Tuckey, M.R. & Brewer, N. (2003). The effect of schemas, stimulus ambiguity, and interview schedule, on eyewitness memory over time. Journal of Experimental Psychology: Applied, 9, 101-118.

People have pre-existing ideas about what typically happens during crimes (see Vicki Smith's kicking research on this topic) and they are often interviewed about crimes after substantial delays. Furthermore, a typical eyewitness is interviewed several times. This article presents theory driven research on how these factors interact in producing accurate and inaccurate eyewitness testimony.

Two theoretical frameworks guided the work the authors present here. First, they relied on propositional network theories. These theories picture the mind as being made up of generic and episodic nodes with links between them indicating degree of association. When someone witnesses an event the relevant generic and episodic nodes become activated and the links between the relevant nodes become strengthened. Over time, however, the activation decays and we forget events. Events that violate expectations of a schema receive additional processing because the witness tries to make sense of the expectation violation. Events that match the expectations generated by the schema get a moderate amount of activation. Events that are irrelevant to the schema get relatively little activation. At time of test, cues in the test question activate nodes in the network, and that activation spreads throughout the network to information that is relevant for answering the question. If the activation level rises above a threshold an answer is output. Schema inconsistent information is likely to get output because it received the most processing during study. Schema consistent information is likely to get output both because it received some activation during study, and because it gets additional activation from the generic nodes at retrieval.

The second theoretical framework the authors considered was Fuzzy Trace Theory. Fuzzy trace theory predicts the parallel extraction of verbatim and gist representations. Verbatim traces represent specific item level surface details. Gist traces represent general senses and meanings. Retrieval of a verbatim trace can lead to the acceptance of targets through a process called an identity judgment (e.g. I saw that exact thing) or to rejection of meaning preserving lures through a process called recollection rejection (e.g. The robber had a handgun not a rifle). Gist traces can lad to acceptance of both targets and meaning preserving lures through a process known as a similarity judgment (e.g. I saw something like that). Under some circumstances gist traces can also lead to an illusory experience of recollection called a phantom recollection. Note that fuzzy trace theory characterizes memory traces as sets of bound features with verbatim details decaying more rapidly that gist details. However, repeated testing can lead to trace redintegration.

These theoretical perspectives led Tuckey and Brewer to the following predictions that they test in two experiments:

• Accurate memory should decline across the delay
• Schema irrelevant details should decay more rapidly than schema consistent and inconsistent information
• Repeated testing should refresh memory traces
• Schema consistent false alarms should be more common for ambiguous events than unambiguous events (recollection rejection)
• More remember judgments for schema inconsistent details
• Fewer remember judgments across the delay
• More remember judgments for accurate memories than false memories

Experiment 1

Participants watched a short video depicting a bank robbery. The video included schema consistent, inconsistent, and irrelevant actions. In addition half of the subjects saw a video in which there were a set of ambiguous details that could be interpreted in a schema consistent manner. The other half of the subjects saw a video in which those details were disambiguated.

Participants responded to an open ended free recall section and then a series of cued recall questions. Participants were interviewed according to one of four interview schedules that allowed for the manipulation of retention interview between the event and first interview, as well as for manipulation of repeated interviews.

• Immediate, 8 weeks, 12 weeks
• Immediate, 4 weeks, 12 weeks
• 4 weeks, 12 weeks
• 8 weeks, 12 weeks

Central Results

• For both free and cued recall, accurate recall was reduced when the initial interview was delayed.
• Contrary to the hypothesis, there was not a significant item type (Conistent, Inconsistent, Irrelevant) X Delay interaction for accurate recall.
• For free recall, there was no effect of delay on false memories, although there was a delay by item type interaction because false memories for irrelevant items increased across the delay. For cued recall, false memories increased across the delay.
• For free recall, the amount of accurate schema irrelevant information decreased with the repeated tests more than the other two types of information. In cued recall the amount of accurate schema inconsistent information decreased more than the amount of schema irrelevant information.
• When the first test happened immediately, schema irrelevant information decayed more than schema consistent information. When the first test was delayed, schema irrelevant information increased relevant to the other types of information.
• Schema irrelevant false memories decayed more than schema consistent or irrelevant ones. Fewer false memories occurred when people had an interview at 4 weeks than when people had an interview at 8 weeks.
• More false memories in when evidence was ambiguous than when it was unambiguous.

Experiment 2

Similar to first experiment, however the test now included a written questionnaire in which people made yes/no judgments and cued recall responses to items. They were also asked to make remember/know judgments.

• Fewer accurate responses when the initial interview was delayed
• More decay in schema consistent information at a delayed interview
• Fewer false memories when tested immediately than when tested after a delay
• Remember responses showed the same general pattern as above. Also more remember responses for true than for false memories. Overall number of remember responses declined across the delay. (They have a nice discussion of the pros and cons of making R/K judgments online versus retrospectively)
• As in Experiment 1, ambiguous stimuli led to fewer true memories and more false memories. Moreover, ambiguous stimuli led to more false remember judgments.

Tuckey and Brewer's Discussion

Here are the central conclusions as T&B see it from their study:

• As predicted, less correct information was reported when there was a delay between the event and the first interview. This clearly suggests that police need to interview witnesses as soon as possible in order to archive the witness's recollection before it decays.
• When there was a quick initial test, there was a lot of decay between that test and the next test. When the initial interview was delayed, testimony ended up being more stable across repeated tests. [My Comment: Note that this implies that conducing an immediate interview may result in more accurate and complete testimony, but less testimonial consistency over time.]
• Intrusions appeared to be stable across repeated interviews (consistent with predictions of FTT)
• Ambiguous elements of a crime are likely to be interpreted in a schema consistent manner and experienced in a 'remember' sense.

---

University of Arkansas	Department of Psychology	Graduate Program in Experimental Psychology
Lampinen Lab	False Memory Reading Group	False Memory Reading Group Spring 2004