Two sources of bias explain errors in facial age estimation

Clifford, Colin W. G.; Watson, Tamara; White, David

doi:10.1098/rsos.180841

Cited by 35 publications

(41 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the Guess data we delete all guesses that are more than two times the standard deviation away from the mean age guess on a photograph. We further remove all guesses on photographs that have less than 10 guesses, since it is known that substantial uncertainty in rating ages exists 19,22,23 . This uncertainty mainly arises within guessers among repeated guesses and to a much lower degree among guessers 24 .…”

Section: Data Recordsmentioning

confidence: 99%

“…When they rated ages on AgeGuess.org, they were disappointed to not have performed any different than the users on AgeGuess.org (data not published). Also, when comparing the variance (standard deviation, SD) in the difference between chronological age and perceived age between highly controlled studies 19,22–24 and the AgeGuess data the variance in the difference was comparative to the data generated by the citizen scientists (6–8 in classical studies, 6.9 for the AgeGuess data). This similarity in age estimation might not be expected since the images in the controlled studies have been obtained under strict standardized settings, such as controlled posture, lighting, face expression, clothing, background, distance to camera, make-up, and without accessories such as hats, jewellery, or glasses 8 .…”

Section: Data Recordsmentioning

confidence: 99%

“…3b). A limited age range can also reduce the variance since the raters realize that the persons guessed on are within such an age range 22 .…”

Section: Data Recordsmentioning

confidence: 99%

See 2 more Smart Citations

The AgeGuess database, an open online resource on chronological and perceived ages of people aged 5–100

Jones

Nash

Vieillefont³

et al. 2019

Sci Data

View full text Add to dashboard Cite

In many developed countries, human life expectancy has doubled over the last 180 years. Underlying this higher life expectancy is a change in how we age. Biomarkers of ageing are used to quantify changes in the aging process and to determine biological age. Perceived age is such a biomarker that correlates with biological age. Here we present a unique database rich with possibilities to study the human ageing process. Using perceived age enables us to collect large amounts of data on biological age through a citizen science project, where people upload facial pictures and guess the ages of other people at www.ageguess.org. The data on perceived age we present here span birth cohorts from the years 1877 to 2012. The database currently contains around 220,000 perceived age guesses. Almost 4500 citizen scientists from over 120 countries of origin have uploaded ~4700 facial photographs. Beyond studying the ageing process, the data present a wealth of possibilities to study how humans guess ages and who is better at guessing ages.

show abstract

Section: Data Recordsmentioning

confidence: 99%

Section: Data Recordsmentioning

confidence: 99%

See 1 more Smart Citation

The AgeGuess database, an open online resource on chronological and perceived ages of people aged 5–100

Jones

Nash

Vieillefont³

et al. 2019

Sci Data

View full text Add to dashboard Cite

show abstract

“…6 Studies have shown that facial aging is the most important factor in perceived age, which accounts for up to 20% of the variance in the social evaluation of faces. 10,11 It has been suggested that a significant proportion of facial cosmetic surgery patients seek antiaging procedures due to negative societal views about aging. 12 According to the American Society of Plastic Surgeons, >400,000 facelift, forehead lift, neck lift, and blepharoplasty surgeries were performed in 2018.…”

Section: Introductionmentioning

confidence: 99%

How Old Do I Look? Exploring the Facial Cues of Age in a Tasked Eye-Tracking Study

Liao

Ishii

Chen

et al. 2020

Facial Plastic Surgery & Aesthetic Medicine

View full text Add to dashboard Cite

Importance: This is the first eye-tracking study to use a tasked age estimation paradigm to explore the facial cues of age as seen by casual observers. Objectives: Determine where observers gaze on faces when tasked with estimating an individual's age. Design, Setting, and Participants: This was a prospective controlled experiment, which took place at an academic tertiary referral center. In total, 220 casual observers (80 untasked, 140 tasked) viewed frontal facial images of women while an infrared eye-tracking monitor recorded their eye movements and fixations in real time. Main Outcomes and Measures: Multivariate Hotelling's analysis followed by planned posthypothesis testing was used to compare fixation durations for predefined regions of interest, including the central triangle, upper face, midface, lower face, and neck between tasked and untasked observers. Results: A total of 80 observers (mean age 23.6 years, 53% female) successfully completed the first untasked eyetracking experiment. A total of 140 observers (mean age 26.1 years, 60% female) successfully completed the second age estimation experiment. On multivariate analysis, there were significant differences in the distribution of attention between observers in the two experiments (T 2 = 99.70; F(5,2084) = 19.9012, p < 0.0001). On planned posthypothesis testing, observers attended significantly more to the lower third of the face (0.20 s, p < 0.0001, 95% confidence interval (CI) 0.14-0.27 s) and neck (0.05 s, p = 0.0074, 95% CI 0.01-0.08 s) and less to the upper third of the face (À0.27 s, p < 0.0001, 95% CI À0.40 to À0.14 s) when tasked. There was no significant difference in time spent on the whole face in the two experiments, suggesting that peripheral elements such as hair color or jewelry did not significantly influence gaze patterns. Conclusions and Relevance: Humans form judgments about others every day of their lives, and age perception colors their every interaction. To our knowledge, this study is the first to use eye tracking to investigate facial cues of age. The results showed that when tasked with estimating age, casual observer visual attention was shifted toward the lower face when compared with those who were untasked. These data inform our understanding of facial age perception and potential areas to target for facial rejuvenation.

show abstract

“…Although we prevented any systematic biases of block order by counterbalancing blocks, slightly different results might have been obtained with randomized presentation. This is because age ratings for both faces(Clifford, Watson, & White, 2018;Schweinberger et al, 2010) and voices(Zäske & Schweinberger, 2011) show sequential dependencies, in the form of contrastive adaptation. Note also that blocking stimuli by age (as compared to randomized presentation) would be expected to promote contrastive adaptation, and thus might have counteracted (rather than produced) the abovementioned biases of overestimating young speakers' ages and underestimating old speakers' ages.…”

mentioning

confidence: 99%

The Jena Speaker Set (JESS)—A database of voice stimuli from unfamiliar young and old adult speakers

et al. 2019

View full text Add to dashboard Cite

Here we describe the Jena Speaker Set (JESS), a free database for unfamiliar adult voice stimuli, comprising voices from 61 young (18-25 years) and 59 old (60-81 years) female and male speakers uttering various sentences, syllables, read text, semi-spontaneous speech, and vowels. Listeners rated two voice samples (short sentences) per speaker for attractiveness, likeability, two measures of distinctiveness (Bdeviation^-based [DEV] and Bvoice in the crowd^-based [VITC]), regional accent, and age. Interrater reliability was high, with Cronbach's α between .82 and .99. Young voices were generally rated as more attractive than old voices, but particularly so when male listeners judged female voices. Moreover, young female voices were rated as more likeable than both young male and old female voices. Young voices were judged to be less distinctive than old voices according to the DEV measure, with no differences in the VITC measure. In age ratings, listeners almost perfectly discriminated young from old voices; additionally, young female voices were perceived as being younger than young male voices. Correlations between the rating dimensions above demonstrated (among other things) that DEV-based distinctiveness was strongly negatively correlated with rated attractiveness and likeability. By contrast, VITC-based distinctiveness was uncorrelated with rated attractiveness and likeability in young voices, although a moderate negative correlation was observed for old voices. Overall, the present results demonstrate systematic effects of vocal age and gender on impressions based on the voice and inform as to the selection of suitable voice stimuli for further research into voice perception, learning, and memory.

show abstract

Two sources of bias explain errors in facial age estimation

Cited by 35 publications

References 32 publications

The AgeGuess database, an open online resource on chronological and perceived ages of people aged 5–100

The AgeGuess database, an open online resource on chronological and perceived ages of people aged 5–100

How Old Do I Look? Exploring the Facial Cues of Age in a Tasked Eye-Tracking Study

The Jena Speaker Set (JESS)—A database of voice stimuli from unfamiliar young and old adult speakers

Contact Info

Product

Resources

About