Intoxicated eyewitnesses: The effect of a fully balanced placebo design on event memory and metacognitive control

Abstract

Few studies have examined the impact of alcohol on metacognition for witnessed events. We used a 2 × 2 balanced placebo design, where mock witnesses expected and drank alcohol, did not expect but drank alcohol, did not expect nor drank alcohol, or expected but did not drink alcohol. Participants watched a mock crime in a bar-lab, followed by free recall and a cued-recall test with or without the option to reply "don't know" (DK). Intoxicated mock witnesses' free recall was less complete but not less accurate. During cued-recall, alcohol led to lower accuracy, and reverse placebo participants gave more erroneous and fewer correct responses. Permitting and clarifying DK responses was associated with fewer errors and more correct responses for sober individuals; and intoxicated witnesses were less likely to opt out of erroneous responding to unanswerable questions. Our findings highlight the practical and theoretical importance of examining pharmacological effects of alcohol and expectancies in real-life settings.

And one more for the road: Commentary on the Special Issue on alcohol and eyewitness memory

“…The adapted AMT is an extension of the original AMT, whereby, due to the tendencies of intoxicated individuals to focus on salient and to neglect peripheral details, overall fewer details are recalled. Both Altman et al () and Gawrylowicz et al () reported results in line with the adapted AMT. However, the findings of Flowe et al () did not support the adapted AMT and suggested that intoxicated individuals were able to process both salient and peripheral cues as well as individuals who were not intoxicated.…”

“…Whereas many studies have included a placebo condition in which participants only thought that they were intoxicated but actually had not received any alcohol (i.e., placebo), no study had until now included a condition in which participants did not think that they were intoxicated but actually had received some alcohol (i.e., reverse placebo). Gawrylowicz et al () addressed this issue and conducted a study with a fully balanced placebo design, in which participants watched a mock crime in a bar lab. They found that although drinking alcohol (BAC 0.06%), both knowingly or unknowingly, affected the number of correct details that the participants freely recalled, it did not affect the accuracy rate on the free‐recall task, because they did not provide more incorrect details.…”

“…They found that although drinking alcohol (BAC 0.06%), both knowingly or unknowingly, affected the number of correct details that the participants freely recalled, it did not affect the accuracy rate on the free‐recall task, because they did not provide more incorrect details. However, Gawrylowicz et al () found on the cued‐recall task that the participants in the newly introduced reverse placebo condition performed worse (fewer correct answers and more errors) than the participants in the other conditions, suggesting that people who do not know that they have consumed alcohol might be less careful than people who know that they have consumed alcohol. Finally, when given questions to which they could not know the answer, participants who were intoxicated chose less often the don't‐know option and therefore more often made errors than participants who were not intoxicated.…”

“…Like Gawrylowicz et al (), Flowe et al () used a fully balanced placebo design. Female participants underwent an interactive rape scenario either intoxicated or sober while thinking that they had consumed either alcohol or tonic.…”

“…This Special Issue covers studies with an impressive range of experimental designs. Some studies ensured that the design was completely balanced by either including a reverse placebo condition (Flowe et al, ; Gawrylowicz, Scoboria, Teodorini, & Albery, ) or counterbalancing intoxicated, placebo, and sober conditions at both encoding and retrieval (Evans et al, ). Other studies opted to have more naturalistic settings by testing participants in bars (Altman, McQuiston, & Schreiber Compo, ; Sauerland, Broers, & Van Oorsouw, ) or simulated bars (Mindthoff, Hagsand, Schreiber Compo, & Evans, ).…”