Introduction
Information is the lifeblood of many investigations, whether about crime, accidents, public health, national security, or a host of others. Some of that information may result from analysis of physical records (e.g., instruments, a patient’s medical chart, skid marks, explosive devices, the plane’s black box), and some will come from interviewing people who have first-hand knowledge of the event (e.g., pilots, soldiers, physicians, victims, bystander witnesses, informants). The present chapter focuses on eliciting information from human sources.
On the surface, it might seem that if interviewers are knowledgeable about the event to be investigated, i.e., they know what kind of information to elicit (e.g., the robber’s appearance, the train’s speed, which terrorists attended a meeting) it should be relatively easy to extract that information – assuming, of course, that the interviewee is motivated to provide the requested information (see Brandon & Wells, this volume, Leins & Zimmerman, this volume for interview techniques with uncooperative respondents). Thus, for instance, if a police investigator needs to find out the criminal’s ethnicity, the police officer/interviewer can simply ask: Was he White, Black, Hispanic, or Asian? And, in fact, that is how many investigative interviewers proceed. That is, investigators think in terms of content (the information to be gathered) and then ask specific questions targeting each to-be-gathered fact, e.g., how fast was the car going? (Fisher, Geiselman, Raymond, Jurkevich, & Warhaftig, 1987). Although this method of interviewing (asking specific questions) is common among police – and others – it is not an effective method to gather information from human sources. Geiselman and Fisher therefore set out to devise an alternative method of interviewing, one that revolved around witnesses’ mental processes (rather than around the facts to be gathered) to elicit more information from human sources. The resulting procedure, called the Cognitive Interview (CI), was born out of this need, to devise an interview process based on the witness’s and the interviewer’s psychological processes.
The original version of the CI was developed specifically for criminal investigation. That the CI took this direction was mainly happenstance. It emanated from a brief conversation we had in Ed Geiselman’s office about what profession could best profit from applying psychological researchers’ knowledge about memory-enhancing techniques. Ed’s immediate thought was that police could benefit most from enhancing witness memory, because solving crimes depends mainly on witness evidence – despite the more exotic, but uncommon, solutions often found on television shows. And so, we devised the CI to be used by police to enhance witness recall of crime-related events.
When we first examined the problem, of witnesses failing to remember key elements of a crime, we thought about it mainly as a witness memory problem. As such, the original version of the CI (Geiselman et al., 1984) was composed exclusively of techniques that we believed would enhance witnesses retrieving information about past experiences (see Table 1.1: Active respondent participation; Report everything; Varied Retrieval). We later realized that witnesses’ under-reporting criminal events was not only a reflection of their using inefficient memory-retrieval mechanisms, but also that the interviewers were not processing information efficiently (e.g., listening to and notating the witness’s statements, keeping track of questions to be asked, developing a hypothesis of how the crime occurred, etc.). Moreover, other non-memory factors also contributed to witnesses’ under-reporting events, including the social dynamics between the interviewer and the witness (e.g., inadequate rapport), and the difficulty of communicating some ideas (e.g., describing the odor of a fire). Thus, whereas the initial version of the CI was a simple collection of memory-enhancing mnemonics, the CI evolved into a more wide-ranging approach (the so-called “Enhanced CI”) that addresses (a) the social dynamics between the interviewer and the witness, (b) the witness’s memory, and other cognitive processes, and also the interviewer’s cognitive processes, and (c) the witness and the interviewer communicating effectively with one another. The core elements of the CI and the psychological processes we intended to enhance are presented in Table 1.1. (For a complete description of the CI, see Fisher & Geiselman, 1992.)
Table 1.1 Elements of the Cognitive Interview
CI Element | Description | Psychological processes enhanced1 |
Rapport | Develop rapport between respondent and interviewer | Social Dynamics |
Active respondent participation | Respondent actively to generate information (not merely to answer) Interviewer’s questions) | Social Dynamics |
Report everything | Include all recollections in response; do not edit out unimportant details2 | Memory & Communication |
Reinstate context | Reinstate the context of the original experience | Memory |
Describe in detail | Instruct respondents to provide a detailed account3 | Communication |
Close eyes | Instruct respondents to close their eyes4 | Cognition |
No interruptions | Do not interrupt the respondent’s narration5 | Social Dynamics & Cognition |
Don’t guess | Instruct respondents not to guess (OK to say “I don’t know”) | Cognition |
Open-ended questions | Ask primarily open-ended questions (closed questions as follow-up) | Social Dynamics & Cognition |
Multiple retrieval | Encourage respondents to search through memory more than once | Memory |
Varied retrieval | Encourage respondents to search through memory in different ways | Memory |
Respondent- compatible questions | Ask questions that are compatible with respondent’s currently accessible information | Memory |
Avoid suggestive questions | Avoid asking questions that suggest a specific answer | Memory |
Code-compatible output | Allow respondents to output their knowledge in the same form as it is stored (often non-verbal) | Communication |
As is documented in other chapters in this book and also in published reviews and meta-analyses (e.g., Fisher & Geiselman, 2018; Köhnken, Milne, Memon, & Bull, 1999; Memon, Meissner, & Fraser, 2010), the CI has been extremely successful in eliciting more information from witnesses to crime. The present chapter explores how the CI can be adapted to other, non-criminal investigations. Because the CI is a purely process-oriented approach, we expected – in truth, we hoped – that its effectiveness would extend to other kinds of investigation that depend on the same underlying psychological processes that the CI was intended to bolster, namely, (a) the social dynamics between the interviewer and the respondent, (b) the memory and cognitive processes of the respondent and the interviewer, and (c) communication between the interviewer and the respondent. Clearly, there are many kinds of investigative interviews that depend on these three underlying psychological processes, and which may benefit from the CI, including, debriefing military combatants after a mission, gathering human intelligence information from informants, interviewing drivers and passengers about vehicular accidents or interviewing workers about industrial accidents, interviewing patients about their medical histories, etc.
In exploring these non-criminal applications of the CI, we report both our own research efforts and also those of our colleagues, several of whom are contributors to this book. Most of these research studies, naturally, were conducted in controlled, laboratory environments with college students as the respondents in a simulated situation. Some research studies, however, were also conducted outside the laboratory, in the field, where the experimental participants were “real people” engaging in their normal, real-world (non-laboratory) lives. Finally, we describe a few instances in which high-ranking professional investigators incorporated elements of the CI into their investigations of real-world critical events.
Survey of Empirical Research
We take as our jump-off point into reporting the research on non-criminal events the most basic kind of experience: A single, external event that occurs in a specific episodic context (time and place) and the respondent is a cooperative, untrained observer. Within this basic experience, we examine both low-arousal events (e.g., memory for conversations) and high-arousal events (e.g., traffic accidents). From there, we explore events that vary systematically from the basic experience, either because they are collections of many similar events or because the respondent is a trained observer (e.g., police officer). We then examine a more common experience, in which a person’s goal is to extract meaning (e.g., listening to a story). Finally, we explore internal, mental events (e.g., decisions, plans, and emotions), and we speculate on the CI’s use in a therapeutic context.
Before presenting the experimental findings, let us describe the typical research paradigm, including the different variations. Generally, experimental participants experience an event, and then are interviewed about their memory for the event. The researcher typically measures how much the interviewee recalls and the accuracy of his/her recollection. Within this standard procedure there is considerable variation: (a) The experimental participants are usually college students, but in some studies they are young children, or older or non-student adults. (b) The experienced event is usually shown on videotape but sometimes it is experienced live; generally the participants observe the critical event passively, but sometimes they actively participate in the event; the event is usually fashioned by the experimenter, but in some studies, it is a naturally-occurring event from the participant’s life. (c) The interviews are usually conducted immediately or within minutes of the critical event, but in some studies, several days, months, or years intervene. (d) The interview is conducted either as a CI or a “control” interview, which can either be a “Structured Interview” (a technique that incorporates generally accepted principles of interviewing, but does not include the memory-enhancing techniques unique to the CI) or a procedure that simulates a typical police interview. Many of the studies that claimed to conduct the CI used only some, but not all, of the CI elements (see Table 1.1), and so it is important when reading the individual studies to note exactly which CI elements were used. We address this issue later in the chapter (Retrospective Comments on the CI Research).
Single, external events
Low-arousal events: Laboratory studies using the CI have examined, among others, memory for innocuous events like blood donations and health-related events.
Blood donations: In some of the earliest studies on the CI, Köhnken and colleagues assessed the original version of the CI in a blood donation event. Experimental participants were shown a videotape of a typical blood donation event and then were interviewed with either a CI or a Structured Interview. In both Köhnken, Thürer, and Zoberbier (1994) and Mantwill, Köhnken, and Aschermann (1995), interviewees remembered more blood donation details when interviewed with the CI than with the Structured Interview, and at comparable levels of accuracy. This occurred both for participants who had donated blood in the past and for those who had not donated in the past (Mantwill et al., 1995). In Köhnken et al. (1994), not only were the interviewees tested for recall of the blood donation event, but also the interviewers were tested on their recollections of what the interviewees reported during the interview. Interviewers who had conducted a CI recalled more (of what the interviewees had reported) than interviewers who conducted a Structured Interview. Thus the CI not only facilitated observers’ recollection of an external event, but also the interviewers’ recollections of what transpired during the interview.
Accidents: A more common experience that should lend itself to the CI is one in which the witness is a passive bystander observer of either a vehicular or industrial accident. The CI was examined in at least four laboratory studies in which experimental participants watched a video of a car accident (Brock, Fisher, & Cutler, 1999; Chapman & Perry, 1995; Ginet & Verkampt, 2007; Milne, Clare, & Bull, 1999) and one in which the witnesses viewed a video of an industrial accident (MacLean, Stinson, Kelloway, & Fisher, 2011). In all four car accident studies, the CI elicited considerably more information than a Structured Interview, and at comparable accuracy rates. In the one study of memory for an industrial accident, the results were mixed, with the CI eliciting more information than the control interview, but at a lower accuracy rate. Given that the information-processing requirements for watching a video of an industrial accident should be similar to those of watching a car accident, what accounts for the different findings? We suspect that the more encouraging results of the car accident studies reflected that the CI and control interviews were conducted as interactive, face-to-face interviews, whereas the mixed results of the industrial accident reflected that the “interview” was a pre-printed set of instructions to which the witnesses wrote a written response. Moreover, in this study, several of the CI prompts were closed multiple-choice questions, which are prone to error and therefore generally discouraged when conducting a CI. Finally, a critical element of the CI, instructing witnesses not to guess if unsure, was missing from the written CI instructions. We will describe shortly, when discussing high-arousal events, a more compelling examination of the CI in which the witnesses were victims of real car accidents (Ginet, Teissedre, Verkampt, & Fisher, 2016).
Conversations: Relatively little research has been conducted on recalling conversations. It is unfortunate that memory for conversations has been overlooked, given its applied value in several domains: For national security, interviewers may want to learn what transpired during meetings of terrorist groups; business leaders or attorneys may want to know who said what during high-level business discussions; in current vogue in Washington politics, investigators want to know about conversations between American government officials and representatives of foreign governments. At least four studies have been conducted to examine the CI’s effect on participants’ recollections of earlier conversations. In Campos and Alonso-Quecuty (2008), Spanish college-aged students watched a video of two criminals discussing in detail their plans to commit a crime. The participants attempted to remember the conversation 15 minutes later, when they participated in either a CI or an interview modeled after a typical Spanish police interview. The CI-intervi...