Age estimation of faces: a review

Face age modulates gaze following in young adults

Gaze-following behaviour is considered crucial for social interactions which are influenced by social similarity. We investigated whether the degree of similarity, as indicated by the perceived age of another person, can modulate gaze following. Participants of three different age-groups (18-25; 35-45; over 65) performed an eye movement (a saccade) towards an instructed target while ignoring the gaze-shift of distracters of different age-ranges (6-10; 18-25; 35-45; over 70). The results show that gaze following was modulated by the distracter face age only for young adults. Particularly, the over 70 year-old distracters exerted the least interference effect. The distracters of a similar age-range as the young adults (18-25; 35-45) had the most effect, indicating a blurred own-age bias (OAB) only for the young age group. These findings suggest that face age can modulate gaze following, but this modulation could be due to factors other than just OAB (e.g., familiarity).

Evidence for an own-age bias in age estimation from voices in older persons

Previous studies have investigated the effect of ageing on age estimation from faces as well as the occurrence of an own-age bias in such age estimation from faces. To the best of our knowledge, the occurrence of an own age effect on age estimation from voices has never been examined earlier using an experimental design in which the age of participants (young vs. old) and the age of voice stimuli (young vs. old) were crossed. Results revealed an own-age bias in older adults only. In comparison with younger adults, older participants showed age estimation abilities that are preserved for voices from their own age group and impaired for younger voices. This own age bias was absent in younger participants.

What’s on Your Mind?

Lie detectors can be applied in a wide variety of settings. But this advantage comes with a considerable cost: False positives. The applicability of the Concealed Information Test (CIT) is more limited, yet when it can be applied, the risk of false accusations can be set a priori at a very low level. The CIT assesses the recognition of critical information that is known only by the examiners and the culprit, for example, the face of an accomplice. Large effects are obtained with the CIT, whether combined with peripheral, brain, or motor responses. We see three important challenges for the CIT. First, the false negative rate of the CIT can be substantial, particularly under realistic circumstances. A possible solution seems to restrict the CIT to highly salient details. Second, there exist effective faking strategies. Future research will tell whether faking can be detected or even prevented (e.g., using covert measures). Third, recognition of critical crime details does not necessarily result from criminal activity. It is therefore important to properly embed the CIT in the investigative process, while taking care when drawing conclusions from the test outcome (recognition, not guilt).

Natural experience modulates the processing of older adult faces in young adults and 3-year-old children

Just like other face dimensions, age influences the way faces are processed by adults as well as by children. However, it remains unclear under what conditions exactly such influence occurs at both ages, in that there is some mixed evidence concerning the presence of a systematic processing advantage for peer faces (own-age bias) across the lifespan. Inconsistency in the results may stem from the fact that the individual's face representation adapts to represent the most predominant age traits of the faces present in the environment, which is reflective of the individual's specific living conditions and social experience. In the current study we investigated the processing of younger and older adult faces in two groups of adults (Experiment 1) and two groups of 3-year-old children (Experiment 2) who accumulated different amounts of experience with elderly people. Contact with elderly adults influenced the extent to which both adult and child participants showed greater discrimination abilities and stronger sensitivity to configural/featural cues in younger versus older adult faces, as measured by the size of the inversion effect. In children, the size of the inversion effect for older adult faces was also significantly correlated with the amount of contact with elderly people. These results show that, in both adults and children, visual experience with older adult faces can tune perceptual processing strategies to the point of abolishing the discrimination disadvantage that participants typically manifest for those faces in comparison to younger adult faces.

So you think you look young? Matching older adults’ subjective ages with age estimations provided by younger, middle-aged, and older adults

Perceived age plays an important role in the context of age identity and social interactions. To examine how accurate individuals are in estimating how old they look and how old others are, younger, middle-aged, and older adults rated photographs of older target persons (for whom we had information about objective and subjective age) in terms of age and other characteristics. Whereas the older targets had quite accurate perceptions of how old they look, the raters’ age estimations for those targets were rather inaccurate. Older raters and those with frequent contact with older adults provided the most accurate age estimations. Targets that were rated more favorably with respect to attractiveness or fitness were perceived as younger than those with negative ratings.

Discriminant learning through multiple principal angles for visual recognition

Canonical correlation has been prevalent for multiset-based pairwise subspace analysis. As an extension, discriminant canonical correlations (DCCs) have been developed for classification purpose by learning a global subspace based on Fisher discriminant modeling of pairwise subspaces. However, the discriminative power of DCCs is not optimal as it only measures the "local" canonical correlations within subspace pairs, which lacks the "global" measurement among all the subspaces. In this paper, we propose a multiset discriminant canonical correlation method, i.e., multiple principal angle (MPA). It jointly considers both "local" and "global" canonical correlations by iteratively learning multiple subspaces (one for each set) as well as a global discriminative subspace, on which the angle among multiple subspaces of the same class is minimized while that of different classes is maximized. The proposed computational solution is guaranteed to be convergent with much faster converging speed than DCC. Extensive experiments on pattern recognition applications demonstrate the superior performance of MPA compared to existing subspace learning methods.

Coherence of Personal Narratives across the Lifespan: A Multidimensional Model and Coding Method

Personal narratives are integral to autobiographical memory and to identity, with coherent personal narratives being linked to positive developmental outcomes across the lifespan. In this article, we review the theoretical and empirical literature that sets the stage for a new lifespan model of personal narrative coherence. This new model integrates context, chronology, and theme as essential dimensions of personal narrative coherence, each of which relies upon different developmental achievements and has a different developmental trajectory across the lifespan. A multidimensional method of coding narrative coherence (the Narrative Coherence Coding Scheme or NaCCS) was derived from the model and is described here. The utility of this approach is demonstrated by its application to 498 narratives that were collected in six laboratories from participants ranging in age from 3 years to adulthood. The value of the model is illustrated further by a discussion of its potential to guide future research on the developmental foundations of narrative coherence and on the benefits of personal narrative coherence for different aspects of psychological functioning.

Experts on age estimation

The aim of this investigation was to study the biases and accuracy in age estimation of persons selling alcohol. Two experiments are reported, both suggesting that the accuracy in age estimation of Swedish alcohol salespersons is higher than that of control persons. This expertise in age estimation is probably the result of the extensive training Swedish alcohol salespersons go through as a natural part of their profession. Nonetheless, their estimates were not free from bias. Salespersons overestimated the age of target persons below 20 years of age and thus too young to buy alcohol. The results also revealed that controls, in contrast to salespersons, assimilated their estimates towards their own age (i.e. an own-anchor effect). Furthermore, female participants were shown to estimate the age of old target persons (56-65 years) more accurately than male participants. These results are discussed in relation to previous findings on training in age estimation and present jurisdiction.