You may have heard of face-blindness (known formally as prosopagnosia), which is when someone has a particular difficulty recognising familiar faces. The condition was first noticed in brain-damaged soldiers and for a long time psychologists thought it was extremely rare and primarily caused by brain damage. But in recent years they’ve discovered that it’s actually a relatively common condition that some (approximately two per cent of the population) otherwise healthy people are born with. Now research on the related condition of phonagnosia – an impairment in recognising familiar voices – is catching up. A new survey reported in Brain and Language, the largest of its kind published to date, estimates that just over three per cent of the population are born with phonagnosia, many of them probably without even realising it.
Bryan Shilowich and Irving Biederman at the University of Southern California tested 730 people, mostly students, online. Before the voice recognition test began, the participants were first shown pictures of 100 named celebrities and had to say whether their voices were familiar to them. Many participants said they were familiar with all or nearly all the celebrities’ voices. The voices of the 50 most familiar celebrities were then chosen for the main voice recognition test.
On each test item, the participants saw an array of either two or four celebrity faces and underneath were two 6-8 second clips of someone speaking (chosen to ensure there were no clues to their identity, besides the sound of the voice) – one voice belonged to one of the celebrities, the other voice belonged to a non-famous person who was matched for sex, race, age and accent. The participants’ main challenge was to listen to both voices, to identify the celebrity voice, and then to say which of the pictured celebrities the voice belonged to. Finally, there was also an imagination test, in which the participants were asked to say how easily they could imagine the voices of five more celebrities of their choosing, as well as other sounds like breaking glass.
Overall, the participants successfully matched 76.7 per cent of the celebrity voices to the correct celebrity. As you might expect, participants tended to do better at identifying voices belonging to celebrities who they’d expressed greater familiarity with at the start. Also participants who generally had a greater familiarity with more celebrities tended to perform better. There was also a tendency for men and older people to perform better.
However, the researchers were most interested in the spread of scores on the test, once they had statistically accounted for differences in the participants’ levels of familiarity with the different celebrities. This analysis showed that 3.2 per cent of the survey sample (23 people) scored substantially lower than the average – in statistical terms, they scored more than 2.28 standard deviations lower. If the spread of scores were distributed evenly in the population, just like height or weight, we would expect only 8 people, or 1.1 per cent of the sample to score this low. A spike in extreme low scores of this kind is indicative of a subpopulation of people who have an unusually pronounced difficulty with voice recognition.
Indeed, Shilowich and Biederman believe these 23 people likely have developmental phonagnosia, meaning that they were born with the deficit, or it appeared early in their lives, rather than it being caused by brain damage (in which case it would be called “acquired phonagnosia”). Consistent with this, 18 of the 23 low scorers on the recognition test also reported extreme difficulties imagining familiar celebrity voices, even though they had no trouble imagining other sounds. A similar problem with imagining voices was also found to afflict “AN” – one of the few detailed case studies of developmental phonagnosia – presented at a conference in 2013 by Biederman and his colleagues.
The present study wasn’t designed to address what causes developmental phonagnosia, but comprehensive tests with AN suggest that it is not a perceptual deficit as such, but has to do with an inability to match up the perceptual representation of a heard voice with the personal identity information, held in memory, for the person that the face belongs to. However, the reality is we still know very little about developmental phonagnosia. One thing that’s helped research on prosopagnosia or face-blindness is greater public awareness of the condition. Hopefully news of this study might similarly contribute to a greater awareness of phonagnosia thus aiding even more research into the condition.