Drunk eyewitnesses are more reliable than expected

Imagine you are on a jury: would you trust the testimony of a drunk eyewitness? In a surprising new study, Angelica Hagsand and her colleagues report that drunk witnesses performed just as reliably as sober witnesses at recognising a criminal in a line-up.

One hundred and twenty-three students (60 per cent were women; average age 25) were split into three groups - one third drank orange juice for 15 minutes; another group spent the same time drinking enough orange juice mixed with vodka to reach a blood alcohol concentration (BAC) level of .04 per cent; the final group drank enough vodka and orange to reach a BAC level of .07 per cent. This last value is just below the legal drink driving limit in the UK and USA, and is approximately equivalent to an average-sized man drinking two or three shots of vodka in that time.

Five minutes after they'd finished drinking, the participants watched a five-minute video of a man kidnapping two women at a bus stop, shot from the perspective of a witness. Close views of the man's face were available for a total 31 seconds during the film.

A week later, the participants were invited back and completed a surprise identification task. In a sober state, they saw an 8-man line-up on a computer screen that either did, or did not, feature the kidnapper who they'd seen in the film. The test administrator didn't know which condition participants were in, nor whether the culprit was present. Each participant had to say whether the culprit was in the line-up, answering either "yes", "no the culprit is not present" or "do not remember".

Although better than chance, overall performance was poor, consistent with a great deal of past research showing the limited accuracy of eyewitness memory. Crucially, for both the culprit-present and culprit-absent conditions, there was no difference in accuracy across the different participant groups. This result held even after excluding participants who answered that they could not remember.

In fact, although not a statistically significant difference, the most intoxicated (.07 per cent BAC) participants actually achieved a higher accuracy percentage than the controls in both the culprit-present (47.1 per cent vs. 38.5 per cent) and culprit-absent (56.3 per cent vs. 41.7 per cent) line-up conditions. These results contradicted the researchers' expectations. Based on alcoholic myopia theory (a loss of memory for peripheral details), they predicted that the intoxicated participants would match the controls when the culprit was present, but would make more incorrect identifications when he was absent.

The results also clash with the common sense beliefs of the general public that drunk witnesses will be less reliable than sober witnesses. Given how common it is for witnesses to crimes to be intoxicated, there's been surprisingly little research on how alcohol affects eyewitness performance. Sure, this study has its limitations - the alcohol levels used were only moderate and the crime wasn't a real event - but it makes a welcome contribution to a neglected research area.

_________________________________

Angelica Hagsand, Emma Roos-af-Hjelmsäter, Pär Anders Granhag, Claudia Fahlke, and Anna Söderpalm-Gordh (2013). DO SOBER EYEWITNESSES OUTPERFORM ALCOHOL INTOXICATED EYEWITNESSES IN A LINEUP? The European Journal of Psychology Applied to Legal Context: http://www.usc.es/sepjf/images/documentos/Volumen_5/hagsand%20et%20al.pdf

Post written by Christian Jarrett (@psych_writer) for the BPS Research Digest.