Prevalence of face recognition deficits in middle childhood

ABSTRACT

Approximately 2–2.5% of the adult population is believed to show severe difficulties with face recognition, in the absence of any neurological injury—a condition known as developmental prosopagnosia (DP). However, to date no research has attempted to estimate the prevalence of face recognition deficits in children, possibly because there are very few child-friendly, well-validated tests of face recognition. In the current study, we examined face and object recognition in a group of primary school children (aged 5–11 years), to establish whether our tests were suitable for children and to provide an estimate of face recognition difficulties in children. In Experiment 1 (n = 184), children completed a pre-existing test of child face memory, the Cambridge Face Memory Test–Kids (CFMT-K), and a bicycle test with the same format. In Experiment 2 (n = 413), children completed three-alternative forced-choice matching tasks with faces and bicycles. All tests showed good psychometric properties. The face and bicycle tests were well matched for difficulty and showed a similar developmental trajectory. Neither the memory nor the matching tests were suitable to detect impairments in the youngest groups of children, but both tests appear suitable to screen for face recognition problems in middle childhood. In the current sample, 1.2–5.2% of children showed difficulties with face recognition; 1.2–4% showed face-specific difficulties—that is, poor face recognition with typical object recognition abilities. This is somewhat higher than previous adult estimates: It is possible that face matching tests overestimate the prevalence of face recognition difficulties in children; alternatively, some children may “outgrow” face recognition difficulties.

KEYWORDS:

• Prosopagnosia
• Face recognition
• Object recognition
• Cognitive development
• Developmental disorders

Notes

¹In many cases, age was reported in whole years rather than exact years and months. As such, the mean ages reported here are likely to be lower than the actual mean age of children tested.

²Certainly, children who completed the tests in our lab scored somewhat higher than their school-based counterparts: There was an average difference of 6% for Year 1–3 students, and 11.37% for Year 4–6 students. However, the lab sample was relatively small (N < 10 per age group), and therefore statistical comparisons between the groups lacked power to detect differences.

³The decision to exclude based on attention was made either based on researchers’ observations at time of testing, or if a child responded faster than 300 ms to 10 or more trials on either the face or the bike matching test.

⁴Since transformations alter the relative distance between data-points, 95% CIs are reported rather than standard deviations to provide a more accurate representation of the distribution of data after retransformation.

⁵While an individual child may be classified as above/below the quoted cut-offs from raw scores alone, due to the nature of transformed scores (which change the relative distance between scores) we would not recommend estimating the magnitude of difference from the mean (i.e., z-scores or similar) based on an individual’s raw score. Researchers wishing to quantify results for individual cases should either apply appropriate data transformations or rely on the percentile rank method (Crawford & Garthwaite, 2009; see Supplemental Material for more details).

⁶Unfortunately, familiar face score and parental ratings were not available for the majority of the children who showed face memory difficulties, so we were unable to determine whether these measures were a more effective predictor of memory difficulties.