Poor Sleep May Make It Harder to ID Faces: Study

sleep and faces

The ability to compare faces is impaired with poor sleep, according to a recent study. While it has previously been known that memory for faces is impacted by a lack of sleep, the findings show that non-memory-related tasks like face identification—which are much more common and more important than facial memory—are also not performed as effectively.

The study consisted of two experiments, both of which used the same Glasgow Face-Matching Task (GFMT). The first experiment used 102 participants who reported as having insomnia symptoms. The participants were made to complete the GFMT, which is a face-matching test that also assesses how confident a person is in their choices. They also completed some questionnaires to evaluate their level of insomnia symptoms. The results of the first experiment showed that the normal sleepers accurately matched faces about 80% of the time, while those with insomnia symptoms were correct about 70-75% of the time. In terms of confidence, the normal sleepers expressed confidence for 75% of their correct matches and about 60% of their incorrect matches. Those with insomnia symptoms expressed confidence around 80% of the time for correct matches and 70% for incorrect matches.

The second experiment used a sleep diary to track the sleep habits over three days of 50 participants. Those who were identified as short sleepers were compared to normal sleepers after taking the GFMT. The normal sleepers were correct about 90% of the time and the short sleepers only 80%. In terms of confidence, the normal sleepers had around 80% and 70% confidence for their correct and incorrect answers, while the short sleepers had about 75% and 65% confidence, respectively.

What This Means

The small sample size makes it tricky to draw generalizations from the findings, but the results do suggest certain trends. Most notable is the fact that poor sleep patterns or symptoms of insomnia are associated with lower performance in face matching, even if memory isn’t required. While the difference in raw accuracy percentages is small, context is important. For a security guard who is checking ID and passport photos at an airport, for instance, that 5-10% difference could mean a few hundred or thousand mistakes, possibly more depending on how busy it is. Also troubling is the suggestion that confidence in face matches might actually increase even as effectiveness goes down. The topic is going to need more probing with larger populations, but the results here do suggest that it’s an issue worth investigating.

Beattie, L., et. al., Perceptual impairment in face identification with poor sleep, Royal Society Open Science, 2016; 10.1098/rsos.160321.