OF FALSE MEMORY
ALAN J.PARKIN
UNIVERSITYOFSUSSEX
ABSTRACT:
This article reviews recent research into the neuropsychology of
false memory as revealed by studies of individuals who have suffered lesions to
the frontal lobes. It is shown that frontal lobe damage gives rise to confabulation
along with a number of laboratory-based
demonstrations
of memory error
including source amnesia and impaired metamemory. It is further shown that
frontal lobe damage gives rise to a disorder known as pathological false recognition in which patients produce large numbers of false alarms during recognition
testing. It is argued that the most plausible account of these various findings is
that frontal lobe damage serves to prevent the creation of an appropriate and
properly focused retrieval context which may arise because either retrieval or
encoding factors have been compromised.
That month was a dim dark nightmare. Blankness was interspersed with bizzarre and uneasy happenings. I remember mad things, uncharacterisitic reactions to things, recollections seemingly lasting for hours instead of seconds. I
was unable to distinguish between fact and fiction, imagination and reality.
Consequently I made up utterly believable stories, and related these with such
conviction that no one knew whether they were coming or going. I was convinced that my parents had moved house and that people on the television
were my friends (Anonymous 1996, p. 1485).
The above quote comes from a British Army officer who suffered frontal lobe dam-
age during the Gulf War. It reminds us that brain damage not only causes the loss
of memory but that it can also lead to the fabrication of memory. Although identified clinically for many years this form of memory impairment has received scant
attention from experimental neuropsychologists
until very recently. The purpose
of this article is to provide the reader with a review of what recent work on adult
lo: Alan J.
Parkin,
Laboratory
of Experimental
Psychology,
342
LEARNINGANDINDIVIDUAL DIFFERENCES
patients with frontal lobe damage has revealed about the nature of false of recollection and the reasons for it. While the present article is specifically concerned
with data and theories derived from the adult neuropsychological
literature there
may well be important imphcations for understanding the nature of false memory
in children. It is well known that children exhibit some of the false memory phenomena associated with frontal lobe damage (Lindsay, Johnson, & Kwon 1991;
Wellman 1978) and that important changes in frontal lobe function take place
between the ages of 5 and 10 years (Case 1992) with fuI1 maturation of the frontal
cortex not occurring until the teenage years (Smith, Kates & Vreizen 1992). It is
therefore possible that the theoretical insights gained from the adult neuropsychological literature may have important implications for understanding
why the
memory of children is so vulnerable to distortion. More specifically it may be possible to develop a framework for understanding this area within the context of
developmental
neuropsychology
(see Schacter, Kagart, & Leichtman 1995 for a
discussion of this viewpoint).
343
NElJROPSYCHOLOGYOFfALSEMEMORY
pp. 277). A more recent and extensively investigated patient of this kind is JB (Parkin 1997; Parkin, Bindschaedler, H arsent, & Metzler 1996; Parkin, Leng, Stanhope,
& Smith 1988) who suffered a ruptured aneurysm of the anterior communicating
artery (ACoA). ACoA aneurysms are one of the most frequent causes of fantastic
confabulation (DeLuca & Diamond 1995; Fischer, Alexander, DEsposito, & Randall 1995) because the resulting infarction due to ACoA disruption and vasospasm
of the anterior cerebral artery. JB presented florid fantastic confabulation for about
two years despite performing quite well on standard clinical assessments of memory. A feature of his confabulation was that he tended superimpose aspects of his
more remote stable past on to his present situation thus producing bizzarre memories in which, for example, the hospital he was in became a college:
Al: Tell me a bit about (name of hospital).
JB: Well its a nuthouse for want of a better phrase - lunatic asylum I would
imagine.
Al? What makes you say that?
JB: Theres some funny people in there. I want out.
AI? You were telling me about (name of hospital). Why do you think youre
there?
JB: I dont really know. I have asked several people but never got an answer
that satisfies me. Dont know why Im there at all.
AI? What do you do during the day there?
JB: Same as everyone else. Bit of maths, bit of English, bit of religion first thing
in the morning.. .depends whats on the timetable-normal
school lessons.
AI: If its a nuthouse why are you having lessons?
JB: Brain stimulation I suppose.
Al? Brain stimulation?
JB: Got to get the brain working again.
JBs variable insight interacting with confabulation also cropped up a little later
in the same conversation.
JB: Ive been chugging along over the last three weeks....Its
ory.
344
DIFFERENCES
VOLUMEQ,N~~BE~4,1997
JB: Quite regularly. Sort of get in the car and think where am I going? I must
sit and work it all out. But its not as bad as all that-Ill
away.
JB: Well I give myself a clue and I can work it out from there. Well Ive got in
the car and looked at my timetable where it says on the first page which school
Im going to. I havent turned up at the wrong one yet.
AI: How many different schools do you go to?
IB: Well theres been three in as many months (Parkin 1997).
ORDER
345
Corsi, & Leonard 1991) who presented subjects with pairs of stimuli and at various intervals a test card showing a ? between the two stimuli. Subjects had to
decide which stimulus had been most recently presented. On some test trials
both stimuli had been previously presented but at different times (e.g., 4 versus
8 trials ago) thus requiring a recency discrimination. However, when one of the
stimuli was novel the task was just recognition. The principal result was that
the frontal patients were impaired on recency discrimination but not recognition-thus
they had difficulty remembering the temporal order of the stimuli.
Shimamura, Janowsky, and Squire (1990) presented frontal patients and controls with a 15 word list and then required the subjects to reconstruct the list
order. Frontal patients found reconstructing list order much more difficult even
though their recognition memory for the words was normal. The same study
also showed that fronta lesions affect the ability to correctly sequence facts
about the past even when general knowledge of those facts is intact-this
was a
phenomenon that I also observed during JBs confabulatory phase (Parkin et al.
1988). Thus he would show quite detailed and accurate recollection of specific
periods of his life (e.g., names and characteristics of colleagues) but be unable to
place these life periods in their correct temporal sequence unless they were logically constrained.
METAMEMORY
~~fj~e~~~cf~~~
of the frontal cortex and it is clear that disruption of this aspect of
mnemonic processing renders the memory system uncertain and allows the emergence of false memories. Greater understanding of the mechanism(s) underlying
the deficits caused by this malfunction wili be cruciai in our explanations of how
the veracity of memory is established and, more broadly, have possible implications for false recollection observed in non-clinical populations.
STlJDlES
While false recollection can often be observed within clinical populations it is difficult to use clinical observations as the basis for developing a theoretical account.
Source monitoring errors can be observed in controlled experimental conditions
but here false recollection is being observed indirectly and in many instances the
error may arise from guessing rather than a false source attribution. In order to
make more direct observations
of false recollection it has been necessary to
develop new paradigms. At the forefront of this has been use of thefalse recoglrifiolz
task. In a false recognition test subjects first see a series of target items (usually
words or pictures) followed by a recognition test in which target items are interspersed with novel distractor items. The subject is required to identify only previously exposed targets and a false recognition response arises when a distractor
item is incorrectly identified as a target-this
is known as a false alarm.
There are now four individual case studies indicating that lesions to the frontal
cortex can give rise to pathological false recognition (PFR)--this can be defined as the
production of abnormally high levels of false alarms. The first observations of this
phenomenon came from Delbecq-Derousne,
Beauvois, and Shallice (1990) who
reported that RW, a patient who suffered a ruptured ACoA aneurysm resulting in
frontal damage, made large numbers of false alarms on a forced choice recognition
test-importantly
RW was as sure about his false alarms as he was his correct
responses. On recall he performed reasonably although it was notable that he
tended to produce more false information in his recall (intrusions) than controls.
JB (see above) produces high rates of false alarms on a range of recognition
memory tests with a hit rate averaging about 80% (within normal range) and a
false alarm rate of 40% (extremely abnormal). JB also tends to be confident about
his incorrect responses and, like RW, intrusions feature significantly in his recall
(Parkin et al. 1996). Schacter, Curran, and Galluccio (1996) describe BG who also
had circumscribed damage to the frontal lobes On several tasks BG made excessive numbers of false alarms but, simiIar to JB, his hit rate was normal. Again BG
is as sure about his false alarms as his correct responses. Finally, Parkin, Bradley,
Hughes, and Squires (forthcoming) have described MR-a patient with a progressive demyehneating
illness affecting the frontal lobes. MR reported having
illusory memories. In one instance he decided to paint the kitchen door and, when
his wife returned, he announced that the job was done. However, nothing had
been done even though MR had a convincing (to him) recollection of doing the job.
NEUROPSYCHOLOGYOFFALSEMEMOt?Y
347
FALSERECOGNITION:ATHEORETlCALPERSPECTlVE
Although many theories of frontal lobe function have been developed the most
widely discussed view is that the frontal lobes operate as an executive goveming the more complex and flexible aspects of behavior. This approach has been
most extensively developed by D.A. Norman and Shallice (1986) who suggest that
the majority of mental activity involves routine automatic responses (e.g., driving)
which require no conscious intervention. However, when uncertainty prevails
and a decision made about various options is required, routine operations are
inadequate and control passes to the Supervisory Activating System (SAS). Shallice (1988) has suggested two ways in which the SAS could be involved in memory. First the SAS is involved in setting up descriptions which subsequently allow
access to specific events in memory. At retrieval descriptions may be ill-formed
but, via interaction with the stored contents of memory, they become increasingly
specific until the target memory is retrieved. The second function of the SAS is to
verify whether a retrieved memory is appropriate enabling, for example, an imagined event to be distinguished from a real memory. Thus, according to this theory
patients with frontally mediated impairments of the SAS could exhibit deficits
arising from the poor formulation and implementation
of descriptions or from
impairment to the verification process.
K. Norman and Schacter (1996) have taken exception to the Shallice (1988) view
by arguing against description/verification
dichotomy. They argue that information underpinning the description process involved in accessing memories must
be the same as that responsible for verification-the
argument being that the criteria for verification of an -event must logically be derived from representation of
that event thus making separation of description and verification impossible.
Instead they propose that frontal lobe damage can result in a poorlyfocused retrieval
system. The basis of this account is that, at retrieval, the frontal system is involved
in the creation of a retrieval context which will serve to verify the content of memories retrieved in relation to that context. Acuracy of recollection thus depends on
how well focused this context is with poor focusing allowing a too liberal response
bias and thus the generation of false recollection. Thus in the case of a typical false
recognition experiment it is assumed that the retrieval context is little more than I
saw a bunch of words thus allowing many distracters to be falsely recognized as
targets.
K. Norman and Schacters theory is attractive, not only because of its explanation of false recognition, but also because it also explains why frontal patients
typically have poor recall-the
argument being that a focused retrieval context is
just as important for deciding what to recall as it is for making recognition decisions. Similarly, errors insource monitoring and temporal order judgments could
348
DiFFERbVCES
also arise from poorly focused retrieval contexts. In addition, it is not difficult to
incorporate metamemory impairments into this framework on the grounds that
responses such as feeling of knowing may represent the partial implementation of
retrieval contexts which are insufficient
to promote recollection but whose
strength of activation provides a reliable cue to subsequent recognition performance. A patient capable of only poorly focused retrieval wouId, by definition,
have poor metamemory because the origin of this judgment is deficient. Finally,
confabulation may represent the most extreme form of poorly focused retrieval
resulting in the recollection of wildy inaccurate memories.
The focused retrieval theory has so far been developed as one which posits the
locus of frontally mediated false recognition impairments at the retrieval stagethe implication being that memories are normally encoded but damaged retrieval
mechanisms fail to derive an appropriately focused retrieval context. However,
since the theory proposes that retrieval context is a necessary interaction between
the content of the event memory and retrieval mechamisms, it follows that PFR
might be observed as a consequence of encoding rather than retrieval difficulties,
i.e. that PFR might arise because the retrieval system is operating on a highly deficient event memory which logically precludes a focused retrieval context.
NEUROPSYCHOLOGYOFFALSEMEMORY
349
DEESE PARADIGM
Deese (1959) described an experiment in which subjects studied word lists such
as bed, rest, awake, tired, dream, wake, snooze, blanket, doze, slumber, snore
and nap. Subsequent free recall showed that between 30% and 40% of subjects
falsely remembered that the word sleep had been presented in the list. Roediger and McDermott (1995) have replicated Deeses paradigm producing false
recall rates of up to 55%. It has also been shown, by means of the rememberknow paradigm (Gardiner, 1988), that subjects feel that they are actually recollecting these false recalls (e.g., Payne, Elie, Blackwell, & Pradere 1996; Read
1996).
Experiments using the Deese paradigm have led to two theoretical accounts
of how these false recalls arise. The implicit associative response (IAR) theory
basically argues that presentation of associates to the non-presented item leads
to an activation of that item which is sufficient to engender subsequent recall
(Underwood
1965). An alternative account is known as fuzzy trace theory
(Brainerd, Reyna, & Brandse 1996) which maintains that subjects develop two
separate representations
of an item at encoding, a verbatim memory trace and
a gist representation.
False recalls are explained by arguing that, at recall,
subjects access the general gist of their memory for the list (e.g., sleep related
350
words) and from this readily produce the strong associate sleep as a
response.
The Deese paradigm,
and the theoretical
framework
it offers might be
expected to inform the theoretical interpretation of false memory phenomena in
frontal patients. However, a number of obstacles prevail at present. First, and
most important perhaps, is that no frontal patient has yet been assessed on the
Deese paradigm. Until this is done there is no sure way of knowing that
accounts such as IAR or fuzzy trace have any bearing on the phenomenon of
PFR. Recent research does, however, suggest that these theories may not
explain pathological false alarm rates. At the heart of both IAR and fuzzy trace
theory is the idea that false recalls are a consequence
of semantic overlap
between associated words. Parkin (1997) describes data showing that JBs false
alarm rate in recognition
is unaffected by the degree of semantic overlap
between targets and distracters. Thus in one experiment, JBs false alarm rate
was no higher when distracters were highly related synonyms of targets compared with unrelated words. In addition, JB made just as many false alarms in a
nonword recognition task in which no basis for the semantic mediation of recall
is possible. Schacter et al. (1996) report similar findings for case BG and, collectively, these data indicate that the type of theoretical frameworks being derived
from the Deese paradigm may not have much relevance to the inte~retation
of
pathological false alarm rates.
NEUROPSYCHOLOGYOFFALSEMEMORY
351
tion of the auditory neocortex. This finding suggests that true memories involve
some perceptual record of the event along with any conceptual information that
is also present.
We have seen above that JBs false alarm rate is not influenced by the relationship between targets and distracters. This suggests that his memory traces are
essentially shallow, being based on the physical features of the stimuli. Evidence
supporting this is reported by Parkin et al. (1996) in which JB was given two versions of a recognition test for abstract shapes. In the non-overlapping condition the
target and distractor sets were made up from completely different pools of physical features while in the overlapping condition any given target-distractor pair
differed by only one physical feature. In the non-overlapping
condition JB performed normally and made no false alarms whereas on the overlapping condition
his performance was extremely poor and characterised by an extremely high false
alarm rate. The important point about this experiment is that it shows that only
low level physical differences between stimuli affect JBs false alarm rate. Thus the
idea that JBs false memory arises because he encodes only conceptual aspects of
stimuli is again not upheld.
It might, howewver, be argued that certain aspects of normal false memories are
physical in nature. Payne, Elie, Blackwell, and Neuschatz (1996) showed that false
recognition responses resulted in almost as many remember judgments as those
given to hits. Within Payne et al.s study a remember response was defined in
terms of various criteria such as how loud it was, how it was pronounced, and
what it reminded them of. Obviously some of these attributes are perceptual but
the lack of objective evidence that subjects actually experienced these attributes at
recollection means that we cannot be sure that real false memories induced in normal subjects can have perceptual qualities.
352
lEARN/NGANDlNDlVl~UAi DHfRfNCES
A. Married.
Q. How long have you been married?
A. About four months.
Q. Whats your wifes name?
A. Martha.
Q, How many children do you have?
A. Four. (He laughs.) Not bad for four months.
Q. How old are your children?
A. The eldest is 32, his name is Bob, and the youngest is 22, his name is Joe.
Q. How did you get these children in 4 months? (He laughs again.)
A. Theyre adopted
Q. Who adopted them?
A. Martha and I.
Q. Immediately
children?
A. Before we were married we adopted one of them, two of them. The eldest
girl Brenda and Bob, and Joe and Dina since we were married.
Q. Does it all sound a little strange to you, what you are saying?
A. (He laughs.) I think it is a little strange.
Q. I think when I looked at your record it said that youve been married for over
30 years. Does that sound more reasonable to you if I told you that?
A. No.
Q. Do you really believe that you have been married for 4 months?
NEUROPSYCHOLOGYOFFALSEMEMORY
353
A Yes
(from Moscovitch,
1989)
FRONTALLYMEOIATEDDEFICITSIN
RECALL
While the emphasis of this article has been on false recollection associated with
frontal lobe disease it is also the case that frontal lesions can also impair recall
in the absence of any false recollection. Hanley, Davies, Downes, and Mayes
(1994) describe the case of ROB. This woman, who suffered from a lesion in the
left caudate, showed normal levels of recognition memory and did not make
false alarms (even on tests similar to those generating high levels of false alarms
in case JB). There are many other examples of defective recall as a result of frontal lesions. Rochetta and Milner (1993), for example, showed that frontal
patients are far more disrupted than normal subjects by part-list cueing at
recall-this
is the phenomenon where presentation of part of the list as a cue for
recall inhibits recall of the rest. The greater impairment of frontal patients suggests that their retrieval strategies are impaired and thus they are less able to
reject the provided cues and seek alternative routes into memory. This lack of
flexibility in retrieval is perhaps most evident when frontal patients attempt
autobiographical
recall. Here there is plenty of evidence that recall is severely
impaired and that the patients tend to offer very general and vague responses.
The frontal patient CB (Parkin, Yeomans, & Bindschaedler
1994), for example,
was asked to retrieve specific episodic recollections to cue words but found this
close to impossibe. Thus for the cue flowers he could only respond I like
flowers.
CB, like many other frontal patients, shows enormous improvements in recall
when given prompts. Thus, with prompting, he was able to recall many events
accurately. The need to use so much prompting when eliciting recall from frontal patients means, in turn, that the patients memory is vulnerable
to
suggestibility via the prompting process. Thus, poor recall should be considered
354
VOLUMES.NUMBER4.1997
via suggestibility
ACKNOWLEDGMENTS:
Preparation of this article was supported by grants from the
Wellcome Trust and the UK Economic and Social Research Council.
NOTES
1. The term metamemory is also used to describe an individuals knowledge
how memory works (e.g., conditions that do or do not promote learning).
about
2. At this point it should be noted that the concept of poor focusing has considerable
affinity with Shimamuras (1995) idea that frontal lobe memory impairments arise from an
inability to inhibit irrelevant information.
REFERENCES
Anonymous. (1996). Desert Storm. Brifish Medical Bulletin.
Benson, D.F. & D.T. Stuss. (1990). Frontal lobe influences on delusions: A clinical perspective. Schizophrenia Bulletin, 16,403-411.
Brainerd, C.J., V.F. Reyna, & E. Brandse. (1996). Are childrens false memories more persistent than their true memories ? Psychological Science, 6, 359-364.
Case, R. (1992). The role of the frontal lobes in the regulation of cognitive development.
Brain and Cognifion, 20, 51-73.
Damasio, A. R. & H. Damasio. (1993). Cortical systems for retrieval of concrete knowledge. In Large scale neural theories of the brain, edited by C. Koch & J. L. Davis. Cambridge, MA: MIT Press.
Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal ofExperimental Psychology, 58, 17-22.
Delbecq-Derousne,
J. M. F. Beauvois, & T. Shallice. (1990). Preserved recall versus
impaired recognition. Brain, 213,1045-1074.
DeLuca, J. & B. J. Diamond. (1995). Aneurysm of the anterior communicating artery: A
review of neuroanatomical and neuropsychological
sequelae. Journal of Clinical and
Experimental Neuropsychology, 17,100-121.
Fischer, R. S., M. P. Alexander, M. DEsposito, & 0. Randall. (1995). Neuropsychological
and neuroanatomical correlates of confabulation. Journal of Clinical and Experimental
Neuropsychology, 17,20-28.
Fletcher, P. C., C. D. Frith, P. M. Grasby, T. Shallice, R. S. J. Fracowiak, & R. J. Dolan. (1995).
*Brain systems for encoding and retrieval of auditor-verbal
merno~.~ Brain, 118,
401416.
Fujii, D.E.M. & I. Ahmed. (1996). Psychosis secondary to traumatic brain injury. Nueropsychiafry, Neuropsychology, and Behavioral Neurology, 9, 133-138.
Gardiner, J.M. (1988). Functional aspects of recollective experience. Memory & Cog~jf~on,
16,309-313.
Hanley, J. Ii., A. D. M. Davies, J. J. Downes, & A. R. Mayes. (1994). Impaired recall of verbal material following rupture and repair of an anterior communicating artery aneurysm. Cognitive Neuropsycho~o~, 27,543--578.
Hanley, J. R. & A. M. D. Davies. (1997). Impaired recall with normal recognition. In Case
studies in the neuropsychology ofmemory, edited by A. J. Parkin. Hove, UK: Psychology
Press.
Harlow, J. M. (1993). Recovery from the passage of an iron bar through the head. History
of Psychiatry, 4,271-281.
Incisa DellaRocchetta, A. & B. Milner. (1993). Strategic search and retrieval inhibition: the
role of the frontal lobes. Neuropsychologia, 32(6), 503-524.
Janowsky, J. S., A. P. Shimamura, & L. R. Squire. (1989). Memory and metamemo~: Comparisons between patients with frontal lobe lesions and amnesic patients. Psychobiology, 27,3-11.
356
Johnson, M. K., M. A. Foley, A. G. Suengas, & C. L. Raye. (1988). Phenomenal characteristics of memories for perceived and imagined autobiographical events. Iorrrrrnl of
Experimerz trill
Psychology:
Gcwcml, 117, 371-376.
Johnson, M. K., S. Hashstroudi, & D. S. Lindsay. (1993). Source monitoring. Ps~~~~?~~~~~~~ca~
B~f~/~fj~~, 2 14, 3-28.
Kapur, S., F. 1. M. Craik, E. Tulving, A. A. Wilson, S. Home, & G. M. Brown. (1994). Neuroanatomical correlates of encoding in episodic memory: Levels of processing effect.
Proc. N&l. Acnd. 5-i. USA, 92, 2088-2011.
Payne, D.G., C.J. Elie, J.M. Blackwell, J.S. Neuschatz. (1996). Memory illusions: Recalling,
recognising and recollecting events that never occurred. fowmf
nfMcmorj/ IZJI~ L~JJgiiagc,
35, 261-285.
Roediger, H. L. & K. B. McDermott. (1995). Creating false memories: remembering words
not presented on lists. Jolrrnnl of Erperimntal
Psychology:
Lenrrkg,
Mmory
and Co@
boil, 21, 803-814.
NEUROPSYCHOLOGY
OF FALSE MEMORY
357