An Introduction to Applied Cognitive Psychology - David Groome, Anthony Esgate, Michael W. Eysenck (2016)

Chapter 7. Everyday memory

David Groome


Scientific studies of memory have been taking place for well over a century, but until fairly recently most memory research was restricted to laboratory experiments carried out under highly artificial conditions. It is only in the past few decades that psychologists have begun to investigate memory function in everyday life. As explained in Chapter 1, Neisser (1976) argued that cognitive psychologists should carry out research in real-life settings. This plea for ‘ecological validity’ led to a new interest in the study of everyday memory, and this approach has grown over the years as a body of research that is largely separate from laboratory studies and yet complementary to it. Such real-life studies provide a ‘reality check’ for the accepted laboratory findings, offering an opportunity to confirm or disconfirm their validity in the real world. Real-life memory studies can also provide findings that may be applied in the real world, and they can sometimes identify memory phenomena that have not emerged from laboratory experiments.

Conway (1991) points out that there are certain fundamental differences between memory performance in real-life settings and memory performance in a laboratory experiment. One important difference is that memory in real life tends to involve personal experiences that hold considerable significance for the individual, whereas the test items presented in a lab experiment are usually of little interest to them. Conway and Jobson (2012) argue that memory in real life also tends to be goal directed, whereas memory in the laboratory usually has no purpose from the participant’s point of view.

Koriat and Goldsmith (1996) make the observation that laboratory experiments usually involve quantitative measures of memory (e.g. the number of words recalled from a list), whereas in real life there is more emphasis on qualitative aspects of a memory trace (e.g. recalling your holiday experiences). A similar point is made by Neisser (1996), who notes that participants in lab experiments are usually required to retrieve as many test items as possible, whereas in real life memory may involve more selective retrieval and more personalised motives. For example, sometimes we may wish to recall events that will be helpful or reassuring to us, or incidents that we can use to impress or amuse other people. These motives will often result in a tendency to remember selectively rather than simply aiming for maximum recall.

All of these factors should be borne in mind as we examine the research on various types of everyday memory, starting with autobiographical memory, which is the store of memory we all have for the events and experiences of our own lives.



Autobiographical memory refers to our memory for the events we have experienced in our own lives. Williams et al. (2008) suggest that autobiographical memory has three main functions. First, there is the directive function, meaning that past experiences are used to help direct our subsequent behaviour and decisions. Second, there is the social function, whereby our memories help with our social interactions and group cohesiveness. Third, there is the self function, providing us with a sense of identity, enabling us to understand who we are and our place in the world.

Since autobiographical memory concerns our personal experiences, the usual lab techniques for studying memory (e.g. recalling word lists) are not appropriate, and researchers have had to develop new testing methods. Autobiographical memory is sometimes tested by free recall, but more often participants are provided with a retrieval cue of some sort, which provides more control over the type of items to be retrieved. The first experiments on autobiographical memory made use of verbal cues, starting with the ‘cue-word’ technique, which was introduced in the earliest days of psychology by Galton (1879). Photographs have also been used as retrieval cues (Bahrick et al., 1975).

You can easily test your own autobiographical memory. For example, try to write down the names of all of the children in your class at primary school. Better still, go and fish out an old school photograph like that in Figure 7.1, and see how many of the children you can name.

These tasks require you to draw upon memories that have probably remained undisturbed for many years, and yet you will probably be surprised how many names you can produce. You will also probably find that many other related memories are dragged up along with the names you recall, because of the interconnections between these related memories.

A more scientific version of this experiment was carried out by Bahrick et al. (1975), who investigated the ability of American adults to remember their old high school classmates. Bahrick et al. used a number of different types of test, including the accuracy with which subjects could recall names to match the faces in college photographs (the ‘picture-cuing’ test) and their ability to match photos with a list of names supplied to them (the ‘picture-matching’ test). On the picture-matching test they found that most of their participants could still match up the names and photos of more than 80 per cent of their college classmates even 25 years after graduation, and there was no significant decline in their performance over the years since the actual period when they had graduated. Although scores did drop off slightly at longer retention intervals, they still remained above 70 per cent despite an average time lapse of 47 years. The picture-cuing test not surprisingly yielded lower overall recall scores since names had to be generated by each participant spontaneously, but performance was still quite impressive and again showed relatively little decline over the years. Bahrick et al. concluded that memory for real-life experiences is often far more accurate and durable than memory for items tested in a laboratory experiment. In fact Bahrick et al. suggested that some of our autobiographical memories are so thoroughly learnt that they achieve ‘permastore’ status, and remain intact indefinitely.


Figure 7.1 An old school photograph, taken in 1956. The author is three rows back, somewhat left of centre, and with a cross drawn on his chest (NB the author did not actually go around with a cross on his chest; he added the cross with a pen at some later date).

More recently, Bahrick et al. (2008) asked older people to try to recall the grades they obtained in college. Despite the passage of about 50 years since their college days, they were able to recall their grades with 80 per cent accuracy. Again it would appear that memory performance in real life tends to be far more lasting than in laboratory experiments.

This durability has not generally been found in laboratory studies, where forgetting is typically found to be far more rapid. One possible explanation for this discrepancy may be the fact that (as mentioned in Section 7.1), autobiographical memories have far greater personal significance to the individual than do the test items in a lab experiment (Conway, 1991). Autobiographical memories also tend to benefit from more extensive learning periods. For example, you may have spent several years with your school friends, but in a typical lab experiment you would have only a few minutes to learn the test material.

One reason for studying autobiographical memory is to find out whether it conforms to the same general principles that we find in laboratory studies of memory, and Bahrick et al.’s discovery of a relatively permanent store of personal memories does indeed differ from the usual laboratory findings.


One of the biggest problems with the testing of autobiographical memory is that we do not usually have a precise and detailed record of the events an individual has experienced during their life. This means that we do not know what we can reasonably expect them to remember, and we cannot easily check the accuracy of the events they may recall. A person may appear to remember events from the past, but it is entirely possible that the memories they report are incorrect. In an effort to overcome these problems, some investigators have deliberately kept detailed diaries of their own daily experiences over long periods, thus providing a suitable source of memories to be tested later, which could be checked for their accuracy.

Linton (1975) used this diary technique, noting down two or three events every day over a 6-year period. At the end of each month she chose two of those events at random and attempted to recall as much as possible about them. Linton found that memories grew weaker as time went by, and that memories for similar events or repeated experiences became harder to distinguish from one another, merging together to form a more general composite memory. Linton also discovered that items were far more likely to be recalled if they had been tested on previous occasions. This finding is consistent with the ‘testing effect’ reported in laboratory studies (Allen et al., 1969; Carpenter et al., 2008), which was covered in Chapter 6.

Wagenaar (1986) used a diary technique similar to Linton’s, again recording daily events over a 6-year period. However, he took the additional precaution of recording retrieval cues for later use. Aided by these retrieval cues, Wagenaar was able to recall about half of the events recorded over the previous 6 years. His study also revealed that the likelihood of retrieving an item depended on the number and quality of the retrieval cues available, a finding that is broadly consistent with the encoding specificity principle proposed by Tulving and Thomson (1973).

Wagenaar found that the most effective retrieval cues were those that related to the details of the actual event (Who? Where? What?), but not its timing (When?). For example, you will probably find it easy to recall the last time you saw a close friend or visited a restaurant, but if I asked you what you did on the 15th of June you probably won’t have any idea.

Both Linton and Wagenaar noted that their recall of past events showed a strong bias towards the recall of pleasant events rather than unpleasant ones. There are a number of possible explanations for this retrieval bias. Psychoanalytic theory suggests that we tend to repress our more unpleasant memories as a form of defence mechanism, to protect us from distressing thoughts (Freud, 1938). An alternative theory is that unpleasant memories are often acquired in stressful situations, which may tend to inhibit memory input (Williams et al., 1988; Hertel, 1992). A third possibility is that people prefer to think about pleasant memories when they are reminiscing about the past, so pleasant memories are likely to benefit from more frequent retrieval and rehearsal than unpleasant memories (Searleman and Herrmann, 1994).


Crovitz and Schiffman (1974) used the cue-word approach to investigate whether there were certain periods in a person’s life that were more likely to stand out in their memory. They found that their participants were far better at recalling events from the recent past than from the distant past, and indeed there was a roughly linear relationship between the amount retrieved from a given year and its recency. However, this study was carried out on relatively young adults. Rubin et al. (1986) found a somewhat different pattern when older subjects were tested. People in their seventies tended to recall a large number of events from their early adult years, especially events they experienced between the ages of 10 and 30. This phenomenon is sometimes referred to as the ‘reminiscence bump’, as it appears as a bump on the graph of retrieval over time (see Figure 7.2).

Rubin et al. (1998) found that the reminiscence bump not only occurred for personal events but also for more general public events such as news items, books and academy award winners. Schulkind et al. (1999) found that older people were also better at recognising songs that were popular in their youth rather than those from more recent times. They also rated those older songs as more emotional, which may help to explain their heightened memorability. Janssen et al. (2012) found that people even showed a reminiscence bump when asked to name their favourite footballers, as they would usually choose players who were active during the early adulthood of the respondent.

One possible explanation for the reminiscence bump is that an older person may find their earlier years more memorable because they were more eventful and involved more novel experiences. Most people have fairly vivid memories for their first trip abroad, their first date or their first marriage, whereas subsequent dates or trips abroad (or marriages) tend to lose their novelty value and thus become less distinctive memories. Pillemer et al. (1988) found that novel experiences do tend to be particularly memorable, and of course the young adult stage of life involve far more of these novel experiences.


Figure 7.2 The reminiscence bump. Retrieval scores for personal autobiographical events from different periods of a person’s life (Rubin et al., 1986).

Source: reprinted with permission from Cambridge University Press.

It is also possible that memories from early adulthood are remembered because they are more pleasant, and a study by Gluck and Bluck (2007) confirmed that the reminiscence bump does tend to involve mostly happy memories. Since older people are more likely to prefer reminiscing about the happier moments of their younger days, these pleasant memories are likely to benefit from frequent retrieval.

Conway et al. (2005) compared the characteristics of the reminiscence bump across five different cultures (Britain, USA, China, Japan and Bangladesh). They found broadly similar reminiscence bumps in all five of these national groups, but with slight variations that seemed to reflect cultural differences. For example, Chinese participants tended to recall more events of collective or cultural significance, whereas Americans tended to recall more events of personal significance.

Chu and Downes (2000) found that odours could act as powerful cues to the retrieval of events from early life. This finding has been referred to as the ‘Proust phenomenon’ as it reflects the observations of Proust about the evocative nature of odours (for a scientific evaluation of Proust’s account, see Jones, 2001). Chu and Downes also reported a marked reminiscence bump, but noted that memories related to odours peaked at around 6–10 years of age, whereas for verbal cues the peak occurred between 11 and 25 years of age.

Koppel and Berntsen (2015) identified two separate reminiscence peaks, depending on which type of retrieval test was employed. With a cued recall test they found a reminiscence bump around the age period of 8–22 years, whereas the peak occurred at 15–28 years with an uncued test, in which participants were simply asked to list the most important events of their lives. At present it is not clear whether this double peak merely reflects two different testing procedures or whether it indicates two distinct reminiscence peaks with different causes.

Most people have some regrets about some of their actions as they look back on their lives, but a study by Gilovich et al. (2003) has shown that the types of regret we feel tend to vary according to the part of the lifespan being reviewed. They found that when considering earlier periods of their lives, people are more likely to regret things they have not done, rather than regretting things they have done. However, when reviewing more recent actions and decisions, the opposite is usually found, and people tend to regret their actions rather than their inactions.


Early infancy is one period of life that appears to be particularly difficult to remember. In fact most adults appear to remember nothing at all from the first 2 or 3 years of their lives (Waldfogel, 1948; Pillemer and White, 1989). This phenomenon is known as ‘infantile amnesia’, and there are a number of theories about its possible causes. One possible explanation is the fact that brain structures involved in memory storage have not completed their physical development in early infancy, such as the prefrontal cortex (Maguire et al., 2001) and the hippocampus (Richmond and Nelson, 2007). Josselyn and Frankland (2012) note that the period of infantile amnesia corresponds with a time when many new hippocampal neurons are developing. They suggest that the growth of new neurons may interfere with the formation of new memories, possibly by replacing existing neurons and neural connections. However, these findings offer only part of the explanation of infantile amnesia, as they do not tell us which aspects of memory processing are missing in young children.

There is clear evidence that young infants are actually capable of creating new memories, but these memories somehow become inaccessible over the next few years. Nelson (1988) found that 2-year-old children were able to register and retrieve information, which subsequently became inaccessible. A possible reason for this subsequent forgetting is that young children have a restricted form of memory for previous events. Nelson and Ross (1980) reported that very young children are able to remember general facts (i.e. semantic memory) but not specific events (i.e. episodic memory). Their earliest memories thus tend to be based on schemas and scripts for general or typical events, but not for specific episodes of their own personal lives. Bauer and Larkina (2013) confirmed that young children recall few details of time, place or the significance of a memory, and that more detailed memory representations did not appear until about 7 years of age.

Newcombe et al. (2000) argue that young infants retain implicit memories, which can affect their later behaviour but without any conscious recollection of the original causative event. Newcombe et al. suggest that a possible reason for the inability of young infants to form explicit episodic memories may be the incomplete development of the prefrontal cortex, which is known to be important in the formation of episodic autobiographical memories.

Another possible cause of infantile amnesia is that young children lack the language skills required to verbalise memories in later years, and there is evidence that younger children do perform very poorly on verbal memory tests but are far better at non-verbal memory tests (Simcock and Hayne, 2003).

An interesting cross-cultural study by MacDonald et al. (2000) compared childhood recall among three different New Zealand sub-cultures. When asked to report their earliest memories, NZ Maoris were able to report memories from earlier in their lives than NZ Europeans or NZ Asians. This finding suggests that early memory formation (or at least the tendency to report it) may be affected by cultural influences, which in this case might possibly be related to the heightened significance accorded by Maoris to the past.

Another possible explanation of infantile amnesia is the suggestion of Howe and Courage (1997) that children do not develop a ‘sense of self’ until the age of about 20 months, as indicated for example by their inability to recognise themselves in a mirror or photograph. Howe and Courage argue that this kind of self-identity may be crucial for the formation of personal autobiographical memories, which are characterised by their reference to the self. Morrison and Conway (2010) have shown that the earliest formation of lasting autobiographical memories seems to coincide with the point in childhood where children are starting to form a sense of their own identity.



It has often been said that most Americans remember what they were doing at the moment when they heard the news of the assassination of President Kennedy (Figure 7.3), because it came as such a terrible shock to the entire nation. Brown and Kulik (1977) decided to investigate this claim scientifically, and they found that all but one of the eighty participants they tested were indeed able to report some details of the circumstances and surroundings in which they heard the news of Kennedy’s death.

Similar findings have been reported for a range of other major news events, including the explosion of the space shuttle ‘Challenger’ (Neisser and Harsch, 1992), the death of Princess Diana (Davidson and Glisky, 2002; Hornstein et al., 2003), the terrorist attack on the World Trade Center (Talarico and Rubin, 2003; Candel et al., 2003) and the deaths of Michael Jackson and Osama Bin Laden (Demiray and Freund, 2015).

The capacity of an important and shocking event to illuminate trivial aspects of the observer’s current activities and surroundings is known as ‘flashbulb memory’. The fact that a major news event is itself well remembered is hardly surprising, but the significance of flashbulb memory is that people are also able to remember trivial details of their own lives at the time of the event, such as where they were and what they were doing. These trivia of daily life are in some way illuminated by the simultaneous occurrence of a highly significant and shocking event, hence the term ‘flashbulb memory’.


Figure 7.3 President John F. Kennedy shortly before he was assassinated.

Source: photo by Rolls Press/Popperfoto.


In an effort to explain the occurrence of flashbulb memory, Brown and Kulik (1977) suggested that a special memory process might be involved, which is fundamentally different from the mechanism involved in normal memory. This hypothesis was based on Brown and Kulik’s observation that flashbulb memories appeared to be not only remarkably accurate but also immune to normal forgetting processes. This special process was assumed to be activated only by an event that was very shocking, and it was thought to create a permanent and infallible record of the details relating to that event. It was assumed that such a memory process might have evolved because it would offer a survival advantage, by enabling an individual to remember vivid details of past catastrophes that would help them to avoid similar dangers in the future.

The notion of flashbulb memory as a special process has been challenged by studies showing that flashbulb memories actually seem to be subject to errors and forgetting just like any other type of memory. Researchers have been able to demonstrate this by testing their participants’ memories immediately after a disaster, in order to provide a baseline measure for comparison with later tests of flashbulb memory. For example, Neisser and Harsch (1992) tested people the day after the ‘Challenger’ explosion, to establish precisely what they could recall at that initial stage. The same people were tested again 3 years later, and a comparison of these results with the initial test data revealed that their flashbulb memories were by no means immune to forgetting over this time period. In fact, roughly half of the details recalled in the 3-year retest were inconsistent with the information recalled on the day after the crash.

A number of subsequent studies have confirmed the fallibility of flashbulb memories. The announcement of the verdict in the O.J. Simpson murder trial generated flashbulb memories in a sample of American participants (Schmolk et al., 2000), but these memories suffered a rapid drop in their accuracy over the months following the verdict. Talarico and Rubin (2003) reported that flashbulb memories following the World Trade Center attack (Figure 7.4) also showed a decline in their accuracy over the months that followed, and were in fact no more accurate and lasting than the normal everyday memories of their participants for other occasions unrelated to any kind of disaster.


Figure 7.4 The World Trade Center attack.

Source: photo by Robert J. Fisch.

Conway et al. (1994) argued that the fallibility of flashbulb memories reported in many of these studies might simply reflect a lack of interest in the key event in some of the participants tested. Conway et al. suggested that flashbulb memories might only occur in individuals for whom the event in question held particular personal significance, especially for those who perceived the event as having major consequences for their own lives. For example, many Americans would have perceived the Kennedy assassination or the attack on the World Trade Center as having major consequences for them personally, since these events were likely to have a major impact on life in America thereafter. On the other hand, an event such as the ‘Challenger’ disaster, although shocking, would probably have no direct consequences for most people’s lives, so it was not entirely surprising that flashbulb effects were not so clearly found. In an effort to explore this possibility, Conway et al. (1994) investigated flashbulb memories for the resignation of British Prime Minister Margaret Thatcher. They discovered that the Thatcher resignation produced significant flashbulb effects in a sample of British people (for whom the event might have important consequences), but people from other countries showed very little evidence of a flashbulb effect. Other studies have confirmed that flashbulb memory tends to be related to the level of personal significance or importance that the event holds for the perceiver, as for example in a study about hearing news of the death of the first Turkish president (Tekcan and Peynircioglu, 2002), or hearing news of a nearby earthquake (Er, 2003).

Most studies suggest that flashbulb memories are in fact subject to normal forgetting processes and errors over time, as with other kinds of autobiographical memory. Although the available research is not totally conclusive, at present there does not seem to be any clear justification for regarding flashbulb memory as being fundamentally different from other types of memory. However, this is not to deny the existence of flashbulb memory. While it may involve the same basic neural processes as other forms of memory, flashbulb memory is still distinguished from other autobiographical memories by its unusual degree of vividness and detail (Conway, 1995).


Neisser (1982) rejected the notion of flashbulb memory as a special process, arguing that the relatively high durability and detail of flashbulb memory could be simply a consequence of frequent retrieval. Neisser argued that a memory for a very significant event would probably be subjected to frequent review and retelling, which would help to strengthen the memory trace. A number of studies have confirmed a relationship between flashbulb memory and the extent of repetition and rehearsal (Tekcan and Peynircioglu, 2002; Hornstein et al., 2003).

It has also been suggested that flashbulb memory could be seen as a form of context-dependent learning, but one in which a very dramatic and memorable event provides a powerful contextual cue for more trivial aspects of the occasion (Groome, 1999). For example, the rather unmemorable slice of toast you happen to be eating one morning could become extremely memorable when consumed in the context of a news report announcing the outbreak of war or the death of your country’s leader. Unlike more typical examples of context-dependent memory, in this instance the major news event is serving as a context for other trivial memories. Davidson and Glisky (2002) have proposed a similar explanation, suggesting that flashbulb memory can possibly be regarded as a special case of source memory.

Some researchers have reported a relationship between flashbulb memory and the severity of emotional shock reported by the participants. Hornstein et al. (2003) found that flashbulb memories relating to the death of Princess Diana were greater for individuals who had been very upset by the news. However, Talarico and Rubin (2003) reported that participants’ ratings of their level of emotional shock following the World Trade Center attack predicted their confidence in the accuracy of their flashbulb memories but not the actual accuracy of their retrieval.

Demiray and Freund (2015) found that flashbulb memories were stronger when they related to an event of great personal significance (for example, hearing about the death of a relative), whereas flashbulb memories for public events (such as the death of Michael Jackson) tended to be less intense.


Candel et al. (2003) investigated the flashbulb memories of amnesic Korsakoff patients in the aftermath of the attack on the World Trade Center. Despite being severely amnesic for most daily events, most of the Korsakoff patients were able to remember some information about the attack, but they did not appear to have any flashbulb memories for the details of their own personal circumstances when they heard the news of the attack.


Figure 7.5 Michael Jackson.

Source: copyright Featureflash/

Davidson and Glisky (2002) found that flashbulb memories following the death of Princess Diana were weaker for older people than for younger people. It is not entirely clear why older people should be less prone to flashbulb memory, though this finding may possibly reflect the general decline found in the memory performance of older people. Davidson and Glisky found no relationship between flashbulb effects and measures of frontal or temporal lobe function, which might have been expected since these brain areas are known to be involved in context retrieval and memory storage, respectively. However, Bourne et al. (2013) reported that the activation of a flashbulb memory was associated with increased activation of the amygdala, which is a region of the brain known to be involved in intense emotional responses. Rimmele et al. (2012) showed that flashbulb memory involves a selective enhancement of certain aspects of a memory, an effect that would be expected in a highly emotional situation.

One intriguing piece of speculation is that the basic phenomenon of flashbulb memory may also be responsible for the occurrence of intrusive memories in certain clinical disorders (Conway, 1995; Sierra and Berrios, 2000; Budson and Gold, 2009). For example, one of the main symptoms of PTSD (post-traumatic stress disorder) is the occurrence of extremely distressing memories of some horrifying experience, which are unusually persistent and long-lasting, and also very intense, in fact so powerful that they cannot be kept out of consciousness. Sierra and Berrios (2000) suggest that the flashbulb mechanism may be involved in a range of intrusive memory effects, including the intrusive memories found in PTSD, phobia and depression, and even perhaps in drug-induced flashbacks. At present this view is largely speculative, but if evidence is found to support it, the mechanism of flashbulb memory would acquire a new level of importance.



One form of memory that has particular importance in real life is eyewitness testimony. In a court of law the testimony given by eyewitnesses is often the main factor that determines whether or not the defendant is convicted. Kebbell and Milne (1998) carried out a survey of British police officers and found that they considered eyewitness accounts to be generally quite reliable and accurate, but there is a considerable amount of evidence suggesting that eyewitness testimony is often unreliable and does not justify the faith placed in it by the courts.

The introduction of DNA testing has provided a new method of establishing the guilt or innocence of defendants, and it has all too frequently demonstrated that eyewitnesses have made mistakes. Wells et al. (1998) described forty cases where DNA evidence had exonerated a suspect who had been wrongly identified by eyewitnesses, and in five of these cases the wrongly convicted person had actually been on death row awaiting execution. More recently, Brewer and Wells (2011) reported that 258 Americans convicted of serious crimes had subsequently been exonerated by DNA evidence and freed. Of these cases, 200 had been convicted on the basis of eyewitness testimony, which had evidently been mistaken.


The experiments of Bartlett (1932) showed that recall is prone to distortion by prior knowledge and expectations, and he warned that this effect probably affected the testimony of eyewitnesses. Research focusing more specifically on eyewitness testimony has established that eyewitnesses are indeed susceptible to such reconstructive errors. For example, Tuckey and Brewer (2003a) found that eyewitnesses are better at recalling information that is consistent with their expectations and schemas than those that are not.

More recent research has shown that eyewitness recall of an event is not only subject to distortion from pre-event knowledge and expectations, but it can also be changed and distorted by something experienced after the event. The distortion of eyewitness testimony by post-event information is known as the ‘misinformation effect’, and it was first demonstrated by Loftus and Palmer (1974). They showed participants a film of a car accident, and later asked them a series of questions about what they had seen. Their answers were found to be strongly influenced by the wording of the questions. For example, participants were asked to estimate how fast the cars had been travelling at the time of the collision, but the wording of the question was varied for different subject groups. Those who were asked how fast the cars were travelling when they ‘smashed into one another’ gave a higher estimate of speed on average than did those who were asked how fast the cars were travelling when they ‘hit one another’. They were also far more likely to report having seen broken glass when questioned a week later, even though no broken glass had actually been shown. In another similar experiment, Loftus and Zanni (1975) found that after viewing a filmed car crash, participants were far more likely to report seeing a broken headlight if they were asked if they saw ‘the broken headlight’ rather than ‘a broken headlight’ (again no broken headlight had actually been shown in the film). The experiment demonstrated that merely changing a single word in the questioning could be sufficient to influence retrieval, essentially by making an implicit suggestion to the witnesses about what they should have seen.

Loftus et al. (1978) found that eyewitness memories became increasingly vulnerable to misinformation with increasing time intervals between the witnessed event and the contaminating input. A possible explanation for this finding is that as time passes the original memory trace becomes weaker and more fragmented, which makes it easier for the gaps to be filled by input from some other source. In fact there is clear evidence that eyewitness testimony (like other types of memory) becomes more unreliable with the passage of time. Flin et al. (1992) reported that eyewitness reports became less accurate after a 5-month delay, and although this applied to all age groups tested, they found that small children were particularly susceptible.

The ability of eyewitnesses to recall the appearance of an individual also seems to be subject to contamination effects. For example, Loftus and Greene (1980) showed that post-event information can significantly alter a witness’s recall of the physical characteristics of an actor in a staged event, such as their age or their height. This contamination would be likely to have a detrimental effect on the witness’s ability to provide the police with an accurate description of a suspect, or to identify that suspect at a later time. The accuracy of eyewitness identification of faces is covered in Chapter 3.

A number of studies have shown that it is possible to inhibit the retrieval of a particular piece of information by omitting it from a subsequent post-event presentation. Again, children seem to be particularly susceptible to this effect. Wright et al. (2001) presented children (aged 9–10) with a video depicting a series of events (such as a drink-driving incident), and then showed them the same video again later with a short scene missing. The children were then asked to imagine the event or (in a second experiment) to create a story about it, and subsequent testing revealed that the children often failed to recall the omitted scene.

Williams et al. (2002b) found a similar effect when reference to a particular scene was omitted from a post-event interview. Their participants (a group of young children aged 5–6) were far more likely to forget a scene if it was omitted from the post-event interview.

From the findings outlined above, it is easy to see how a witness to a real-life crime might suffer contamination from suggestions contained in questions posed long after the event by a police officer or a lawyer. Another possible source of post-event contamination is the testimony reported by other witnesses, which appears to have occurred in the case of the Oklahoma bombing, providing a real-life example of the occurrence of such misinformation effects.


On 19 April 1995, a huge bomb exploded beside the Alfred P. Murrah Building in Oklahoma City, killing 168 innocent people and injuring over 600 more. This was the worst act of terrorism ever to occur on American soil up to that time, and it caused profound shock throughout America. At first there were rumours that Middle Eastern terrorists were responsible, but two days after the explosion an American citizen called Timothy McVeigh, who was actually a Gulf War veteran, was arrested and accused of carrying out the bombing (Figure 7.6).

Timothy McVeigh had been stopped by chance for a routine traffic offence, but his appearance was found to match descriptions given by eyewitnesses and from video footage captured on security cameras. McVeigh, who had connections with a right-wing anti-government racist group known as the Aryan Republican Army, subsequently confessed to the bombing. After a lengthy court case Timothy McVeigh was found guilty of the bombing, and he was executed by lethal injection on 11 June 2001.


Figure 7.6 Timothy McVeigh, the Oklahoma bomber.

Source: Bob Daemmrich/Getty Images.

The Oklahoma bombing raises a number of important issues about the reliability of eyewitness testimony in real-life cases, most notably the apparent errors made by eyewitnesses in deciding whether McVeigh had an accomplice. The main eyewitnesses in this respect were three employees of the car rental shop where McVeigh had hired the truck used in the bombing. All three claimed to have seen McVeigh come in to hire the truck together with a second man. The FBI spent over a year searching for McVeigh’s accomplice, but he was never found. While it remains a possibility that McVeigh may have had an accomplice, Memon and Wright (1999) found evidence that the witnesses’ memories were probably contaminated by a subsequent event. On the day after McVeigh’s visit to the body shop, two other men had come in to hire a truck. These two men were quite unrelated to the bombing, but it is possible that the witnesses might have confused the memory of their visit with that of Timothy McVeigh. A further complication in this case was the apparent cross-contamination of the testimony given by different witnesses. When the three workers at the body shop were first interviewed by police officers, only one of them claimed to have seen a second man with McVeigh. The other two witnesses made no mention of a second man at this stage, but subsequently both of them came to believe that they had seen two men hiring the truck after they had discussed the event with the first witness. It appears that their recall of events had been influenced by the witness who described seeing a second man with McVeigh, since the three witnesses worked together and had discussed the incident extensively among themselves.

The occurrence of cross-witness contamination has since been demonstrated in carefully controlled laboratory studies (Wright et al., 2001; Wright et al., 2009). Furthermore, Edelson et al. (2011) showed that the testimony provided by a witness can be changed by simply informing them that other witnesses have reported the event differently. Wright et al. (2009) suggest three possible explanations for cross-witness contamination. First, witnesses may respond to social pressure, because they want to fit in with everyone else. Second, hearing a contradictory report from another witness may actually cause them to doubt the accuracy of their own recall. And finally, there is the possibility that the information provided by another witness may be mistakenly incorporated into the first witness’s memory; in other words, the misinformation effect.


In some circumstances it is actually possible to create entirely false memories in the mind of a witness by the use of suggestion effects. By using vivid forms of suggestion, such as the use of instructions to create a detailed visual image of some imaginary scene, it is possible to persuade some people to believe that they have a genuine personal recollection of an event that did not actually take place. For example, Loftus and Pickrell (1995) asked people to read several descriptions of things that had happened to them in early childhood, but they also included a fictitious event, which was a story about getting lost in a shopping mall. When questioned later, about a third of the participants described this fictitious event as having actually happened to them, even though this was not correct.


Figure 7.7 Lost in a shopping mall. Did this happen to you?

Source: copyright Dmitrijs Dmitrijevs/

Following the London bus bombing of 2005, Ost et al. (2008) asked a sample of British people whether they had seen TV footage of the bus exploding. About 40 per cent of the sample replied that they had, despite the fact that no footage of the explosion actually existed. Again it would appear that suggestive questioning and incomplete memory traces may have led to the creation of a false memory.


Another kind of false memory is the occurrence of a false confession. There have been many court cases in the past where the person accused has confessed to a crime that was later found to have been committed by someone else. Indeed Kassin et al. (2012) reported that in roughly 25 per cent of DNA exoneration cases, the conviction had been made on the basis of a false confession. Not surprisingly, a confession greatly increases the chances of a defendant being found guilty. Kassin et al. reported that a false confession not only tended to convince jury members of the defendant’s guilt, but it also biased their interpretation of other evidence in the case, so that even relatively weak evidence was more likely to be seen as incriminating. One famous example described by Kassin (2012) was the case of Amanda Knox, who confessed to the 2007 murder of Meredith Kercher, her room-mate in Perugia. However, an appeal court quashed her conviction in March 2015, accepting the claim that police pressure had caused Amanda Knox to make a false confession.

Kassin (2008) pointed out that some false confessions were made out of mere compliance, whereby the defendant confessed to a crime under pressure, despite believing themselves to be innocent. However, in many cases the individual making the false confession apparently comes to believe that they actually committed the crime, probably because the evidence is so strong that they assume that they must have forgotten or repressed the memory of their criminal act. In a lab study, Shaw and Porter (2015) have recently demonstrated that it is indeed possible to induce false memories of having committed a crime in an innocent person, by including strong suggestion and misinformation in an interview.


Figure 7.8 Amanda Knox.

Source: copyright © WENN UK/Alamy Stock Photo.


The contamination of eyewitness testimony by a subsequent input has now been clearly established by experimental studies, though the exact mechanism underlying this phenomenon remains uncertain. One possible explanation (Loftus, 1975) is that parts of the original memory are actually replaced by the new input and are thus permanently lost from the memory store. Some support for this hypothesis comes from the finding (Loftus, 1979) that when people recalled events wrongly, further guessing or the use of a recognition test containing the correct item did not usually help them to retrieve the lost memory. However, Dodson and Reisberg (1991) found that the use of an implicit memory test would sometimes facilitate the retrieval of the correct information, which the eyewitness could not retrieve explicitly. This suggests that, at least in some cases, the original memory has not been totally lost but has merely been rendered inaccessible to normal explicit retrieval processes.

Lindsay (2008) suggests that the misinformation effect may be caused by source misattribution, whereby the witness confuses the information from two different events. For example, in the case of the Oklahoma bombing described above, witnesses confused two different car hire customers with one another.

It has also been suggested that RIF (retrieval-induced forgetting – see previous chapter) could play a part in the misinformation effect. MacLeod and Saunders (2005) confirmed that RIF can affect the retrieval of crime descriptions, noting that the misinformation effect is stronger for items that are subjected to RIF inhibition. If a witness rehearses post-event information, and thus strengthens access to it, this could lead to the inhibition of memory traces for the original witnessed event. RIF could also be responsible for the finding that witnesses tend to forget scenes that are omitted from a subsequent re-showing of the incident (Wright et al., 2001). In this case the strengthening of rival memory traces for items included in the re-showing might have inhibited the memory traces for the omitted items.

Chan and Lapaglia (2013) suggest that the misinformation effect could reflect the fact that the retrieval and reactivation of a memory trace (e.g. during police questioning) renders it more vulnerable to change, a phenomenon known as reconsolidation.


There is a considerable amount of evidence indicating that small children are more prone to suggestion, contamination and memory distortion than adults. As noted earlier, Flin et al. (1992) found that the accuracy of children’s eyewitness reports deteriorated more rapidly over a 5-month period than did those of adults. Poole and Lindsay (2001) found that children were especially susceptible to post-event misinformation. In their study, children aged from 3 to 8 years took part in a science demonstration, after which they listened to parents reading a story that contained some events they had just seen and some that they had not. Subsequent testing revealed that many of the fictitious events were recalled as though they had been actually experienced. When asked to think carefully about the source of their memories (known as ‘source monitoring’), many of the older children withdrew their incorrect reports, but this did not occur with the younger children in the sample. Davis and Loftus (2005) conclude that the possibility of contamination from post-event information is a particular concern with very young child witnesses, who seem to have difficulty in monitoring the source of a memory trace.

Dodson and Krueger (2006) reported that elderly people tended to provide a less accurate account of witnessed events than younger adults, and the elderly were also more likely to produce false memories. This probably reflects a deterioration in memory performance in older adults. Indeed, Zhu et al. (2010) found that individuals with poor memory function and low IQ provided less accurate testimony and were more prone to the misinformation effect.


Witnessing a crime often takes place under stressful conditions, sometimes even involving a direct threat or actual violence towards the witness. Loftus and Burns (1982) showed that eyewitness testimony tends to be particularly prone to distortion when the events witnessed involve violence, since witnesses are likely to be less perceptive when in a frightened state.

More recent studies have confirmed that stress and anxiety significantly impair the accuracy of eyewitness memory (Valentine and Mesout, 2009) and the ability of a witness to identify the perpetrators of a crime (Morgan and Southwick, 2014).

Loftus (1979) also found that when a weapon was involved in a crime, witnesses tended to narrow down their attention to concentrate on the weapon, and became less aware of other, more peripheral aspects of the scene. Loftus called this phenomenon ‘weapon focus’, and it has been observed in many eyewitness studies both in the laboratory and in real-life crime cases (Fawcett et al., 2013).


A witness who expresses great confidence in their testimony (e.g. ‘there is no question in my mind that this is the man’) is more likely to convince a jury. However, although lab studies have generally suggested that confident witnesses are more likely to be right, in real-life cases it has been shown that very confident witnesses are often mistaken (Smalarz and Wells, 2015). Smalarz and Wells argue that witness confidence should not be given undue weighting. They also suggest that the discrepancy between the lab and real-life findings may reflect the fact that witnesses in real cases are more likely to receive feedback from other witnesses and from police officers, often encouraging them to believe that their recollection of events is correct. In addition, real cases often involve very long delays between crime and trial, during which witness memory deteriorates.

In fact, there are several fundamental differences between laboratory and real-life eyewitness studies. A witness to a real crime is likely to be under high stress at the time of the crime event, and they will also be aware that their testimony in court may help to convict someone of a serious crime (Eysenck and Keane, 2015). In lab studies the witnesses are under little stress, and they know that their testimony will have no important consequences. In view of these differences we should be cautious about drawing inferences from lab studies of witness testimony, as they will not necessarily be representative of witness performance in a real-life case.


Kassin et al. (2001) carried out a survey of sixty-four experts on eyewitness testimony, and found that there was a clear consensus view (using a criterion of 80 per cent of the experts being in agreement) that certain findings were now supported by sufficient evidence to be presented in court as reliable phenomena. These included the contamination of testimony by post-event information, the importance of the wording of questions put to witnesses, the influence of prior attitudes and expectations on testimony, and the suggestibility of child witnesses. All of these were regarded as established phenomena that could be legitimately stated in a court of law by an expert witness in support of their case.

A number of lessons can be learnt from the research summarised in this section, which have important implications for those who are involved in the process of obtaining eyewitness testimony. It is clear that the memory of a courtroom witness can easily be affected by contamination from subsequent information, which might be included in police questioning, newspaper articles or discussions with lawyers or other witnesses. Judges and juries therefore need to understand that witnesses cannot be expected to have infallible memories, and the court should not place too much reliance on the evidence of eyewitness testimony alone. In order to minimise the risk of post-event contamination, statements should be taken from witnesses as soon as possible after the incident in question, and witnesses should be allowed to use notes when giving their evidence in court at a later date rather than relying on memory. Police interviewers should be particularly careful about their methods of questioning, and should avoid the use of leading questions or suggestions that might implant misleading information in the witness’s head. Finally, there is a need for particular care when obtaining eyewitness testimony from young children, because of the difficulty they tend to have in distinguishing between real events and imagined or suggested events.

Those involved in the legal process require a clear understanding of these established phenomena in order to minimise the risk of a miscarriage of justice through the fallibility of courtroom testimony. Benton et al. (2006) reported that most judges have very little knowledge of the findings of eyewitness research, and jurors tend to know even less.

However, there has been progress in at least convincing courts of the value of eyewitness research. Loftus (2013) points out that in recent years the chances of a defendant receiving a fair trial have greatly improved, because of a better understanding of the fallibility of witnesses. The increased use of expert witnesses has also helped, as for example when psychologists are called on by the court to clarify the extent to which a witness can be expected to remember events accurately. Loftus concludes that ‘25 years of eyewitness science finally pays off’.

One important technique, which has been shown to improve the accuracy of police interviewing, is the ‘cognitive interview’, which is covered in the next chapter.


•  Memory for real-life autobiographical experiences tends to be far more accurate and durable than memory for items tested in a laboratory experiment.

•  Recent events and experiences are generally easier to remember than events and experiences from the distant past, but older people recall more events from their early adult years. This phenomenon is known as the ‘reminiscence bump’.

•  Early infancy is one period of life that appears to be particularly difficult to remember, and most people recall virtually nothing from the first 2 years of their lives. This phenomenon is known as ‘infantile amnesia’.

•  Most people retain very vivid and lasting memories of where they were and what they were doing at the time of hearing news of a shocking event. This phenomenon is known as ‘flashbulb memory’.

•  Although some researchers argue that flashbulb memory involves a special encoding process, this view has been challenged by the finding that flashbulb memories are subject to errors and forgetting, as with other forms of memory.

•  The recollections of eyewitnesses are not very reliable, and are responsible for the wrongful conviction of many innocent people.

•  Eyewitness testimony has been found to be vulnerable to contamination from post-event information (the ‘misinformation effect’), and it is also susceptible to conformity effects and cross-witness contamination.

•  While young children can provide accurate information under the right circumstances, they are particularly susceptible to suggestion and prone to reporting events that did not actually occur.

•  It is possible to implant a completely false memory in the mind of a witness, which they cannot distinguish from a true memory.


•  Williams, H.L., Conway, M.A. and Cohen, G. (2008). Autobiographical memory. In G. Cohen and M.A. Conway (eds), Memory in the real world. Hove: Psychology Press.

•  Wright, D.B. and Loftus, E.F. (2008). Eyewitness memory. In G. Cohen and M.A. Conway (eds), Memory in the real world. Hove: Psychology Press.