar method. Bower (1977) discovered paired associates given to subjects were recalled more easily than those not using imagery (imagery - 80% recall, non-imagery - 45% recall). He found that the more bizarre the pairs were the higher the recall rate. Anderson (1995) also said that the more time and effort spent making the image and interacting with the image the more effective the recall.
Pavio (1985) asked the question, "Why are images better remembered?". There is a dual-codes model of memory:
Sensory codes = visual = imagen
Verbal codes = words = logogen For the same reason concrete words are recalled better, they use imagen and logogen. Abstract words are not recalled well because they only use logogen.
I think that imagery is a valid theory for the organisation of information in memory as shown by the results of Bower's paired associates experiment ( the subjects that used imagery had a 35% higher recall rate than those that did not use imagery).
Collins and Quillian (1972) put lexical memory into a " hierarchical network model ". They used nodes or concepts and attached them to features.
The subjects are given sentence verification tasks, for example:
"A canary can sing" fast reaction time
"A canary can fly" longer reaction time (further away)
"A canary can breathe" even longer reaction time (at the top of model)
Subjects were also given false sentences: "A canary has gills" in theory this should take a very long time to work out, but in practise it does not.
The confounding variable here is the category size. Some items are more typical to the category. The ethnocentricity is not taken into account (episodic and semantic memories). There are other explanations, the relative frequency of concepts being related to each other, for example:
"A bear is an animal" is verified more quickly than "A bear is a mammal" which goes against the model.
The hierarchical network model is not particularly valid because there are many examples that go against it, but the recall is good and it is effective because there is little redundancy.
In 1978 Broadbent developed the matrix, words given in diagrammatical form in categories:
The matrix also gives good recall but has similar problems to the hierarchical network model.
The Spreading Activation Model (designed by Collins and Loftus (1975)) measures semantic distance in your mind between words. Hierarchical relationships are out. Collins and Loftus explained that activation spreads in all directions. A lot of energy starts at the prime and then spreads out and decreases as it goes ( as shown in the lexical decision test)
Schemas are "maps" or "packets of information" for different circumstances in your life. For example going to a restaurant. We all have schemas for situations in our lives, but if something happens that does not normally occur then our schemas are now no good. Bartlett read Native American stories to white Americans. Later they were asked to recall the stories and they had changed them to fit their own schemas. Similarly Allport and Postman (1947) showed white subjects a picture of a white man attacking a black man. They then "played Chinese whispers" and the result was the description of the picture had been changed to a black man attacking a white man.
There are two types of schema:
Neisser (1981) studied the testimony of John Dean at the Watergate inquiry. He found that the testimony did not match what had actually been said. Dean had used his schemas of what usually happened in the Oval Office to unintentionally distort the truth.
Bransford and Johnson (1972) explained that if you do not have a schema then it is difficult to understand the information. For example, a passage about washing clothes was read to two groups. One group had the title (schema) "Washing Clothes" the other group was not given the schema. They were then required to recall as many pieces of information from the text. The group that did not have the schema recalled less items whereas the group that did, recalled many more.
Schank and Abelson (1977) suggested that each schema is a type of script for the event (eg. the restaurant script). The scripts enable us to remember situations and parts that are missing are implicit. This was demonstrated by Bower et al (1976). The subjects were told restaurant stories but the information was in the incorrect order. When they recalled the stories, they had been put in the correct order and extra bits had been put in that the subjects felt appropriate and fitted the script.
I think these theories do have some advantages, but they also have some criticisms. They do not tell us about how the schemas are formed or how they change. It is a vague concept that is not defined clearly and the theories over-emphasize the inaccuracy of memory (ie. Memories get changed to fit your schema, the theory does not explain being able to remember unusual, memories.)
Each of these methods have their highlights and downfalls. The imagery theory is quite effective but if the images are not vivid enough there is the risk that the information will not be recalled sufficiently. The Hierarchical network model seems to have the most problems as there is more evidence against it than there are for the theory. For example there are some examples that do not follow the model. The matrix idea has good recall but has the same kind of problems as the hierarchical network model. Feature models have good recall and the theory appears to work in most cases. The spreading activation theory has some valid points, for example, when you think of a piece of information, it often "sparks off" thoughts of other related pieces of information. Finally, schemas or scripts have some advantages but the theory is quite thin and vague.