People often misremember, or forget completely. What can be deduced about the processes of memory on the basis of everyday memory failures?
People often misremember, or forget completely. What can be deduced about the processes of memory on the basis of everyday memory failures?
It would be difficult to think of a cognitive process which could be carried out without a memory system. However, people often complain that they often do forget the things which they would like to remember. Therefore an understanding of the way memory works is especially important when looking at why people misremember and forget. Because of this, much research has been carried out over the years on memory. Research in the 1970's tended to be based on the structural view of memory based on the modal model. This was later replaced with the idea that we have different types of knowledge with memory systems and memory processes. The most influential approach to what governs the complex pattern of remembering and forgetting has come from schema theory. This essay will look at how schema theory explains why people misremember and forget.
Schema theory was first proposed by Bartlett in 1932 and was largely ignored until the 1980's. Schema theory suggests that the information we have stored in memory is organized in such a way as to incorporate all the knowledge of a given type of object or event that we have acquired from past experience. Schemas are packages of information stored in memory representing general knowledge about all kinds of events, actions, objects or situations in the world around us. The knowledge about simple things such as the shapes of letters to more complex knowledge about events of everyday life are all stored in memory as schemas.
New information relevant to the schema is absorbed and remembered to provide a framework which can be added to whenever necessary. Information, which is not relevant to the schema in operation, may be disregarded and forgotten. The information in memory tends to be changed from the specific to the general. In this way, if we forgot something about an event which we had a schema for, we would tend to fill in the gaps in our knowledge about that event with what usually happens in a given situation. This is because schema theory emphasizes the fact that what we remember is influenced by what we already know. This is demonstrated by Loftus (1975) in an experiment which showed that new information is absorbed with memory representations which are already present.
In this experiment Loftus gave misleading information to eye-witness subjects after they had been shown a film of a car accident. The misleading information was then absorbed by the witnesses and integrated into their memory of the event. The misleading information had a significant influence on memory of the event as the fictitious information had been integrated with the real memory.
In a later experiment Loftus, Miller and Burns (1978) the correct information was altered and replaced with false information. Loftus showed in her experiments on eye witness testimony that the memory representation of an event can be modified by subsequent information. However in a further experiment, Loftus (1979), showed that in certain conditions the memory is more resistant to outside interference. For instance when it was obvious that the information was misleading, witnesses were less likely to take the fictitious information on board, and were more likely to be on their guard against any further attempts to mislead.
The most important point to come out of this research is that once the original memory has been tampered with, the original memory is not very likely to be recovered at any stage. This has implications for all theories about memory and not just eye-witness testimony.
Another area of research which is relevant was produced by Harris and Monaco (1976). They showed that pragmatic implications affect the way that information is stored in memory. For instance when information is given it is comprehended in such a way as to include what was directly asserted as well as what was already stored as a schema for that situation. Harris (1978) investigated how pragmatic implications might affect members of a jury by misleading them to believe that something which was implied had been asserted as definitely true. He concluded that it is very easy to mislead ...
This is a preview of the whole essay
Another area of research which is relevant was produced by Harris and Monaco (1976). They showed that pragmatic implications affect the way that information is stored in memory. For instance when information is given it is comprehended in such a way as to include what was directly asserted as well as what was already stored as a schema for that situation. Harris (1978) investigated how pragmatic implications might affect members of a jury by misleading them to believe that something which was implied had been asserted as definitely true. He concluded that it is very easy to mislead juries in this way. This shows that when remembering verbal information, particulars which were only implied may be confused with facts that were actually related.
Research on eye-witness testimony has tended to concentrate on the way that memory can be altered and so only gives one side of the argument. The witnesses were not only integrating prior knowledge from schemas which they held, but were also combining information from two different external sources, namely the observed event and the verbal information about that event. This shows that we can deduce that memory is not all schema based, but can sometimes be a combination of different sources of information.
Another area which schema theory has been used to demonstrate how the processes of memory work is in the area of everyday slips and errors. Absentminded slips of action occur when the schema for the wrong action sequence is activated instead of the schema for the correct action sequence. Slips and errors were investigated by Reason (1979). He suggested that slips and errors tend to occur when carrying out everyday tasks which are highly practiced and have become automatic. Automatic actions differ from actions which have to be well attended to. This is because we have to pay attention to things which we do not know how to do very well, whereas those things which we do everyday become automatic with practice. This means that attention can be on executing another activity while performing the one which has become automatic, and in this way errors and slips can occur.
Norman (1981) suggested that schema theory can account for slips of action. Norman's model is one where schemas work together in organized groups which he called hierarchical organization. The same principle of schemas applies but instead of events or places, the schema represents knowledge about motor actions. The causes of errors in both Reason's and Norman's examples are not very distinct. Reason suggests that absentminded slips of action occur when the schema for the wrong action sequence is selected instead of the correct one. Whereas Norman's model indicates that an error could occur when there is faulty specification of the overall intention, faulty triggering, or faulty activation of the schemas.
This is demonstrated by Brewer and Treyens (1981) in which they showed that people's memory for a scene is affected by the schema which is appropriate for that scene. Their experiment consisted of leaving people in an office and later asking them to recall the things that were present in the office. They showed that while people successfully recalled items which they would have expected to have been in the office, some subjects made errors by recalling items which they would have expected to be in an office, but in reality were not there. This shows that the subjects were using there schema based knowledge to try to recall what was in the office. What is more surprising is that a significant number of subjects recalled accurately that there was a skull in the office, this shows that recall is not entirely schema based. Brewer and Treyens show that schema theory is not very useful for explaining how people actually remember things in everyday activities.
In conclusion it could be said that schema theory definitely has its strong points and can be used to demonstrate how people do forget and misremember. Schema theory and can be used to explain many things such as how relatively easy it is to confuse eye-witnesses, and why we make slips and errors in everyday tasks. On the other hand, what it is not so good at, as Brewer and Treyens pointed out, is in explaining how we do remember odd events, such as the skull in the office. It should be remembered, however, that schema theory can account for the everyday slips and errors which occur when people 'fill in the gaps' with their own knowledge by using the schemas they have for that particular event or action.
References
G. Cohen, G. Kiss & M. Le Voi. Open Guides to Psychology: Memory Current Issues (2nd edition)
The model of person recognition proposed by Bruce and Young (1986) suggested that there are independent routes for the processing of expressions, facial speech and identities. They also suggested that there is a sequence of stages within the identification route. The sequence of stages which they proposed for identification stood up to an attempt at falsification. However, the phenomenon of covert recognition in prosopagnosia suggests that the stage model of identification offered by Bruce and Young requires modification.
To examine whether the evidence supports the model proposed by Bruce and Young it is necessary to look closely at the model to see exactly what it states. The model put forward by Bruce and Young (1986) was a stage model which put the stages of facial recognition into the broader context of other uses made of facial information.
The task of recognising individual identity from a face is logically separate from that of recognising an emotional expression from a face
The component stages of person identification appear to be related and move in sequence. The stage model of Bruce and Young is similar in structure to that of the bottom up model of object recognition proposed by Humpheys et al (1988). In Humphreys model the first stage of recognition consists of matching the structural characteristics of the object to a stored structural description. This is then assessed in the second stage, semantic representation. Then the result of these two stages is signalled to the name stage so that a name can be accessed for the object.
The way that the Bruce and Young model suggests faces are processed is that when we look at a face which is familiar, information from the face recognition unit should become available. This is the equivalent of the first stage of Humphreys model. This should be followed by information about the person which is available from the person identity node, which is equivalent to Humphreys second stage. This information includes such things as their occupation or where they were first encountered or anything about the person which is relevant. The persons name is then retrieved from the name generation component.
This means that if a person is familiar then they should be recognised faster than decisions based on person identity nodes. This was found to be the case by Young et al (1986a) who found that subjects were able to decide that a face belonged to a politician much faster than they were able to name the person.
It is important to make clear the relationship between identification of people from their faces and other uses made of facial information in interpersonal communication. Facial expression analysis and lip reading, which Bruce and Young call facial speech, are independent of each other, whereas the component stages of person identification appear to be related and to proceed in sequence.
Facial recognition is an extraordinary development of the ability to recognise individuals within the same subordinate level. It involves within category discriminations categorising differently patterns which all share the same overall structure.
Structural encoding is where the evaluation is made which is either a view centred or an expression centred description.
Expression analysis is where an individuals emotional state can be determined from looking at the information obtained from the facial features.
Facial speech analysis is what is usually known as lip reading and is the information obtained from the detailed observation of a speaker's lip movements.
Directed visual processing is to determine what the facial features are like. Specific facial information may be processed selectively.
Face recognition units contain the information required to recognise people who are known to the viewer.
Person identity nodes provide information about the person such as their interests. friends and other relevant information.
Name generation is stored separately from other information.
The cognitive system is where all the additional information needed in facial recognition is contained, for instance that actors and actresses tend to have attractive faces. The cognitive system also plays a part in determining which of the components receive attention.
Experimental evidence
The main theoretical assumption made by Bruce and Young was that familiar and unfamiliar faces are processed in different ways. Some components of the model can be used with both familiar and unfamiliar faces whereas other components can only be used with familiar faces. For instance, face recognition units, name generation and person identity nodes can only be used with familiar faces, and facial speech analysis, directed visual processing can be used with both familiar and unfamiliar faces.
Malone, Morris, Kay and Levin (1982) found that there were patients who showed good recognition of familiar faces but poor recognition of unfamiliar faces and other patients with the reverse pattern. One patient showed the ability to recognise photographs of famous statesmen but could not match unfamiliar faces. A second patient was able to match unfamiliar faces but had great difficulty recognising famous people from photographs. From this it would seem that there is evidence for the different processes involved in recognition of familiar and unfamiliar faces.
Another area in which it is possible to test the theory of Bruce and Young is the area of name recollection. The way that they suggest name generation happens is that the name would not be able to be recalled without other information about the person being available at the same time, for instance their job or where they were seen in the past, etc.. Young, Hay and Ellis (1985) found that of the people that they studied not one could recall a name without other information about the person being available at the same time (diary studies) page 68 According to the model another problem which should be encountered when the face recognition unit is activated but the person identity node is not, is that of the feeling of familiarity alongside the inability to recall any other relevant information about the person. This was found to be the case on 233 occasions in the Young et al studies. This evidence is also supported by Flude, Ellis and Kay (1989) who found that a patient EST was able to name the occupations of 85% of very familiar people when shown photographs of familiar people but could only recall 15% of their names.
The model, although influential, was not very specific in some areas. Bruce and Young admitted that the component labelled "cognitive system" was a place where all the aspects of processing not reflected in other components of the model was put. Also, the way that the model suggests unfamiliar faces are processed is not as detailed as the processing of familiar faces.
One of the features of the model is that there is a separate store for names which can only be accessed with the information from the person identity node. Evidence contradicting this was put forward by Haan, Young and Newcombe (1991) who studied a patient with amnesia who was able to recall the names of 88% of famous people, but was unable to recall any autobiographical information. This should not have been possible according to the model because the inability to retrieve autobiographical information should have prevented her from recalling the names.
Burton, Bruce and Johnston (1990) and Burton and Bruce (1992) revised and developed the Bruce and Young model in a number of ways. They suggest a model which has three main pools of information. These are Face recognition units, person identity nodes and semantic information units. Within these three pools all the other information about the person is stored. The face recognition unit is activated by the presentation of any familiar face and is view independent. The person identity nodes are domain and modality free paths into semantic information. The information about the individual, such as name and occupation are stored in the semantic information units.
There are links between the pools which are bi-directional, but within each pool the units are linked to the others by inhibitory connections. The main differences between this model and the original one is that there is now no separate name store and information about the individual is stored in the same way as other information about the person. The way that names are stored is poorly integrated with the other semantic information because it is unique to the individual concerned and is therefore difficult to access. Familiarity judgements are made at the person identity nodes rather than at the face recognition unit, this means that it is the person rather than the face which is recognised.
The whole model is now more precise because in place of the generalities of the original model Burton et al have provided a detailed interactive model.
There are several advantages of the revised model in that the findings of Haan et al can now be accounted for. The fact that names are harder to remember is also accounted for in the revised version because name information is poorly integrated with other information. According to the revised version, names might be easier to remember than other relevant information pertaining to the person if they were more semantically integrated. This could not happen with the original model. Cohen (1990) found that faces produced better recall of names than occupations where the names were meaningful and the occupations meaningless.
The revised model is easier to evaluate because it is more explicit than the original model. This enabled Burton et al (1990) to produce a computer simulation of the revised model which led to improvements in the model which could also be tested empirically.
7
