MEMORY

Coding - the format in which information is stored in the various memory stores.

Capacity - the amount of information that can be held in a memory store.

Duration - the length of time information can be held in memory.

Short-term memory - the limited-capacity memory store. Coding is mainly acoustic (sounds) and capacity is between 5 and 9 items on average. Duration is only between about 18 and 30 seconds.

Long-term memory - the permanent memory store. Coding is mainly semantic (based on meaning). It has unlimited capacity and can store memories for up to a lifetime.

Once information gets into the memory system, it is stored in different formats depending on the memory store. The process of converting information from one form to another is called coding.

Alan Baddeley (1966) gave different lists of words to four groups of participants to remember:

• Group one: acoustically similar words such as cat, cab, can.
• Group two: acoustically dissimilar words such as pit, few, cow.
• Group three: semantically similar words such as great, large, big.
• Group four: semantically dissimilar words such as good, huge hot.

Participants were shown the original words and asked to recall them in the correct order. When they had to do this recall task immediately after hearing it (short-term memory recall), they tended to do worse with acoustically similar words.

If participants were asked to recall the word list after an interval of 20 minutes (long-term memory recall), they did worse with semantically similar words.

Digit Span: Joseph Jacobs (1887) developed a technique to measure digit span. The researcher gives 4 digits and then the participant is asked to recall these in the correct order out loud. If this is said correctly, the researcher reads out 5 digits and so on, increasing the number of digits each time. This is continued until the participant cannot recall the order correctly. This determines their digit span. Jacobs found that the mean span for digits across all participants was 9.3 items, the mean span for letters was 7.3.

Span of memory and chunking: George Miller (1956) made observations of everyday practice. He noted that things come in sevens: 7 notes on the musical scale, 7 deadly sins and so on. This suggests that the capacity of STM is about 7 items (plus or minus 2). However, Miller noted that people can recall 5 words as well as they can recall 5 letters. They do this by chunking - grouping sets of digits or letters into units or chunks.

Duration of STM: Margaret and Lloyd Peterson (1959) tested 24 undergraduate students and each student took part in eight trials. On each trial, the student was given a consonant syllable to remember and was also given a 3-digit numer. The student was then asked to count backwards from that 3-digit number until told to stop to prevent any mental rehearsal of the syllable (which could increase their memory). Their findings suggest that STM may have a very short duration unless we repeat something over and over.

Duration of LTM: Harry Bahrick and colleagues (1975) studied 392 participants from Ohio who were aged between 17 and 74. High school yearbooks were obtained from the participants. Recall was tested in various ways, including photo-recognition test of 50 photos and a free recall test where participants recalled all the names of their graduating class. Participants who were tested within 15 years of graduation were about 90% accurate in photo recognition. After 48 years, recall declined to about 70% for photo recognition and free recall was even worse. After 15 years, free recall was about 60% accurate, dropping to 30% after 48 years.

Coding:

• Artificial stimuli: Baddeley's study used quite artificial stimuli rather than meaningful material. The word lists had no personal meaning to participants, meaning we should be cautious about generalising the findings to different kinds of memory task.

Capacity:

• Lacking validity: Jacobs's study was conducted a long time ago and early research in psychology often lacked adequate control. This could mean the results may not be valid because there were confounding variables that were not controlled.
• Not so many chunks: Miller's research may have overestimated the capacity of STM. Cowan (2001) reviewed other research and concluded that the capacity of STM was only about four chunks.

Duration:

• Meaningless stimuli in STM study: in Peterson and Peterson's study, the stimulus material was artificial and does not reflect most real-life memory activities. We may say this study lacked external validity.
• Higher external validity: Bahrick et al's study has higher external validity as real-life meaningful memories were studied.

Richard Atkinson and Richard Shiffrin (1968, 1971) created the multi-store model which describes how information flows through the memory system.

SENSORY REGISTER --> SHORT-TERM MEMORY --> LONG-TERM MEMORY

Sensory register: a stimulus from the environment will pass into the sensory registers along with lots of other sights, sounds, smells and so on. The two main stores are called iconic memory (visual information is coded visually) and echoic memory (sound information is coded acoustically). Duration is less than half a second. The sensory registers have a high capacity. Very little of what goes into the sensory register passes further into the memory system.

Short-term memory: a limited capacity store and can only contain a certain number of 'things' before forgetting takes place. The capacity of STM is somewhere between 5 and 9 items of information, though research suggests is might be more like 5 rather than 9. Information in STM is coded acoustically and lasts about 30 seconds unless it is rehearsed. Maintenance rehearsal occurs when we repeat material to ourselves over and over again. If we rehearse it long enough, it passes into the long-term memory.

Long-term memory: permanent memory store for information that has been rehearsed for a prolonged time. Psychologists believe that its capacity is unlimited and can last for years. LTMs tend to be coded semantically. Although this material is stored in LTM when we want to recall it, it has to be transferred back into STM by a process called retrieval. None of them are recalled directly from LTM.

• Supporting research evidence: the MSM is supported by research studies that show that STM and LTM are indeed qualitatively different. Baddeley found that we tend to mix up words that sound similar when we are using our STMs but mix up words with similar meanings when we use our LTMs.
• There is more than one type of STM: the MSM states that STM is a unitary store, meaning there is only one type of short-term memory. Evidence from people suffering from amnesia shows that this cannot be true. Shallice and Warrington (1970) studied a patient with amnesia and found that his recall of digits was better when he was able to read them himself and other people with amnesia showed that there could be another short-term store for non-verbal sounds.
• There is more than one type of rehearsal: Craik and Watkins (1973) found that there are two types of rehearsal. Maintenance rehearsal is the type described in the MSM but this does not transfer information into LTM. Elaborative rehearsal is needed for long-term storage and occurs when you link the information to your existing knowledge or think about what it means. 
• Artifical materials: in everyday life we form memories related to useful things but a lot of research studies that provide support for the MSM use none of these things.
• There is more than one type of LTM: we have one long-term store for our memories of facts about the world and another for information such as how to ride a bike.

Episodic memory refers to our ability to recall events from our lives. This has been likened to a diary or a record of daily happenings.

These memories are complex and go in an order:

• First, they are time-stamped so that you remember when they happened.
• Secondly, your memory of a single episode will include several elements such as people and places and all of them are interlinked to produce a single memory.
• Thirdly, your have to make a concious effort to recall episodic memories. You may be able to do so quickly but you are still aware that you are searching for your memory.

This store contains our knowledge of the world. This includes facts and has been linked to a combination of an encyclopedia and a dictionary. It would include knowledge of things like: the taste of an orange, the meaning of words and so on.

Your semantic memory contains your knowledge of an impressive number of concepts.

These memories are not time-stamped and we don't usually remember when we first learned them. Semantic knowledge is less personal and more about facts we all share.

Semantic memory contains an immense collection of material that is constantly being added to.

This is our memory of actions, skills or how we do things. We can recall these memories without conscious awareness or a great deal of effort. A good example of this is driving a car.

Our ability to do this depends on procedural memory. 

These are the sorts of skills we might even find quite hard to explain to someone else. If you try to describe what you are doing as you drive the car, the task may become more difficult.

• Clinical evidence: the case studies of HM and Clive Wearing are relevant here. Episodic memory in both men was severely impaired as a consequence of amnesia, they had great difficulty recalling events that had happened to them in their pasts but their semantic memories were relatively unaffected. Their procedural memories were also intact. This evidence supports Tulving's view that there are different memory stores in LTM.
• Neuroimaging evidence: there is evidence from brain scan studies that different types of memory are stored in different parts of the brain. It was found that episodic and semantic memories were both recalled from the prefrontal cortex. It supports the view that there is a physical reality to the different types of LTM within the brain.
• Real-life applications: Belleville et al (2006) demonstrated that episodic memories could be improved in older people who had a mild cognitive impairment. The trained participants performed better on a test of episodic memory after training than a control group.
• Problems with clinical evidence: there is a lack of control of all sorts of variables in clinical studies which can reduce their validity.
• Three types of LTM or two?: Cohen and Squire (1980) disagree with Tulving's division of LTM and argue that episodic and semantic memories are stored together in one LTM called declarative memory.

The working memory model is an explanation of how one aspect of memory is organised and how it functions. The model consists of four main components, each of which is qualitatively different especially in terms of capacity and coding.

Central executive: essentially an attentional process that monitors incoming data, makes decisions and allocates systems to tasks. The central executive has a very limited processing capacity.

Phonological loop: deals with auditory information and preserves the order in which the information arrives. It is sub-divided into: the phonological store, which stores the words you hear and the articulatory process, which allows maintenance rehearsal to keep them in working memory. The capacity of the loop is believed to be two seconds worth of what you can say.

Visuo-spatial sketchpad: stores visual and/or spatial information when required. It has a limited capacity of around three or four objects. This is sub-divided into: the visual cache, which stores visual data and the inner scribe, which records the arrangement of objects in the visual field.

Episodic buffer: a temporary store of information, integrating the visual, spatial and verbal information processed by other stores and maintaining a sense of time sequencing - recording events that are happening. It can be seen as the storage component of the central executive and has a limited capacity of about four chunks. The episodic buffer links working memory to LTM and wider cognitive processes such as perception.

• Clinical evidence: Shallice and Warrington (1970) completed a case study of patient KF who had suffered brain damage. After this damage, KF had poor STM ability for verbal information but could process visual information normally presented visually - he had difficulty with sounds but could recall letters and digits. This supports the existence of a separate visual and acoustic store.
• Dual task performance: Baddeley et al (1975) showed that participants had more difficulty doing two visual tasks (tracking a light and describing the letter F) than doing both a visual and verbal task at the same time. This is because both visual tasks compete for the same systm, whereas, doing a verbal and visual task simultaneously, there is no competition.
• Lack of clarity over the central executive: Cognitive psychologists suggest that this component of the WMM is unsatisfactory and doesn't really explain anything. The central executive needs to be more clearly specified than just being simply 'attention'.
• Studies of the word length effect support the phonological loop: Baddeley et al (1975) demonstrated that people find it more difficult to remember a list of long words, this is called word length effect. This dissapears when a person is given an articulatory suppression task.
• Brain scanning studies support the WMM: Braver et al (1997) gave their participants tasks that involved the central executive while they were having a brain scan. The researchers found greater activity in an area known as the left prefrontal cortex. The activity in the area increased as the task became harder.

Some forgetting takes place because of interference. This occurs when two pieces of information conflict with each other, resulting in forgetting of one or both, or in some distortion of memory.

Interference has been proposed mainly as an explanation for forgetting in long-term memory.

Psychologists recognise two types of interference:

• Proactive interference occurs when an older memory interferes with a newer one.
• Retroactive interference happens when a newer memory interferes with an older one.

In both cases the interference is worse when the memories are similar, as discovered by John McGeoch and William McDonald (1931).

They studied retroactive interference by changing the amount of similarity between two sets of materials. They had to learn a list of 10 words until they could remember them with 100% accuracy. When the participants recalled the original list of words, their performance depended on the nature of the second list. The most similar material produced the worst recall which shows that interference is strongest when the memories are similar.

• Evidence from lab studies: thousands of lab experiments have been carried out into this exlanation for forgetting. Most of these studies show that both types of interference are very likely to be common ways we forget information from LTM.
• Artificial materials: the stimulus materials used in most studies are lists of words. This is still quite some distance from the things we learn and try to remember in everyday life - faces, birthdays etc. Interference may not be as likely an explanation for forgetting in everyday life as it is in a lab setting.
• Real-life studies: Baddeley and Hitch (1977) wanted to find out if interference was a better explanation for forgetting than the passage of time. They asked rugby players to try to remember the names of teams they had play, some players had missed games for up to three weeks. The results showed that accurate recall did not depend on how long ago the matches took place and much more important was the number of games they played between them.
• Time between learning: Studies are done over a short period of time and this does not reflect how we learn information in real life, which decreases the validity.
• Interference effects may be overcome using cues: Tulving and Psotka (1971) found that recall from a word list was 70% but they were given a cued recall test and the recall rose again to about 70%.

The reason people forget information may be because of insufficient cues.

When information is initially placed in memory, associated cues are stored at the same time.

If these cues are not available at the time of recall, it may make it appear as if you have forgotten the information but this is due to retrieval failure - not being able to access memories that are there.

Endel Tulving (1983) reviewed research into retrieval failure and discovered a consistent pattern to the findings. He summarised this pattern in what he called the encoding specificity principle.

This states that if a cue is to help us to recall information it has to be present at encoding (when we learn the material) and at retrieval (when we are recalling it). If the cues available at encoding and retrieval are different, there will be some forgetting.

Some cues are linked to the material-to-be-remembered in a meaningful way. For example, the cue 'STM' may lead you to recall all sorts of information about short-term memory.

Such cues are used in many mnemonic techniques.

Other cues are also encoded at the time of learning but not in a meaningful way. We will consider two examples of this: context-dependent forgetting and state-dependent forgetting.

Duncan Godden and Alan Baddeley (1975) carried out a study of deep-sea divers working underwater. In this situation it is crucial for divers to remember instructions given before diving underwater.

Procedure: The divers learned a list of words, either underwater or on land. They were then asked to recall the words either underwater or on land, which created four conditions:

• Learn on land - recall on land.
• Learn on land - recall underwater.
• Learn underwater - recall on land.
• Learn underwater - recall underwater.

Findings: Accurate recall was 40% lower in the non-matching conditions. The external cues available at learning were different ones at recall and this lead to retrieval failure.

Procedure: Sara Carter and Helen Cassaday (1998) gave anti-histamine drugs to their participants. The anti-histamines had a mild sedative effect making the participants slightly drowsy. This creates an internal physiological state different from the 'normal' state of being awake and alert. The participants were then asked to learn a list of words and passages of prose and then recall the information, again creating four conditions:

• Learn on drugs - recall when on drugs.
• Learn on it - recall when not on it.
• Learn not on drugs - recall when on it.
• Learn not on it - recall when not on it.

Findings: In the conditions there was a mismatch between internal state at learning and recall and performance on the memory test was significantly worse. So when the cues are absent then there is more forgetting.

• Supporting evidence: there is a range of research that supports the retrieval failure explanation for forgetting. The studies by Godden and Baddeley and Carter and Cassaday are two examples of this research.
• Questioning context effects: Baddeley (1997) argues that context effects are not actually very strong, especially in real life. Different contexts have to be very different before an effect is seen. 
• Recall versus recognition: the context effect may be related to the kind of memory being tested. Godden and Baddeley (1980) replicated their underwater experiment but used a regonition test instead of recall - participants had to say whether they recognised a word read to them from the list, instead of retrieving it for themselves.
• Problems with the encoding specificity principle: ESP is not testable and leads to a form of circular reasoning. If a cue does not result in successful recall of a word, then we assume that the cue was not encoded at the time of learning.
• Real-life applications: when we are having trouble remembering something, it is probably worth making the effort to try and recall the environment in which you learned it first. 

Eyewitness testimony (EWT): the ability of people to remember the details of events, such as accidents and crimes, which they themselves have observed. Accuracy of EWT can be affected by factors such as misleading information, leading questions and anxiety.

Misleading information: incorrect information given to the eyewitness usually after the events. It can take many forms, such as leading questions and post-event discussion between co-witnesses and/or other people.

Leading question: a question, which because of the way it is phrased, suggests a certain answer.

Post-event discussion (PED): occurs when there is more than one witness to an event. Witnesses may discuss what tey have seen with co-witnesses or with other people. This may influence the way the accuracy of each witness's recall of the events.

Procedure: Elizabeth Loftus and John Palmer (1974) arranged for participants to watch film clips of car accidents and then gave them questions about the accident. In the critical question (a leading question) participants were asked to describe how fast the cars were travelling: "About how fast were the cars going when they hit each other?". This is a leading questionbecause the verb 'hit' suggests the speed the car was going.

Findings: The mean estimated speed was calculated for each participant group. The verb contacted resulted in a mean estimated speed of 31.8mph. For the verb smashed, the mean was 40.5mph. The leading questions biased the eyewitness recall of an event.

Why do leading questions affect EWT?: The response-bias explanation suggests that the wording of the question has no real effect on the participants' memories, but just influences how they decide to answer. When the word smashed was used, the participants were encouraged to choose a higher speed estimate.

Loftus and Palmer (1974) conducted a second experiment that supported the substitution explanation - the wording of a leading question actually changes the participant's memory of the film clip.

When co-witnesses to a crime dicuss it with each other, their eyewitness testimonies may become contaminated. This is because they combine information from other witnesses with their own memories.

Procedure: Fiona Gabbert and her colleagues (2003) studied participants in pairs. Each participant watched a video of the same crime, but filmed from different points of view. This meant that each participant could see elements in the event that the other could not. For example, only one of the participants could see the title of a book being carried by a young woman. 

Both participants then discussed what they had seen before individually completing a test of recall.

Findings: The researchers found that 71% of the participants mistakenly recalled aspects of the event that they did not see in the video but had picked up in the discussion. In the control group, where there was no discussion, was 0%. Gabbert et al concluded that witnesses often go along with each other, either to win social approval or because they believe the other witnesses are right and they are wrong. They called this phenomenon memory conformity.

• Useful real-life applications: research into misleading information has hugely important practical uses in the real world, where the consequences of inaccurate EWT can be very serious. Loftus (1975) believes that leading questions can have such a distorting effect on memory that police officers need to be very careful about how they phrase their questions when interviewing eyewitnesses.
• The tasks are artificial: Loftus and Palmer's study used film clips of car accidents which is a very different experience from witnessing a real accident, mainly because clips lack the stress of a real accident. There is some evidence that emotions can have an influence on memory.
• Individual differences: there is evidence that older people are less accurate than younger people when giving eyewitness reports. Anastasi and Rhodes (2006) found that people in age groups 18-25 and 45-45 were more accurate than people in the group 55-78 years.
• Demand characteristics: Zaragosa and McCloskey (1989) argue that many answers participants give in lab studies of EWT are the result of demand characteristics. Participants usually do not want to let the researcher down and want to appear helpful and attentive. 
• Consequences of EWT: Foster et al (1994) point out that what you remember as an eyewitness can have some very important consequences in the real world, but the same is not true in research studies.

Anxiety has a negative effect on recall: Anxiety creates a physiological arousal in the body which prevents us paying attention to important cues so recall is worse. One approach to studying anxiety and eyewitness testimony (EWT) has been to look at the effect of weapons on accuracy of recall of the witness.

Prodecure: Johnson and Scott (1976) led participants to believe they were going to take part in a lab study. In the low-anxiety condition, after hearing an argument in the next room, a man then walked through the waiting area, carrying a pen with grease. Other participants heard the same argument but a man walked out of the room holding a paper knife that was covered in blood.

Findings: The participants later picked out the man from a set of 50 photos. 49% of the participants who had seen the man carrying the pen were able to identify him. The corresponding figure for the participants who had seen the man holding the blood-covered knife was for just 33%. The tunnel theory of memory argues that a witness's attention narrows to focus on a weapon.

The stress of witnessing a crime or accident creates anxiety thought physiological arousal within the body. The fight-or-flight response is triggered which increases our alterness and improves our memory for the even because we become more aware of cues in the situation.

Procedure: John Yuille and Judith Cutshall (1986) conducted a study of a real-life shooting in a gun shop in Vancouver, Canada. The shop owner shot a thief dead. There were 21 witnesses and 13 agreed to take part in the study. The interviews were held 4-5 months after the incident and these were compared with the original police interviews made at the time of the shooting. Accuracy was determined by the number of details reported in each account. The witnesses were also asked to rate how stressed they had felt at the time of the incident.

Findings: The witnesses were very accurate in their accounts and there was little change in the amount or accuracy after 5 monhts - though some details were less accurate, such as recollection of the colour of items and age/height/weight estimates. Those participants who reported the highest levels of stress wre most accurate (about 88% compared to 75% for the less-stressed).

Kenneth Deffenbacher (1983) applied the Yerkes-Dodson Law to EWT. 

Lower levels of anxiety produce lower levels of recall accuracy. But memory becomes more accurate as the level of anxiety experienced increases, just as you would expect from the graph.

However, there comes a point where the optimal level of anxiety is reached.

This is the point of maximum accuracy. If an eyewitness experiences any more stress than this, then their recall of the event suffers a drastic decline.

• Weapon focus effect may not be relevant: Johnson and Scott's study on weapon focus may test suprise rather than anxiety. Pickel (1998) conducted an experiment using scissors, a handgun, a wallet or a raw chicken as the hand-held items in a hairdressing salon video. Eyewitness accuracy was significantly poorer in the high unusualness conditions.
• Field studies sometimes lack control: researchers usually interview real-life eyewitnesses sometime after the event. Other things will have happened to the participants in the meantime that the researchers have no control over. The research does not control extraneous variables.
• There are ethical issues: creating anxiety in participants is very risky. It is potentially unethical because it may subject people to psychological harm purely for the purpose of research.
• The inverted-U explanation is too simplistic: anxiety is very difficult to define and measure accurately. One reason for this is that it has many elements. But the inverted-U explanation assumed only one of these is linked to poor performance - physiological arousal.
• Demand characteristics operate in lab studies: most lab studies show participants a filmed crime. Most of these participants will be aware they are watching a filmed crime for a reason to do with the study.

Ronald Fisher and Edward Geiselman (1992) argued that eyewitness testimony could be improved if the police used better techniques when interviewing witnesses. The cognitive interview is made up of four main techniques:

• Report everything: Witnesses are encouraged to include every single detail of the event, even though it may seem irrelevant or the witness doesn't feel confident about it. These may trigger more important memories.
• Reinstate the context: The witness should return to the original crime scene 'in their mind' and imagine the environment (such as what the weather was like, what they could see) and their emotions.
• Reverse the order: Events should be recalled in a different chronological order to the original sequence, for example, from the final point back to the beginning. This is done to prevent people reporting their expectations of how the even must have happened rather than actual events.
• Change perspective: Witnesses should recall the incident from other people's perspective. For example, how it would have appeared to other witnesses or to the perpetrator. This is done to disrupt the effect of expectations and schema on recall.

The enhanced cognitive interview (ECI): Fisher et al (1987) developed some additional elemants of the CI to focus on the social dynamics of the interaction. For example, the interviewer needs to know when to establish eye contact and when to relinquish it. The enhanced CI also includes ideas such as reducing eyewitness anxiety, minimising distractions, getting the witness to speak slowly and asking open-ended questions.

• The CI is time-consuming: police may be reluctant to use the CI because it takes much more time than the standard police interview. The CI also requires special training and many forces have not been able to provide more than a few hours.
• Some elements may be more valuable than others: Milne and Bull (2002) found that using a combination of report everything and context reinstatement produced better recall than any of the other conditions. This is a strength because it suggests that at least these two elements should be used to improve police interviewing of eyewitnesses even if the full CI isn't used.
• Support for the effectiveness of the ECI: research suggests that the enhanced cognitive interview may offer special benefits. Kohnken at al (1999) combined data from 50 studies. The enhanced CI consistently provided more correct information than the standard interview used by police.
• Variations of the CI are used: studies of the effectiveness of the CI use slightly different CI techniques or use the enhanced CI. The same is true in real life - police forces evolve their own methods.
• CI creates an increase in inaccurate information: the techniques of the CI aim to increase the amount of correct information remembered but the recall of incorrect information may also be increased. Kohnken et al (1999) found an 81% increase of correct information when the enhanced CI was compared to a standard infterview.