Alcohol ‘narrows’ attention — which results in drunk witnesses producing less accurate facial composites

Alcohol intoxication can impair the ability of witnesses to properly recall a person’s facial characteristics, according to new research. The new findings support the alcohol myopia theory — that alcohol reduces a person’s attention to peripheral and non-distinct aspects of a situation.

“During criminal investigations when police don’t have enough evidence to identify a suspect, they help witnesses construct an image of the offender’s face from memory, using computer software,” explained study author Alistair J. Harvey of the University of Portsmouth.

“This results in a facial composite or ‘e-fit’, which is then used to help trigger recognition of the suspect in members of the public familiar with the assailant.”

“While scientific understanding of the factors that contribute to good quality face composites is now well developed, prior to this recent study of ours, no one had considered the role of alcohol intoxication on this important aspect of face processing. Hence, our question: do drunk witnesses produce poor facial composites?” Harvey said.

“In the UK alone, police investigate around 700,000 alcohol related violent crimes per year, yet scientists are only just beginning to understand the impact alcohol has on the quality of witness testimony.”

The study of 32 undergraduate students, which was published in the journal Psychopharmacology, found evidence that alcohol narrowed the attention of intoxicated viewers.

“Previous work on sober participants has shown that witnesses tend to construct more accurate representations of the external features (esp. hair) of a recently viewed unfamiliar face than they do of its internal features (eyes, nose, mouth),” Harvey told PsyPost, “and this is probably because good memory of the hair enables the learner to better discriminate the unfamiliar face later, from a wider range of viewing angles (Frowd et al., 2012).”

“However, in this recent study of ours, participants mildly intoxicated with alcohol only showed this normal advantage for remembering external features when the unfamiliar target face was a female with long, styled (i.e. distinctive) hair.”

The participants watched a video of a man or woman describing his or her interests. The following day, the participants then provided a description of the person and were asked to identify features from his or her face from a selection of photos.

Some participants received an alcoholic drink before watching the video, while others received an alcoholic drink on the second day before reconstructing the person from the video, and some participants received an alcoholic drink on both occasions.

“Alcohol participants who viewed an unfamiliar male target face with short, non-distinct hair tended to construct a poorer likeness of its external features the following day than sober counterparts,” Harvey explained.

“We think this effect is caused by alcohol myopia, which describes a narrowing of attention on to only the most salient features of a visual image or scene. Our work suggests that if a drunk witness sees a perpetrator with short, non-distinct hair then that witness is more likely to focus on the internal features of that face and pay less attention to the hair than a sober witness would.”

“So, the bottom line is that drunk witnesses produce composites that are judged generally to be of a poorer likeness to the target face than sober witnesses; and a drunk witness’s construction of the external features of an offender’s face is likely to be poorer than that of a sober witness if the offender happens to have short, unremarkable hair,” Harvey said.

The study — like all research — includes some caveats.

“Our study has three key limitations that we are planning to address in future work. First, we have no direct evidence that drunk witnesses are paying less attention to short, non-distinct hair than sober witnesses — we need measures of participant eye movements during face learning to test this claim,” Harvey explained.

“Second, the ‘hair effect’ we observed was confounded by the gender of each target face (i.e. only our female target faces had distinct hair; only our male target faces had non-distinct hair), so the effect may be one of face gender rather than hair distinctiveness.”

“Finally, our alcohol participants were only mildly intoxicated (mean BAC = 0.05%) so qualitatively different effects may occur at higher levels of alcohol intoxication,” Harvey said.

The study, “Do intoxicated witnesses produce poor facial composite images?” was authored by S. J. Bayless, A. J. Harvey, W. Kneller, and C. D. Frowd.