Saturday 17 January 2015

Is your face special?

General intelligence runs through psychology like carbon through biology. Nonetheless, the ubiquity of general intelligence does not preclude the existence of some special modular skills, among which the recognition of faces is a strong candidate. Your face is special for reasons you may or may not find palatable: faces have evolved to make people recognisable. We know who you are. More important, we recognise you and remember how you have treated us. No wonder some reviewers want to remain anonymous.

So, it is interesting to find out, with a proper sample, whether face recognition is yet another g loaded task, or something which evolved on its own and is a distinct skill, and not particularly intellectual. Using that old and discredited technique, introspection, it seems to me that recognition is a simple rather than a complex task. Given time and favourable viewing circumstances we can often recognise a face, but no great intellectual insights derive from such recognition. An interesting question is whether face recognition is heritable.

To test that and whether faces are really special it is useful to compare them to the recognition of cars, which are rather recent objects, and not subject to evolutionary influences, not yet anyway. Here are the Plomin gang on the topic of faces, and Nic Shakeshaft’s powerpoint has sent me his presentation below, generously including his notes on many of the slides, which should help you navigate the content. nicholas.shakeshaft@kcl.ac.uk

The genetic specificity of face perception
Shakeshaft, N.C., Schofield, K.L., Plomin, R. King’s College London


Research Question: Specific cognitive abilities are almost invariably found to be highly heritable, and also to correlate substantially with general cognitive ability ('g'), both phenotypically and genetically. Such findings suggest that cognitive abilities in diverse domains share a common genetic aetiology to a considerable extent, falling into a single hierarchy with 'g' at its apex. Recent twin studies have suggested that the ability to recognise faces may be an exception to this pattern, being equally heritable, but phenotypically uncorrelated both with 'g' and with general object recognition. However, genetic associations cannot be determined with confidence from phenotypic correlations alone.
Method: This study presents the first investigation of face perception in a twin sample with adequate power to perform multivariate genetic analyses. More than 2000 19-20 year old UK twins from the Twins Early Development Study (TEDS) completed measures of face perception, object recognition and general (verbal and non-verbal) intelligence. TEDS is a longitudinal cohort study of twins born in England and Wales between 1994 and 1996, representative of the UK's general population. Data were subjected to twin model-fitting analyses to derive estimates
of genetic and environmental influences.
Results and conclusions: Results confirmed the substantial heritability of face perception (62%), and found it to be modestly phenotypically correlated (r < 0.2) both with general object recognition and with 'g'. Model-fitting analyses found that most of the genetic influences on face perception, accounting for most of its total variance, were not shared either with 'g' or with general object recognition.
Discussion: The results suggest that face perception is largely, but not wholly, distinct from other abilities. This makes this ability exceptional among cognitive domains, particularly in terms of its genetic uniqueness, as 'g' typically accounts for the majority of the genetic variance in each domain. Not only does this suggest that faces are 'special', but the existence of a domain largely independent from the traditional cognitive hierarchy may also shed some light on the nature - and limits - of intelligence itself.

https://drive.google.com/file/d/0B3c4TxciNeJZMWR3Zy15YzgzWkE/view?usp=sharing

There is also another paper on this topic, looking at the more general issue of social perception, and tentatively linking it with verbal intelligence.

Are general and social intelligence genetically distinct? Evidence from behavioural genetics. Schofield, K.L., Shakeshaft, N.C., Plomin, R. King’s College London 

Research Question: Studies of human cognitive ability suggest that aspects of social perception, such as the ability to process human faces, are phenotypically and genetically distinct from general cognition. Nonetheless, twin studies indicate that aspects of social perception, including face perception, are significantly heritable. The present study tests the hypotheses that 1) cognitive and self-report measures of social perception are substantially heritable; 2) social perception forms an independently heritable domain, (partially) genetically distinct from general cognitive intelligence ('g'), and more speculatively, 3) social perception is differentially related to verbal and non-verbal g, with the genetic correlation between social perception and g being explained by shared genetic influences with the verbal component specifically.
Method: This research was conducted within the Twins Early Development Study (TEDS), a longitudinal cohort study of twins born between 1994 and 1996. The participants were 1000 pairs of twins (aged 19-20) selected randomly from
the TEDS cohort, who participated via a purpose-built website.
Participants completed a cognitive test of social perception, the Cambridge Face Memory Test (CFMT), and a novel self-report measure, the Aspects of Social Perception (ASP) scale, comprising four subdomains: face identity, object
recognition, face emotion perception, and non-face emotion perception (i.e., from body language/voice). Verbal and non-verbal g, and self-report psychometric measures of social functioning, were completed at previous testing stages.

Results and conclusions: An independently heritable face-processing ability, with heritability of 62%, was confirmed. Self-reported ability in accurately perceiving emotion in faces was 30% heritable, while non-face (body language, voice) emotion perception was 18% heritable. A small (r<0.2)  but highly significant phenotypic correlation between social perception (self reported face and nonface identity and emotion recognition) and verbal g was explained almost entirely by genetic influences; no correlation was found between social perception and non-verbal g. For cognitive measures, the small phenotypic correlation with verbal g was largely genetic in origin, while that with non-verbal g was not.
Discussion: This study introduces a novel measure of social perception, demonstrates the partial independence of social perception from general cognition, and suggests differential associations with verbal and nonverbal cognitive abilities. This is the first in a series of studies intended to assess the genetic influences underlying social intelligence, and the place of this domain in the cognitive hierarchy. Future work will incorporate additional cognitive measures of
social perception. Findings are interpreted in the context of the evolutionary 'bootstrapping' hypothesis of social intelligence, and ongoing work is outlined which further explores the nature of social intelligence, and how it might be characterised in the context of verbal and nonverbal general cognition.

https://drive.google.com/file/d/0B3c4TxciNeJZXzZXblhPZHNuYXc/view?usp=sharing

Summary

Face recognition is highly heritable and weakly related to g so it probably has its own module, or a favoured set of circuits which only partly taps into the central processor power of general intelligence. So, not only is your face is special. but your specialised ability to recognise faces is even more special. Time to look in the mirror, and marvel.

16 comments:

  1. There's more to the story: experience moderates the overlap between face and object recognition: http://bit.ly/1wv2WSA

    ReplyDelete
  2. Any info on the face recognition test? I wonder how much variance there is in this ability (i.e., are there really large individual differences in this ability in a normal sample of people)?

    ReplyDelete
    Replies
    1. Drop Nic a line and he can probably send you some additional data.

      Delete
  3. Emotional memory = face recognition??

    If ''g' exist...


    Santoculto

    ReplyDelete
  4. Thanks for posting this, interesting stuff. I'm high-IQ like the rest of your readers, but well below average in recognizing faces. I've never felt as stupid as I have taking a facial recognition test.

    ReplyDelete
  5. "representative of the UK's general population": does "general" mean 'white'? Or even more specifically 'old British'? Or just ... what?

    ReplyDelete
    Replies
    1. Representing all twins in England and Wales born between 1994 and 1996

      Delete
    2. So, in fact, "general" means nothing whatever - it might as well have been omitted.

      Delete
    3. Yes, probably. However, I think general was used to explain that it was representative of the general population, not a specific clinical sample.

      Delete
  6. My wife may be, I think, better at recognising faces than me in real life, but I am far better at recognising faces on the telly. What can it mean (a) in general, and (b) for the Cambridge test?

    ReplyDelete
  7. Probably a difference between recognising the face (Cambridge test) and recognising the face and putting a name or context to it.

    ReplyDelete
  8. Satoshi Kanazawa in the "Intelligence Paradox" claims that several aspects of social ability are largely uncorrelated with intelligence due to general intelligence's primary use being solving evolutionarily novel problems, although I think other researchers have refuted some of these claims. This study, though, confirms the low correlation between intelligence and face recognition/emotion perception.

    I also thought it was interesting that spatial navigation is entirely uncorrelated with intelligence according to a study he cites on page 69 of "Paradox".

    ReplyDelete
    Replies
    1. Thanks. Its possible that there are some "modules" but I doubt there can be many of them, or the positive manifold would not be encountered so often when a broad range of tests are used on a broad range of people.

      Delete
  9. I refute theory of Kanazawa because to solve novel problems you need solve it and not to be alcoholic or drug-addict. Paradox intelligence or savannah hypothesis start that some incongruent ideas:

    Intelligence= iq only,
    there are novel memes ( novel since what????),
    Alcoholism or homossexuality is to new flying routes of birds bigger heads,
    There is complete free will and or doesn't exist co-pathogenic evolution that regulate and predispose most our behavior.

    Kanazawa mistakes novel routes flying of smarter birds ( higher level of intelligence and creativity and higher perception of errors potentially dangerous linked with natural curiosity) with "novel" behavior. Smarter birds are more curious and higher risk than average birds. Novel behavior can be anything and very probable to be contextual.

    Not correlate with iq doesn't implies that is not intelligence because is evident that is.

    Novel positive behaviors= solve novel problems with new manipulation of perceptual memory and to be drug- user. Can correlate with it but is not true analogy. My opinion.

    Santoculto

    ReplyDelete
  10. Sorry, solve novel problems and NOT TO BE drug-addict.

    Santoculto

    ReplyDelete
  11. ... But the most important here is, what is the motivation to new flying routes??? The answer is simplest, "collective conservation" or survive. Mitigate dangers.

    Santoculto

    ReplyDelete