Greenberg, M.S., Westcott, D.R., & Bailey, S.E., (1998). When believing is seeing: The effect of scripts on eyewitness memory. Law and Human Behavior, 22, 685-694.

• Scripts organize information about events, roles, and actions into a schematic structure. (We know all about these, right?)
• Eyewitnesses have been shown to use schema consistent expectations in order to support their memory for a crime.
• Problem: Much research exists on system variables in the criminal justice system (e.g., retention interval, wording of questions). However, more research needs to be conducted on estimator variables (e.g., eyewitness expectancies, exposure time, & level of arousal).

• Two conditions should facilitate script-related inferences in memory. (1) "Script items are central, important or highly relevant to the script," and (2) a weak episodic memory for the event.
• Short exposure rate and a long retention interval should result in a weak episodic memory.
• Hypothesis: Eyewitnesses who experience a short exposure rate and long retention interval should have higher false recognition for "central script actions."

• This study was conducted to identify people's scripts for a robbery of a convenience store.
• People were given an example of a restaurant script. Then they were told to list 12 actions that they believe might occur during a robbery of a convenience store.

• People described the typical robber as being a white, 24 year-old male. The cashier was described as typically being a white 31 year-old woman. Most people believed that no customers would be in the store and the robbery would take about 7 minutes.

• Sixteen actions were coded into script categories if there was a 25% or greater consensus. This resulted in nine script actions.

• Most participants did not use experience to come up with typical actions of a robbery. Also, the script related actions were almost identical to those obtained by Holst and Pezdek (1992).

• Purpose: The researchers want to see if the script related actions would be used to generate missing information of the testimony, while testing an estimator variable (exposure rate) and a system variable (retention interval)
• Conditions: Central vs. peripheral script centrality of the omitted action, 2 sec. vs. 8 sec. exposure rate, and 5 min. vs. 1 week retention interval.
• Stimuli: 24 sequential slides showing a robbery of a convenience store by a robber (mid 20's, white male) with a gun. He enters, pretends to shop, goes to counter, and the cashier (white woman, early 20's) rings up the "purchase." The robber presents a gun, the cashier gives money to the robber, who pockets it and leaves to enter a getaway car.

• Procedure: Subjects saw one of two versions of the robbery. Half viewed 21 slides with 3 central script actions omitted, and half viewed 21 slides with 3 peripheral script actions omitted.
  Half of the subjects were exposed to the slides for 2 sec. and half were exposed for 8 sec.
  A verbal recognition test was given 5 min. later or 1 week later.
• Recognition test: People responded yes or no to 18 questions about seen and unseen actions in the slide show. Twelve were filler questions and 6 were critical questions.

• More people said yes to central critical questions than to peripheral critical questions.
• More people said yes to these items in the 1 week retention interval condition.
• A short exposure rate (2 sec.) resulted in more false positives as well.
• When the retention interval was long, people tended to falsely recognize central actions more than peripheral actions.

• People tend to falsely recognize central script actions more than peripheral script actions.
• Longer retention intervals lead to a tendency to fill in missing information with scripts.
• Script centrality and exposure rate had no interaction.
  Possible explanation: Either the crime was unexpected or few people had experienced a similar scenario, so during encoding, they may not have adequately activated the robbery script. The script may have been activated further into the slide show.

Limitations:

• Slides are not as realistic as an actual scene of film.
• The study did not test recall, which is more commonly used in police reports.
• The central script actions are limited to those obtained in Study 1 and may not reflect actions that investigators consider most important.

• The media must greatly shape script representations, since < 10% had experienced a similar situation.
• Police should question witnesses right after a crime to avoid having errors of omission and commission.
• Future research should include the investigation of both estimator and system variables and their interaction.
• These results contribute to past findings (List, 1986) by identifying retention interval as playing an important role in eyewitnesses' script-based inferences.