Bartlett used a story "The War of Ghosts" to test his theory: this story is a North American folktale which has subject matter that would be unfamiliar to most of his P's. Bartlett read the story and then tested P's memory of the story. He found that:
Loftus (see study sheet) has conducted many studies to try and understand the reliability of memory in the context of EWT. In one study P's were shown film of a traffic accident, and then asked questions about it. One question was "how fast were the cars going.."; half of the P's were asked "..when they hit each other", and the other half "..when they smashed into each other". Subsequently, when asked to estimate the speed of collision, the "smashed" group estimated a higher speed than the "hit" group. Additionally, when asked whether there had been any broken glass in the film (there hadn't), the "smashed" group were more likely to "remember" having seen glass.
This suggests that the schema being used to reconstruct a memory is influenced by the way questions were put; P's in the "smashed" group were confabulating; that is they honestly believed that they'd seen broken glass.
Nolan & Markham (1998) showed that confident witnesses are more likely to be seen as credible, and so an important implication of Loftus' research is that questioning of eyewitnesses must take account of the consequence of "leading" questions. However, Loftus' experiments could be criticised on the grounds of lack of mundane realism. Following on from this, Geiselman (1985) devised a technique called the cognitive interview and found that police officers using this technique obtained more correct answers than was the case with previous techniques.
Although recognising people by their faces is something people commonly do without having to think about it, the processes involved in face recognition are not simple. For example, a person's face my change quite significantly over a period of years and yet still be recognisable. Similarly, we can recognise faces even though they are capable of showing many different expressions.
People who sufferer from prosopagnasia are unable to recognise faces although they can recognise objects, which suggests that face recognition may be a distinct, or at least specialised, function in the brain. EWT often relies on face recognition, and so the study of this subject has important implications in understanding how reliable memories of faces are.
Two alternative suggestions as to how face recognition works are:
Evidence supporting the holistic view comes from a study by Young, Hellawell & Hay (1987) who showed P's a set of photographs that had been spliced so that the top half of one face was combined with the bottom half of another. P's had more difficulty recognising such "collage" versions than they did when shown just one or other half at a time. This suggests that when presented with the "collage", P's saw a "whole face" and were unable to recognise it.
Evidence supporting the other view comes from Valentine & Bruce (1986) who showed P's a series of photographs and required the P's to state whether or not the face was famous as quickly as possible. This study found that responses were much quicker when the face had a distinctive feature.
Both of these studies could be criticised on the grounds of lack of mundane realism, but taken together suggest that there are aspects of both holistic and feature recognition going on.
Platz & Hosch (1988) performed a study where they had stooges of different ethnic groups visting shops; subsequently the experimenter visited the shops and asked the shop-keeper if he could identify the shopper. The findings of this experiment were that the chances of recognition were highest when the shopper was of the same ethnic group as the shop-keeper, or in other words, "all Chinese look the same" to someone who's not Chinese.
Answer one question from this Section. You should attempt all parts of the question you choose. Notes/books may be used.
Reconstructive memory was investigated by Bartlett, who suggested that memory works not simply as a passive recording device, but that individuals apply a schema based on their own personal experience to information in their memory, so that when they recall information it tends to conform to previous experience and pre-conceptions.
One problem with this explanation is that it is very difficult to test experimentally, and that such experiments as have been done (e.g. Waugh & Norman) do not have high ecological validity. .
The main strength of the multi-store model was the research that it inspired into memory. The model also has significant intuitive appeal, since it ties in with personal experience of how we remember information. Various experiments have confirmed that it is a useful way to view how memory works. Also, case studies do support the multi-store model: Milner studied a patient who suffered from anterograde amnesia following a brain operation; this patient was apparently able to form new STMs but not able to form new LTMs. However more recent research does suggest that the multi-store model may be an overly simplistic view.
Miller, when investigating STM capacity, decided that its capacity averaged seven "chunks", where a "chunk" is a unit of information. While in the simplest case a chunk may be a single digit (which would allow someone to store a 7-digit telephone number in STM), Miller showed that it may be a series of digits, or words, or even phrases, so long as the series could be organised in some way. For example, subjects could apparently store seven separate "years" in STM (e.g. 2001, 1776) where each year consisted of more than one digit. The information required to organise information into chunks must presumably come from LTM, which is not something that the multi-store model describes.
An alternative model of memory, suggested by Craik & Lockhart is the Levels of Processing model. In this model it is suggested that information is more likely to be remembered if it is given some meaning, and they performed a series of experiments which support this hypothesis. This offers an alternative explanation to the "chunking" phenomenon, and their experiments seem to show that "rehearsal" is not a very effective way of remembering (compared to assigning some meaning to the information).
Perhaps one of the main weaknesses for the multi-store model is that much of the research into this model has been carried out in experimental settings which do not correspond to the everyday way in which humans commit information to memory.
Back to class notes index page