Wigmore , J. H. ( 1904 ; revised by J. H. Chadbourn , 1970 ). Evidence in trials at
common law , Vol. 3A . Boston : Little, Brown & Co .
Chapter Nine
The Cognitive Interview: Research and
Practice across the Lifespan
Robyn E. Holliday
University of Leicester
Charles J. Brainerd
Cornell University
Valerie F. Reyna
Cornell University
and
Joyce E. Humphries
University of Leicester
Introduction
There has been a dramatic increase in research on witness testimony in the last
25 years. Much of this research has been concerned with obtaining accurate
testimony from vulnerable witnesses and victims of crime (Ceci
& Bruck,
1995 ; Brainerd & Reyna, 2005 ). This chapter will focus on vulnerable wit-
nesses – children under the age of 16 years, adults aged 65 years and over,
and children and adults with intellectual impairments. In such cases, it is crucial
that these witnesses are interviewed as sensitively as possible so that the infor-
mation that is reported is accurate and reliable.
A number of factors determine the reliability of witness testimony. This
chapter is concerned with one of those factors, investigative interviews. The
manner in which a witness is interviewed is crucial for criminal investigations
Handbook of Psychology of Investigative Interviewing: Current Developments and Future Directions
Edited by Ray Bull, Tim Valentine and Tom Williamson
© 2009 John Wiley & Sons, Ltd.
138
Handbook of Psychology of Investigative Interviewing
and successful prosecutions. This chapter will focus on a set of forensic inter-
viewing techniques which have been tested extensively, namely the original
Cognitive Interview (CI) protocol (Geiselman, Fisher, Firstenberg, Hutton,
Sullivan, Avetissian
& Prosk, 1984) and its revision (Fisher
& Geiselman,
1992 ). First, we shall describe the original CI and the theoretical principles
on which the CI techniques are grounded. This will be followed by empirical
research presented chronologically, beginning with young children through to
old age. We then consider the viability of CI techniques with learning disabled
children and adults, and cognitively impaired older adults. We shall conclude
with a discussion of the state of play in CI research and practice in the twenty -
fi rst century – promising applications of CI research with suspects (Fisher &
Perez, 2007 ; Fisher & Castano, 2008 ), and progress on the development of
shortened Cognitive Interviews for use with children and older adults. CI
research in which these vulnerable individuals have been used is presented in
Table 9.2 .
The Cognitive Interview
Forensic interview protocols characteristically adhere to remarkably similar
structures. For example, all typically adopt a phased (funnel) approach begin-
ning with rapport - building, truth and lies testing, free recall requests followed
by a questioning phase in which open - ended questions are followed by specifi c
questions, and closure (Fisher & Schreiber, 2007 ). A number of interview
protocols are available, such as the guidance documents for the UK
(Memorandum of Good Practice, 1992 – MOGP; Achieving Best Evidence,
2001 and, 2007 – ABE, both produced by the Home Offi ce) and the Canadian
Stepwise protocol developed by Yuille, Hunter, Joffe & Zaparniuk (1993) .
One of the best
- known investigative interview protocols is the Cognitive
Interview (CI). The CI is based on empirical research and principles from
cognitive and social psychology (Fisher & Castano, 2008 ).
The original CI was developed to improve adult (of any age) eyewitness
testimony (Geiselman et al ., 1984 ). It has provided the impetus for much
research (see Geiselman & Fisher, 1997 ; Fisher, Brennan & McCauley, 2002 ,
for reviews). There are three fundamental psychological principles and sub -
principles on which the CI protocol is based. A core principle is memory/
general cognition, which can be broken down into fi ve sub - principles:
(a) Limited cognitive processing resources (e.g., Baddeley,
1986 ), that is,
there is a fi nite amount of cognitive resources available to process infor-
mation. If several cognitive tasks are attempted concurrently, it is likely
that the quality and quantity of a witness ’ s narrative of an event will be
affected negatively. Fisher et al. (2002) advise that interviewers record
the interview and refrain from interrupting an interviewee during his or
her recall narrative.
The Cognitive Interview
139
(b) Witness compatible questioning – witnesses possess unique mental repre-
sentations of their experiences; hence the interviewer must adapt ques-
tions accordingly. In Holliday
’ s research (Holliday,
2003a, 2003b
;
Holliday & Albon, 2004 ), questions were entirely based on the informa-
tion recalled by each child in his or her narrative recollection.
(c) Context reinstatement – mental reconstruction of a witness ’ s physical,
cognitive and emotional states of the to
- be - remembered event will
improve memory (i.e., encoding specifi city; Tulving & Thomson, 1973 ).
(d) Multisensory coding – in addition to conceptual representations, witnessed
events have sensory properties (Paivio, 1971 ) such as smell, sounds, visual
details (Fisher et al ., 2002 ). An example of the implementation of (c)
and (d) is taken from Holliday and colleagues ’ research (Holliday &
Albon, 2004 ). Children were given these instructions: Close your eyes.
Picture yourself back in the room where you watched the video. How were
you feeling? What can you see in the room? What can you hear in the room?
Who were you sitting next to (p. 269) ?
(e) Varied retrieval – memories can be accessed and retrieved by a number
of retrieval paths (Tulving, 1974 ; Anderson & Pichert, 1978 ). Varied
retrieval reduces the likelihood that a witness will use prior knowledge
and expectations to fi ll any gaps in their memory of the witnessed event.
Varied retrieval is facilitated by the instruction to
change perspective :
Recall the event from the perspective of another participant or location
in the same event: You said that Billy opened his presents. Now, I ’ d like
you to be Billy. What did you do fi rst? (p. 269), and change order : Recall
the event again in a different order (e.g., backwards), Now I want you to
tell me about the very last thing you remember in Billy ’ s birthday video.
What happened just before that? This prompt was repeated until a child
could recall no further details or had reached the beginning of the to -
be - remembered event.
A second core principle of the CI is social dynamics . Forensic interviews,
whether of co - operative witnesses or suspects, refl ect an imbalance of social
status (e.g., a police offi cer and a child witness). Social dynamics comprises:
(a) active witness participation – an interviewer
must emphasize to the inter-
viewee that it is the interviewee who is the expert about what he or she has
witnessed; hence the interviewee should actively control the interview (transfer
of control), and (b) development of rapport – interviewers should spend time
in the establishment of rapport (Fisher et al ., 2002 ).
The third core principle of the CI is communication . The interviewer needs
to extract from the witness specifi c details about their experience. Likewise,
the interviewee must convey to the interviewer the specifi c details he or she
witnessed. Communication is facilitated by
promoting extensive, detailed
responses (Fisher et al ., 2002 ). The report everything instruction – informing
the witness that he or she should report all information regardless of whether
he or she considers it relevant or not
– promotes detailed responses. For
140
Handbook of Psychology of Investigative Interviewing
example, I want you to tell me everything you can remember, every little detail
you can remember (Holliday & Albon, 2004 ). Second, an interviewer should
be mindful that some information that a witness wishes to relate might be
non - verbal. In such instances, code - compatible output should be implemented.
Fisher et al. (2002) give this example: ‘ if an event was experienced tactilely
(e.g., brushing against a fabric) then the witness might respond in a similar
tactile mode, by touching various fabrics ’ (p. 268).
Empirical e valuation of CI p rotocols
Numerous laboratory studies of the original and enhanced CI have been con-
ducted in the USA, UK, Australia and Germany since the fi rst CI research was
published in 1984 (Geiselman et al ., 1984 ). In an early test of the original CI,
college students witnessed a staged argument in class and were interviewed
two days later ( ibid .). Findings supported the prediction that students who
were given instructions in the application of the four CI mnemonics whilst
recollecting the witnessed event would recall more correct details than those
in a control condition. No signifi cant increases were observed in the amount
of incorrect or confabulated information reported.
More than 100 studies have evaluated CI protocols since its inception
(Fisher & Castano, 2008 ). There is no doubt that CIs improve witnesses ’
correct recollections of events using a number of different populations such as
those discussed in this chapter (children, elderly, learning - disabled), and in a
number of languages (other than English) including German (K
ö hnken,
Schimossek, Aschermann
& Hofer,
1995 ), Portuguese (Stein
& Memon,
2006 ) and Spanish (Hern á ndez - Fernaud & Alonso - Quecuty, 1997 ; Campos
& Alonso - Quecuty, 1999 ).
Some early studies reported increases in incorrect and confabulated details
along with the improved correct details. Indeed, almost 10 years ago, K ö hnken,
Milne, Memon & Bull (1999) published a meta - analysis of 55 research studies
in which recollections using CI protocols vs. control interviews were com-
pared. The researchers reported that CIs tended to produce small but statisti-
cally signifi cant increases in false information compared to control interviews.
Given that the number of published studies has almost doubled since publica-
tion of K
ö hnken
et al. ’ s paper a new meta
- analysis would be timely (for
reviews, see Geiselman & Fisher, 1997 ; Fisher et al ., 2002 ; Fisher & Schreiber,
2007 ).
Do the p ositive e ffects of a CI e xtend to
c hild w itnesses?
Researchers have reported mixed results as regards the effectiveness of the CI
with children. In general, however, more correct details are remembered with
The Cognitive Interview
141
CIs than with control interviews across early to late childhood (e.g. Geiselman
& Padilla, 1988 ; Saywitz, Geiselman & Bornstein, 1992 ; McCauley & Fisher,
1996 ; Granhag & Spjut, 2001 ; Milne & Bull, 2002; 2003 ; Akehurst, Milne
& K
ö hnken,
2003 ; Holliday,
2003a, 2003b
; Larsson, Granhag
& Spjut,
2003 ). In Holliday ’ s (2003a) study with 4 – 5 and 9 – 10 - year - old children, for
example, more correct person, action and object details were reported in
developmentally modifi ed CIs (MCI) (omitting the ‘ change perspective ’ ) than
in MOGP interviews (Home Offi ce, 1992). In two later studies, Holliday
(2003b) replicated these fi ndings with four
- and eight
- year - old children.
Holliday (2003b) reported evidence of developmental differences in the type
of details recollected with a MCI. The older children recalled more correct
person, action, object and location details than the younger ones. Similarly,
Milne & Bull (2003) reported that children aged eight and nine years recol-
lected more correct person and action details in CIs (omitting CP) than in
control interviews. On the other hand, Memon, Cronin, Eaves & Bull (1996)
found no evidence that a full CI improved six and seven year olds ’ recollections
of an eye examination. Likewise, Memon et al. reported no differences in
children ’ s correct recall when each of the CI mnemonics was compared with
a ‘ try harder ’ instruction.
As noted in early evaluations of CI protocols with adult participants, some
developmental researchers have reported increased incorrect and/or confabu-
lated recollections, as well as increased correct recall, when evaluating CI
techniques (McCauley & Fisher, 1995 ; Memon, Cronin et al ., 1996 ; Hayes
& Delamothe, 1997 ; Memon, Wark, Bull & K ö hnken, 1997 ), although other
researchers have reported no such effects (Milne
& Bull,
2002 ; Akehurst
et al ., 2003 ; Holliday, 2003a; 2003b ; Holliday & Albon, 2004 ). In the next
section, we discuss some exciting new developments in misinformation research
and the CI.
Can CI p rotocols r educe c hildren ’ s s uggestibility?
Holliday and her colleagues raised this important question in the fi rst of four
studies with children ( 2003a, 2003b ; Holliday & Albon, 2004 ). Given that
children under six years of age are disproportionately affected by misinforma-
tion, it is important to determine whether interview protocols such as the CI
can reduce the negative effects of misinformation on child witnesses. Ceci &
Bruck (1993) argued that children ’ s recollections of witnessed events can be
affected by a number of cognitive (e.g., memory, attention) and social (e.g.,
compliance, demand) factors. Many laboratory studies have reported that
children ’ s memories are negatively impacted by misinformation, with very
young children (aged three and four years) disproportionately affected (see
Bruck & Ceci, 1999 , for a review). Much of this evidence has been collected
using
‘ standard ’ and
‘ modifi ed ’ forced
- choice (e.g., Ceci, Ross
& Toglia,
142
Hand
book of Psychology of Investigative Interviewing
1987 ; Holliday, Douglas & Hayes, 1999 , Holliday & Hayes, 2001 ; Zaragoza,
1991 ) or yes/no recognition memory tests (e.g., Holliday & Hayes, 2000 ;
for reviews see Holliday, Reyna & Hayes, 2002 ; Reyna, Holliday & Marche,
2002 ).
A small number of laboratory studies have examined whether CI instruc-
tions minimize the impact of misleading questions on school - aged child wit-
nesses. For example, in Memon et al . ’ s (1996) study, eight - and nine - year - old
children watched a short fi lm. They returned 12 days later and were asked
misleading and neutral questions either before or following an interview
in which the
context reinstatement and
report everything instructions were
used as memory prompts. Memon et al. reported, as they had hypothesized,
no effect of interview type on responses to the pre
- interview questions.
For those children who were questioned post
- interview, however, those
given (prior) CI instructions gave more correct responses to misleading ques-
tions than those given a control interview (see also Milne & Bull, 2003 ).
Hayes & Delamothe (1997) reported that the c ontext reinstatement and report
everything instructions had no effect on suggestibility in six and ten year olds
when the misleading suggestions were presented before the CI instructions.
Holliday (2003a) found that whilst a CI increased fi ve and ten year olds ’
correct recall in comparison with a control interview, no evidence was found
that the suggestibility effects obtained on forced - choice memory tests given
after a CI were infl uenced by interview type. Clearly, one of the keys to these
disparate results could be the timings of the misinformation and of the
interview.
This point was taken up by Holliday (2003b) in two studies with fi ve - and
eight - year - old children. The two studies were identical, with the exception
that the timings of the misinformation presentation and interview. In the fi rst
study, children were misled after interview; in the second study they were
misled before interview. The results were straightforward: reporting of misin-
formation during interviews and on subsequent memory tests was reduced if
the children were interviewed with a developmentally modifi ed CI (omitted
CP) before they were given a memory test. Holliday & Albon (2004) repli-
Handbook of Psychology of Investigative Interviewing: Current Developments and Future Directions Page 27