Our meta-analysis that synthesizes findings from 1,100+ participants in 10 research studies has been published. The results show that alcohol intoxication during memory acquisition reduces the number of correct details recalled about a crime (g = .40) but does not increase the number of incorrect details recalled. Thus, acute alcohol-intoxication at the time of the crime reduces the completeness but not the accuracy of participant witnesses’ memory reports.
The meta-analysis was in order because research on alcohol and eyewitness memory has grown significantly in recent years. All but one of the studies we analysed were published from 2011-2018. Importantly, the results of the meta-analysis contradict the beliefs that people hold on the topic.
Experts were surveyed on their views about factors that affect eyewitness memory accuracy, such as alcohol. The survey results were published in 2001. Crucially, 90% of the experts agreed that alcohol impairs eyewitness performance, with 65% stating that they would be willing to testify before a jury about alcohol's negative effects (Kassin, Tubb, Hosch, & Memon, 2001). However, the body of research on which experts based their opinions was infinitesimal, with only 1 paper on alcohol and recall accuracy existing at the time of the survey. The expert survey is often used by the courts despite it predating nearly all of the research on the topic. It would be of great interest to the courts to find out what experts currently believe about alcohol’s effects given all of the recent research evidence on this topic.
It is crucial that psychologists accurately convey the research findings to the public because of the implications for actual witnesses and victims. For example, one expert recently said to the media: “Mild to moderate drinking does not seem to have much of any effect on memory, but drinking heavily can quickly lead to forgetting,” The expert further opined that alcohol leads to memory gaps and suggested that intoxicated participants may be prone to filling in memory gaps with social information. However, both of these assertions go beyond the research evidence. So far, research has not demonstrated differential rates of forgetting for sober compared to intoxicated people, and research has not found intoxicated individuals are more prone to incorporating misleading information into their memory reports.
Researchers themselves also misstate their own findings. For instance, the abstract of one study we included in the meta-analysis incorrectly stated that alcohol rendered their participant witnesses less accurate. The study actually found that while alcohol intoxication significantly reduced the number of correct details recalled, it did not increase the number of incorrect details recalled. Like most eyewitness recall studies, the researchers found that both sober and intoxicated individuals reported hardly any incorrect details. It is important to accurately and specifically state the research findings to avoid incorrect broad generalisations to actual witnesses and victims about inaccuracy.
We also analysed factors that could moderate the effects of alcohol on recall. One important factor is dose. The effect was largest (g =.72) when participants consumed a high dose (BAC= .10+, which is a level that exceeds the legal limit in many countries, such as the US and UK). Moderate doses of alcohol (BAC = .03-.09) had a smaller, albeit a statistically significant effect (g =.22). Importantly, the dosage effects were restricted to the number of correct details recalled, not the number of incorrect details recalled.
Researchers measured recall in one of two ways: In free‐recall questioning, participant witnesses are allowed to report all that they remember about the event without interruption or further instruction. In cued‐recall questioning, participant witnesses are asked specific questions or given prompts to report certain pieces of information. Most studies incorporated a delay of 3-7 days between the crime and the interview, which was a face-to-face oral interview in the majority of studies. Police interviewers are known to ask both types of recall questions. We found that alcohol decreased the number of correct details recalled on both recall measures, with the effect being larger for cued recall.
We also investigated whether recalled recalling the crime while still intoxicated or after sobering up differentially affected alcohol’s impact on recall. We found that it did not matter – participant witnesses who were interviewed while still intoxicated or after sobering up remembered fewer correct details but they were no more likely to recall erroneous information compared to their sober counterparts regardless of state at interview. These results have been drawn upon by the College of Policing to develop guidance about first accounts with intoxicated witnesses.
Interestingly, the effect of alcohol on recall accuracy depends on whether the information participants are remembering is central (e.g., information about the perpetrator) or peripheral (e.g., information about bystanders). According to alcohol myopia theory, alcohol's pharmacological effects cause an intoxicated person's attention to be allocated to the most central, or immediate and salient, cues in the environment (Steele & Josephs, 1990). Based on this framework, we might expect intoxicated compared to sober individuals to encode fewer peripheral details.
In line with alcohol myopia theory, we found that alcohol reduced only the number of correct peripheral details recalled, not the number of correct central details. For investigators, this finding is quite important. It suggests that investigators should not press intoxicated witnesses for information that is more peripheral to the crime because these witnesses are not likely to know the information. Pressing them for peripheral information might increase recall errors. At trial, the defense may use the witness’ lack of knowledge about peripheral information to undermine the whole of their testimony.
In the paper, we also report analyses of the data in relation to the type of control group employed and interview delay for readers who are interested.
The research indicates that intoxicated witnesses and victims are not more likely than their sober counterparts to in confabulate information during police interviews to make up for being able to recall relatively fewer details about the crime. The findings with intoxicated witnesses align with the rest of the interviewing literature. Factors that strengthen the ability of witnesses’ to recall information about a crime (like being interviewed with the cognitive interview, or being sober during the crime) enable witnesses to report more correct details. Importantly, witnesses who retrieve less information from memory do not recall more incorrect details about the crime.
The paper is free to download until the end of June 2019. Alternatively, the accepted version of the paper can be obtained here.
Jores, T., Colloff, M. Kloft, L., Smailes, H., & Flowe, H. (2019). A meta-analysis of the effects of acute alcohol intoxication on witness recall. Applied Cognitive Psychology, 33, 334-343. DOI: 10.1002/acp.3533
Comentários