Semantic type – word based false memories, are a type of false memory that are normally elicited by the commonly known DRM paradigm.
The classical procedure of the DRM paradigm involves participants studying a selection of lists, in which each list is composed of associates of one non-studied critical word (critical lure). After presentations of each list the participants are asked to recall each list, and then later during a recognition test, recognise the words that they believed, were present in the lists presented to them previously. It was observed that participants falsely recalled and recognised the critical non-studied lure at a high rate (i.e. false memories were produced).
The DRM paradigm might be dependent on the ability of the individual to group associated words into categories. It is therefore possible that an individual’s ability to categorise may be the reason why false memories are produced. As a consequence of this, in the present study a new hypothesis is proposed, which I call “categorisation activation hypothesis”. This new hypothesis may be used to explain why false memories occur in the DRM paradigm
The “categorisation activation hypothesis”, proposes that as a result of an individual ability to group associated words from the lists together into categories during the encoding phase of memory; when later asked to recall or recognise the words from the lists presented to them earlier, it will lead to the activation and retrieval of non-studied words (such as the critical non-studied lures) that are related to the words present on the list but were not studied themselves. Thus the better the individual is at categorizing the more false memories they produce.
This hypothesis was tested by the use of an adapted version on the DRM paradigm, to provoke the participants into producing false memories. In addition to the DRM paradigm, participants were given a categorisation test which tested the individual’s ability to categorise. The categorisation test consists of 11+verbal reasoning questions taken from 11+ secondary school entry exam papers.
It was found that there was a positive correlation seen between the individual’s ability to categorise and the number of critical non-studied words that were falsely recalled (spearman rank correlation coefficient; RS= 0.344 n=39 p <0.05). On the other hand, no significant correlation is seen between an individual's ability to categorise and the recognition of critical non-studied lures in the recognition test (RS=0.278 n=39 p<0.09).
It can be concluded from these results that the “categorisation activation hypothesis” can be applied to the recall of words during the recall test, however not to recognition of words in the recognition test. Although it is not understood why a variation is seen between recall and recognition of words, however the finding observed in this study, does provide a useful benchmark on which future studies can be developed.
False memory is the name given to a phenomenon that involves an individual remembering a past memory inaccurately; or in severe cases fabricating an entire memory that has never occurred . The label of false memory in cognitive psychology is applied to an assortment of different types of memory errors, stretching from misremembering word lists, to inaccurate recall of past events .
In 1990’s a controversy arose, which grabbed the attention of many scientists and the public, leading to the study of false memory to become widely popular. This controversy regarded; trauma-memory-oriented psychotherapies. The suggestive nature of memory recovering techniques used in these psychotherapies showed a disturbing increase in the number of patients, who falsely claimed to have remembered childhood sexual abuse (CSA), but only to later discover that this abuse in fact had never occurred . Many scientists have devoted their time to exploring these findings, and trying to understand why and how false memories occur (e.g. Loftus and Ketcham, 1994). Such work by cognitive psychologists has also led questions arising, about the reliability and distortion of many eye witness testimonies .
The popularity of this phenomenon is new, however false memory has been studied since Bartlett’s (1932) book “Remembering”. Bartlett (1932) is normally attributed for being the first scientist to have demonstrated experimentally the phenomenon of false memory in the form of various serial reproduction experiments (Roediger and McDermott, 1995)(Roediger and McDermott, 1995)(Roediger and McDermott, 1995)(Roediger and McDermott, 1995)(Roediger and McDermott, 1995)(Roediger and McDermott, 1995)(Roediger and McDermott, 1995)(Roediger and McDermott, 1995b)(Roediger and McDermott, 1995). One of the serial reproduction experiments conducted by Bartlett (1932) involved participants being given an adaptive translation of a North American folk tale “The War of the Ghosts”. Bartlett (1932) chose this particular folk tale with special consideration for many reasons, the main one being that he believed this passage will obtain a high amount of transformation as it was from a culture and social environment outside of the subjects; participating in the experiment. The participants were given time to read through the story twice, and then were asked to recall the story repeatedly over a period of time. Bartlett (1932) revealed that participants recalling the story did not elicit a truthful reproduction of the original story. Their recollection of the story became distorted by the influence of their existing knowledge and schemata (an individual’s perception of how an event should occur); leading to the language of the recall to become more conversational, and the participant’s omitting out details that they were not acquainted with and replacing them with words that were more familiar. An example of this will be the replacement of the word “canoe” to the more accustomed “boat” or “hunting seals” was replaced to simply “fishing” . It was through the conclusions obtained from his result’s that Bartlett (1932) introduced the concept of reconstructive memory. Reconstructive memory a contrast to reproductive memory (the accurate recall of a stimulus from memory) is an active process of filling in the gaps to rationalise the memory; so that it makes more sense, and is accustomed to the belief and expectations of the individual remembering the memory . This process of, reconstructing a memory predictably results in frequent memory errors occurring i.e. false memories . Bartlett (1932) thought that remembering events that are more decorative and elaborate such as novels, and past memories will result in reconstructive memory. Whereas remembering simple events such as word lists are more likely to elicit reproductive memory. Initially the book was met with conflicting views, however since then Bartlett’s (1932) renowned book ‘remembering’ has become widely cited and is now seen as an major contributor to the development of cognitive psychology
Over the year’s cognitive psychologists have used Bartlett’s (1939) work, as the basis, on which they have developed various different paradigms; that have been used to study reconstructive memory and the formation of false memories. Examples of such paradigms include the Loftus, Miller and Burns study, were they used slide shows to explore if suggestions can results in generation of false memories. Loftus and her colleagues (1978) tested their hypothesis, by presenting participants with slides that contained information about an automobile accident; after the presentation of the slides the participants were shown a series of misleading evidence. Later when recognition tests were presented, it was observed that the misleading information was incorporated in to the participant’s memory of the automobile accident .
Another example includes the use of sentences . Bransford and Frank (1971) presented their participants with short sentences. After a period of time the participants were asked to recognise the sentences that were presented to them initially from a selection of sentences. A high proportion of the participants recognised the longer sentences that was related to the shorter sentences originally learned, however contained more details. This finding suggested that participants during remembering integrated the short sentences they learned into schemas .
There have also been studies, which do not simply follow the normal tradition of trying to distort true memories (like the above examples); but explore if suggestive interviews can result in the implantation of a whole false memory. For example in Loftus and Pickrell’s study, participants were read 4 past memory events that supposedly occurred to them in childhood. 3 of the memories read were true, which were provided by a family member; however the 4th was a false memory, which told the story of the participant at the age of 5 getting lost at a shopping mall, and later being discovered by an elderly woman, whom reunited the participant with his/her family members. Later during two interviews sessions, where participants were asked to recall the childhood memories that were read to them previously, the results showed that 25% of the participants “remembered” the false memory events either partially or fully . The above examples of paradigms have all successfully provided evidence of generation of false memories. However, there does not seem to be a consistency in the magnitude of false memories produced; which differ from paradigm to paradigm. This could lead to the belief that these false memories can only occur under the paradigm conditions and perhaps do not relate to real life incidences outside of the laboratory .
Bartlett’s (1932) speculation: that word list paradigms largely elicit reproductive memory, had been generally accepted by cognitive psychologists. However, this speculation was bought to question by the findings published by Underwood . Underwood’s (1965) study is well known publication which demonstrates that false memories can be produced in a word list paradigm. Underwood’s (1965) paradigm involved participants firstly studying a list of words; after which they were given a recognition test consisting of 200 words that were related to the words in the lists. In the recognition test the participants were presented with each word audibly via a tape recorder, they were then asked to judge if each presented word was present earlier in the lists that were studied. Underwood (1965) found that the words that were related to words that were studied were falsely recognised as being a studied word (i.e. was present in the lists that were studied).
The paradigms discussed so far have all used recognition tests to study false memories, however an exception to this, is a study published by the experimental psychologist Deese . Deese (1959) constructed a simple procedure in which he used single trail, free recall paradigm to study, sematic type word based false memories. Deese (1959) carried out this procedure as he was interested in observing the probability of intrusions occurring during the single trail; free recall of each list. Deese’s (1959) paradigm involved him developing 36 lists, each list consisting of 12 words. These 12 words per list are primary associated of one non presented (critical) word (e.g. for the critical non-presented word “needle” the words were thread, pin, eye, sewing, sharp ECT.). These words were attained from Minnesota norms that produced the highest incidence of recall in responses to the stimulus (critical non- presented word) of the Kent-Rosanoff word-association lists . In Deese’s (1959) study, there were 50 participants (all of whom were undergraduate students), and each participant was tested individually. Participants were read the lists; after the conclusion of each list, participants were asked to orally recall the words from the list that was read to them (oral recalls were recorded and transcribed later). Deese (1959) after analysing the results, found that there were certain lists in which the Kent- Rosanoff stimulus word (critical non-presented word) intruded at higher rate than other lists. For example the list corresponding to the critical non-presented word “butterfly” produced 0% intrusions, whereas the list corresponding to the critical non-presented word “sleep” produced intrusions 44% of the time. Deese (1959) was interested in why a variation in number of intrusions (of the critical non-presented word) occurred between the lists, and therefore tried to predict the occurrence of the intrusion in each list via a word association test. The word association test consisted of participants being presented with lists of words (these words were the 12 words in each of the 36 lists used in the single trail: free recall test), and then instructed to write down one associated word that each word in the list reminds them of. Deese (1959) calculated the average probability of the 12 words present in each list, leading to the intrusion of the critical non-presented word. He then worked out that the probability of the critical non-presented word intruding in recall, and the mean percentage of the critical word being associated to the words present in the list was 0.873. This led Deese (1959) to conclude, that the strength of the associated words present in the list to the critical non-presented word, is directly proportional to probability of the critical non-presented word occurring as an intrusion during recall of the list.
The observations and conclusion drawn by Deese (1959) in this article, provides a great insight and bears a significant amount of importance to the studies into false memories . However, surprisingly this study went largely unacknowledged when it was first published; with only obtaining 35 citations in total between 1959-1994 .
It wasn’t until 1995 that Deese’s (1959) article became extremely popular. The peak in popularity was due to Deese’s (1959) article being brought to the attention of Roediger and McDermott (1995); who recognized that Deese’s (1959) findings were important, and established that his paradigm may possibly be a promising method in which false memories could be studied. Roediger and McDermott (1995) adapted Deese’s (1959) procedure of creating false memories for study in laboratories , which led to more cognitive psychologists to become aware of Deese’s (1959) findings. In consequence of this the citation rate of Deese’s (1959) article, doubled from the total amount of citation reached by the article at the end of year 1994 .
Roediger and McDermott (1995) conducted two experiments. The first experiment was largely carried out to replicate Deese’s (1959) findings on false recall. Roediger and McDermott (1995) developed, the six lists that had produced the highest rate of false recalls in Deese’s (1959) experiment: chair, mountain, needle, rough, sleep, and sweet. Each of the 6 list, like in Deese’s (1959) paradigm consisted of 12 words that were the strongest associates of the critical non-presented word. These words were mainly obtained from the lists presented in Russell and Jenkins (1954) word association norms. However, a few of the words in the lists were changed by Roediger and McDermott (1995) to words that they had believed were more prone to elicit the critical word. Participants were read each of the six lists; immediately after the presentation of each list, participants were instructed to recall as many of the words they could remember form the list without guessing. Roediger and McDermott (1995) extended Deese’s paradigm by an addition of a recognition test. This test was given to the participants following the recall test, and consisted of 42 items; 12 of which were studied words (words that were present in the 6 lists presented to the participants during the recall teat) and 30 of which were non studied words (words that were not present in any of the lists presented to the participants during the recall test). 6 of the non-studied words were the critical non-presented word of each of the 6 list generated. The participants were told to rate each word on the recognition test on a 4 point scale (4= sure old (studied) 3= probably old 2= probably new 1= sure new) as to their confidence, that the word appeared on the lists presented previously in the recall test.
Roediger and McDermott (1995) via their recall test successfully confirmed Deese’s (1959) findings. They found that the critical non-presented word of the lists was intruded in recall at a probability of 0.40, which was approximately the same probability of recalling a word that was present in the middle of the list (these findings were also echoed by other studies see McDermott, 1996 and Stadler et al., 1999). In the recognition test, the probability of a participant considering a critical non-presented word as being “old” (0.84) was approximately at the same probability of recognition of a studied word (0.86). Additionally over half of the participants (0.58) reported that they were sure, the critical non-presented word was presented in the 6 lists presented to them in recall test (i.e. believed that it was a studied word) . Encouraged by the results attained, Roediger and McDermott (1995) carried out a second experiment.
This second experiment consisted of an extended version of the 6 lists used in experiment 1; and 18 new lists (24 lists with each list consisted of 15 words). Participants were firstly given a recall test which consisted of participants being instructed to: immediately recall at the conclusion of each of the 8 lists, and after the presentation of the another 8 were asked to do a maths problem (the remaining 8 were not studied). Following the recall test, the participants were given a recognition test, which employed Tulving’s remember /know method. This test consisted of words from all 24 lists, including the 8 lists that were not studied and the 24 critical non-presented words. Participants were asked to judge each word, either as being ” old” (studied) or ” new” (non-studied) , and for every word judged as old they were told to attribute if they remembered the word being presented ( R) or knew the word was presented but could not remember the memory of it being presented (K).
In the recall test in Roediger and McDermott’s (1995) experiment 2, the critical non-presented word was recalled at a higher rate (0.55) than seen in experiment 1. Furthermore in the recognition test participants claimed to remember specific aspects of presentation of the critical non-presented word, suggesting that the participants were unable to distinguish between the presentation of a studied word and that of a critical non-presented word. The paradigm developed by Deese (1959), and Roediger, and McDermott (1995), is now commonly referred to as the DRM paradigm (named by Endel Tulving , see Sadler et al 1999 , Roediger, McDermott and Robinson, 1998) . The findings produced by the Deese, Roediger and McDermott via their work have led the DRM paradigm to become widely popular. The DRM paradigm has been used by many congestive physiologists, as the basis of their study to explore various different concepts. For example; hemispheric asymmetry , childhood sexual abuse of women , emotions and even the memory distortion observed in individual’s that have claimed of being abducted by aliens .
Stadler et al. (1999) also adapted the DRM paradigm to produced normative data on the lists normally used in this paradigm to produce false memories. Stadler et al (1999) used 36 lists, where 24 of the lists were form Roediger and McDermott’s (1995) materials for their Experiment 2, and the remaining 12, are the additional lists developed by McDermott . Each list consists of the 15 words that are considered to be the strongest associates of the critical non-presented word. The procedure followed that of the typical DRM paradigm. The participants were read the lists, and at the conclusion of each list, were asked to recall as many words as they could remember without guessing. After the completion of the recall test the participants were given a recognition test. This recognition test is different to that given to participants in Roediger and McDermott’s (1995) study. Instead of asking participants to rate each word on a 4 point confidence scale, participants were simply instructed that they were going to be presented with a list of words; they were then asked to read each word carefully and circle the words that they believe were present on the lists, presented to them during the recall test.
In the recall test Stadler et al. (1999) observed that there was a variation in false recall of the critical non-presented word between the lists. The list corresponding to the critical non-presented word “window” produced false intrusions 65% of the time, whereas the list corresponding to the critical non-presented word “King” only produced the false intrusion 10% of the time. Additionally, it was also seen like in Roediger and McDermott’s (1995) study, that the rate of false recall was approximately the same as the rate of recall of a word presented in the middle of the list. In the recognition test Stadler et al. (1999) observed that there was also a variation in false recognitions between lists. The list corresponding to the critical non-presented word “window” produced high levels false recognition of the critical non-presented word (85%); whereas the list corresponding to the critical non presented word “King” only elicited low levels of false recognition of the critical non presented word (27%).
As mentioned earlier it was generally thought that list learning paradigms generally elicits reproductive memory. Therefore the high level of false memories produced in the DRM paradigm is highly unusual, and bring about the suggestion that perhaps all memory is constructive. There have been many different theories proposed to explain why such intrusion occurs in DRM paradigm. Underwood (1965) proposed an “implicit associative theory”; in accordance to this theory during the encoding phase of memory, when participants are presented with a word from a list, it leads to the participants implicitly activating representation of associates of that word e.g. when subjects are presented with the word “soft” it may lead to participants to think of the word “hard” during encoding of the memory. Thus later in the recognition test when ” hard” is presented as a lure, subjects will claim to have remembered it to occur in the lists due to implicit association of ” hard” to “soft” during encoding of the memory. Brainerd and Reyna suggested the “fuzzy trace theory”. The “fuzzy trace theory” suggests that during the encoding of a memory, two parallel memory pathways are activated: verbatim traces and gist traces. In accordance to this theory verbatim traces represent accurate detail of a stimulus whereas a gist traces represent the general meaning of the stimulus. Thus the accurate recall and recognition of a studied word in the DRM is a result of the activation of the verbatim trace, whereas the false recall of the critical non-presented word is primarily a result of activation of the gist trace . The final theory: “activation-monitoring theory” was suggested by Roediger and his colleagues . This theory purposes that when participants are processing the words presented in the list during the DRM paradigm, it leads to activation of the critical non presented lure, either by spreading activation through a sematic network, or consciously by implicit association of the critical non-presented lure and the word studied in the list. It is then thought that false memories occur due to participants not being able to monitor the source of the activation of the memory, or mistakenly assuming that the activation of the memory occurred as result of the critical non-presented word being present in the lists .
In this present study, a new hypothesis is being proposed; “the categorisation activation hypothesis”. The study of sematic type false memories via the DRM paradigm, as it may have become evident now, maybe dependent on the participants ability to group the associated words of each list together to form categorises. As discussed above, the DRM paradigm elicits high levels of false recall and false recognitions memories. It is therefore inevitable to predict, that the individual’s ability to categorise may result in the individual being more prone to generate false memories. This led to the proposal of “the categorisation activation hypothesis”. This hypothesis proposes that as a result of individual’s ability to group associative words from the lists presented, into categories during the encoding phase of memory; when asked to retrieve the memory via recall or recognition, it will lead to the activation and retrieval of a non- presented word (e.g. critical non presented word) that is associated with the words present in the list. However, the word itself was not present in the list. Consequently if the “categorisation activation hypothesis” is true, than the stronger the individual ability to categorise; the more likely that the individual will elicit false memories.
This study explored this hypothesis, firstly by testing subject’s proneness to word-based (sematic type) false memories which was elicited via a readapted version of DRM paradigm used in Stadler et al (1999). Secondly participants were presented with a categorisation test, which tested the individual ability to categorise by the use of 11+ verbal reasoning questions (taken from 11+ secondary school entry exam papers). The answers produced from this DRM paradigm and categorisation test were then correlated. The paradigm is explained in more detail below.