The inheritance of face recognition (should you blame your parents if you can't recognise faces?)

By Grant Jacobs 03/03/2010 14

This post was chosen as an Editor's Selection for ResearchBlogging.orgYou’d think that recognising faces is one of those things that we all do well, or at least the vast majority of us do, yet in practice our ability to do this varies.

Recent twin studies present evidence that face recognition is heritable and is a distinct cognitive task in it’s own right.

faceless-head-250pxAt one end of the scale is the inability to recognise faces almost entirely, called prosopagnosia, a term first coined in 1947 by Bodamer in his landmark paper fusing the Greek word prosopon, meaning ‘face’ to agnosia, the term for recognition impairment.

The cartoon to the right is drawn by a child from Mr. Taylor’s class at Rutherford Elementary School in Nanaimo, B.C. as part of a large collection of idioms drawn by 9-11 year-olds. Whether this child was prosopagnosic or not, it’s a fine visual image of what the disorder might be like.

This condition is told in the opening story of Oliver Sacks’ book, The Man Who Mistook His Wife For A Hat, where he recounts the case of the musician Dr. P., who increasingly could not recognise the faces of his students, but could recognise them from their voices, and who would see faces where their were none. Dr. Sacks wonderfully describes this as being Magoo-like, for those that know the cartoon, describing how Dr. P. would occasionally pat the heads of water hydrants or parking meters, mistaking them for small children. If you haven’t read any of Dr. Sacks’ works, please do yourself a favour and find a few copies. (For more on Dr. Sacks’ works, see the first footnote.) He describes, trying to diagnose Dr. P. as ’He faced me with his ears, I came to think, but not with his eyes.’ (Dr. Sacks’ emphasis.)

Interesting insights to prosopagnosia can be seen at a website developed by a prosopagnosic to explain her own condition, where she illustrates the problem of recognising faces using stones.

Extra-ordinary as this deficit may seem, it is surprisingly common. Recent studies suggest that perhaps 2.5% (1 in 40) people inherit a congenital form of prosopagnosia (Gräter et al., 2008; Kennerknecht et al., 2006). There is some suggestion that biological motion deficits may accompany prosopagnosia (Lange et al., 2009).

Famous people with prosopagnosia include the naturalist Jane Goodall.

Let me get back to the story of face recognition in the majority of us.

Duchaine’s research group used the Cambridge Face Memory Test (CFMT) and a classic twin study approach, comparing 164 monozygotic (MZ) and 125 same-sex dizygotic (DZ) twins to examine two main questions:

  1. The extent to which face recognition ability is a genetically-inherited ability, and
  2. The extent to which face recognition is distinct from other cognitive functions such as memory and vision.

Twin studies are a classic means of determining the extent to which a condition is determined by genetic or environmental factors that relies on comparing the association of the trait with two different types of twins.

Monozygotic twins (Source: wikipedia.)
Monozygotic twins (Source: wikipedia.)

Monozygotic, or ’identical’, twins are twins occur when one egg is fertilised by one sperm, which later splits into two, resulting in two foetuses that are genetically identical. (They can acquire epigenetic differences through life, but that’s a topic for another article.)

Dizygotic, ’fraternal’, or ’non-identical’, twins occur when two different eggs are released into the uterus at the same time and fertilised by different sperm. As the two foetuses come from different eggs and sperm they are genetically distinct. Dizygotic twins are more common in older mothers, or mothers using in-vitro fertilisation to aid pregnancy.

For most genes, say a gene that code for an enzyme, have variants. Each variant is a called an allele. Variations on genes can affect traits, such as eye colour or lactose tolerance, the ability to drink milk as an adult. (Variants that are associated with diseases are sometimes called ‘disease genes’; it is not that there is a new gene, but a different variant of it.)

Excepting for genes on the sex chromosomes, we have two copies of each gene, one from each of our parents. Each copy of a gene is an allele (they may be the same allele, or different).

Genes with many common variants are called polymorphic alleles; they have many (poly) types or forms (morphs) of this gene in the population.

1-egg-twinsThe key to twin studies are that monozygotic (MZ) twins have twice as many alleles (variants of genes) in common as dizygotic (DZ) twins.

Ignoring genes on the sex chromosome, each parent has two copies for each gene, two alleles. Each sperm or egg has only one copy of each gene. Which of the two copies of each of the parent’s genes is included in the sperm or egg is more-or-less random, with a 50-50 chance of either one being passed on.

Monozygotic twins come from the same fertilised egg and so have the same mix of alleles.

Dizygotic twins come from different fertilised eggs. Roughly one-half of them will have gotten one of the two alleles, one-half the other. Since about a half of different sperm or egg will have the same allele for each gene, dizygotic twins will, on average, have only a half of their alleles in common.

2-egg-twinsIf a trait is genetic, you expect to see it more commonly shared between monozygotic twins than between dizygotic twins.

The extent to which the trait is genetic can be estimated by compared a large number of MZ and DZ twins, and is expressed as a value from 0 to 1.0, with 1.0 meaning that a trait is completely genetically determined.

Jeremy Wilmer and his colleagues estimated the heritability of face recognition to be 0.7, quite strongly genetic. This indicates that genetics would explain 70% of the effects that determine a person’s ability to recognise faces.

Many different traits have been examined using twin studies. Most have some genetic contribution, which varies from trait to trait. Some examples are given in footnote 2.

These researchers also present evidence that face recognition is distinct from other cognitive functions that might be involved.

As they report in their abstract:

Low correlations between face recognition scores and visual and verbal recognition scores indicate that both face recognition ability itself and its genetic basis are largely attributable to face-specific mechanisms.

which is to say (from the Discussion section):

that face recognition ability overlaps little with other visual and memory abilities,

To look at this they tested 120 of their twin subjects and more than 1500 non-twins using the Cambridge Face Memory Test (CFMT), a verbal paired-associates memory test (VPAM) and an abstract art memory test (AAM). The idea is to test if the ability to do well (or not) in face perception was dependent or independent of other abilities such as memory or vision. If the results for the first test tended to be the same results of either of the latter two tests, then the genetic link would be to more that just face recognition, but also memory or vision in general. What they found was this was not the case: there were poor correlations between the CFMT results and the results of the other two tests.

ResearchBlogging.orgIt is interesting that this research suggests this is specifically about faces, and not other factors that might indirectly impinge on ability to recognise faces.

There appears to be some debate if face recognition is a distributed process, involving different parts of the brain, or one involving a centre within the brain devoted to face recognition. This study supports the latter view.

Not discussed in the paper is if it supports a model where there are a range of abilities in face recognition. It’s tempting to infer this from the graphs in the paper, but realistically the statistics for this need to be done, not eye-balling graphs!

While this evidence suggests genetics is the major factor, environmental factors do play a role in face recognition, for example there is a training effect where ability to recognise faces in others races depends on experience.

Either way, it is a reminder that not all of us can recognise faces well, and evidence that this is mostly a genetic trait affecting a face-specific system in the brain, not mostly a deficit in how we were brought up.


Oliver Sacks’ works

Interested readers may wish to explore the cover art for The Man Who Mistook His Wife For A Hat in different editions, for different languages. Aside from how interesting the variations are, it gives you a good impression of just how popular Sacks’ works are.

A full list of Oliver Sacks’ books, audio tapes, documentaries and so on, can be obtained from his website.

Example heritability scores

Below are some estimated heritability rates for a number of well-known traits. I haven’t time, or space, to quote the sources for these, and in any event the exact values vary from study to study; some are no doubt hotly disputed!  There are hundreds, if not thousands, of results of twin studies to estimate heritability. The ones I have chosen are not particularly special, merely a few selected at random.

  • autism: 0.7 – 0.9
  • ASD (autism spectrum disorders): 0.9
  • bipolar spectrum disorder: 0.71
  • myopia: up to 89% (0.89)
  • intelligence: 0.4 – 0.85 (results vary, but most favour the higher end, i.e. ~0.75)
  • depression: 40% for women; 30% for men (Byers et al suggest this is lower still)
  • tinnitus: 0.11 (tinnitus is a ringing sensation in the ear; value from Kvestad et al, 2010.)


I’m not a neuroscientist (scibling Fabiana Kubke is), although I do have an informal interest in neuroscience as I can see connections between this field and my own (computational biology) and possible research opportunities.


Human face recognition ability is specific and highly heritable.

Jeremy B. Wilmer, Laura Germine, Christopher F. Chabris, Garga Chatterjee, Mark Williams, Eric Loken, Ken Nakayama, and Bradley Duchaine

Proceedings of the Academy of Sciences, USA, early edition (online in advance of print, open access, i.e. free)

Neural and genetic foundations of face recognition and prosopagnosia.

Grüter T, Grüter M, Carbon CC.

J Neuropsychol. 2008 Mar;2(Pt 1):79-97. (Subscription required) (PDF available for free from the author’s website.)

First report of prevalence of non-syndromic hereditary prosopagnosia (HPA)

Ingo Kennerknecht, Thomas Grueter, Brigitte Welling, Sebastian Wentzek, Jürgen Horst, Steve Edwards, Martina Grueter

American Journal of Medical Genetics Part A Volume 140A Issue 15, Pages 1617 – 1622 (Subscription only.)

Impairments of biological motion perception in congenital prosopagnosia.

Lange J, de Lussanet M, Kuhlmann S, Zimmermann A, Lappe M, Zwitserlood P, Dobel C.

PLoS One. 2009 Oct 12;4(10):e7414. (Open access, i.e. free)

More research reporting in Code for life:

Deleting a gene can turn an ovary into a testis in adult mammals

Bioinformatics — computing with biotechnology and molecular biology data.

Autistic children and blood mercury levels

Monkey business, or is my uncle also my Dad?


14 Responses to “The inheritance of face recognition (should you blame your parents if you can't recognise faces?)”

  • And of course the forensic science perspective (because I know you know I’m going to give it!) is that knowing how well people can recognise faces is crucial in cases where identity parades (or similar, such as mug shot books) have been undertaken.
    Elizabeth Loftus is, to my mind, the most famous forensic psychologist who successfully and succinctly addresses the problems of eye witness testimony including how convinced people are about the identifications they make (and the problems this causes because there is nothing more compelling than an eye witness who believes what they are saying, even if they are later shown to be wrong – and in so many cases they can never be shown to be wrong). Shae also covers issues such as cross-racial identification and, perhaps less relevant to this post but still interesting, weapons focus. Her books are fascinating reading….

  • The thought to look up the forensic aspects occurred to me as I wrote it actually, but I let it go as it wasn’t my think and the article was enough work to write as it was! I’d love to hear more about that application of it.

    (Sci-fi thought: imagine screening out eye-witnesses if their genetics indicates that they have poor ability to recognise faces. Closer to today, you could test eye-witnesses ability in face recognition…)

  • I’m not sure whether that would be a practical solution. After all, someone with excellent face recognition abilities might still not have gotten an accurate look, whereas someone with relatively poor abilities might have gotten an extremely good look, or been affected by events in such as way as to overcome their lack of natural ability…

  • You’re taking it a little more seriously that I wrote it! 😉

    Talking more seriously for a moment, my understanding is that people with very poor face recognition (attempt to) use other features such as hair, gait, speech, clothing, etc. My own feelings are that as eye-witnesses they would recognition, but not generally of the faces themselves. I’m wary of thinking they can “overcome” their inability to recognise faces so much as compensate for it using other features of the person. Their extra practice in recognising other features may mean they are particularly good at picking these up. Anecdotally many “disabilities” (I don’t really like the term!) result in “over training” talents most people don’t use as well, e.g. I almost certainly lip-read better than most people with better hearing as a consequence of relying on it more than they do.

    Having said that I recall reading that if a face is “framed” in some ways, it can apparently help. (For example, a beard or other clothing close to the face.)

    It’s worth reading Cecilia Burman’s website (see ‘Interesting insights to prosopagnosia’ in my article):

Site Meter