Assessment tools for facial recognition: a systematic review

Vasconcelos, Aline Miranda de; Dantas, Germana Alves de Aguiar; Barros, Marianne Araújo; Gomes, José Aldecyano Lino; Nascimento, Paulo Frassinetti Delfino do; Melo, Géssika Araújo de; Torro-Alves, Nelson

doi:10.1590/0102.3772e41nspe208.en

Special issue: Human Face: Models and applications • Psic.: Teor. e Pesq. 41 (spe2) • 2025 • https://doi.org/10.1590/0102.3772e41nspe208.en linkcopy

Assessment tools for facial recognition: a systematic review

Ferramentas de avaliação para reconhecimento facial: uma revisão sistemática

Authorship SCIMAGO INSTITUTIONS RANKINGS

Face recognition is crucial for social interactions. While the criteria for diagnosing prosopagnosia are not yet firmly established, numerous instruments have been developed in recent years to assess face recognition abilities. In this systematic review, we examined assessment tools used for evaluating face recognition and prosopagnosia. We identified 27 studies in which different tests were described, varying in the specific function assessed (e.g., facial memory versus facial perception), format (paper-based or digital), and administration method (self-administered or examiner-guided). Some of the strengths and weaknesses of these tools are discussed in the paper. These findings are valuable for both clinical and research settings, contributing to the identification of prosopagnosia and guiding appropriate interventions.

Keywords:
face recognition; neuropsychological tests; test validity; literature review

Authors/year	Objective	Tests used for face recognition	Main results
Duchaine & Nakayama (2004)	Testing individuals with developmental prosopagnosia using the BFRT to systematically assess whether patients with developmental prosopagnosia can pass it.	Face One in Ten Test; Faces Old/New 1 Test; Faces Old/New 2 Test; Benton Facial Recognition Test (BFRT).	Patients with DP are generally able to complete the BFRT, but they often require more time to do so compared to healthy patients, although some individuals with DP respond more quickly.
Duchaine, & Nakayama (2006)	Evaluating the validity of the CFMT.	Cambridge Face Memory Test (CFMT); BFRT; Recognition Memory for Faces (RMF).	The comparison of CFMT, BFRT, and RMF revealed that CFMT accurately classified 75% of prosopagnosic individuals, whereas BFRT and RMF correctly classified only 25% and 38%, respectively.
Le Grand et al. (2006)	Investigating what aspects of face processing are impaired/spared in developmental prosopagnosia.	The Famous Face Recognition Tests; “One in Ten” test.	Four participants with DP exhibited a normal pattern of face classification. At least half of the sample displayed impairments in facial identity processing, as evidenced by issues related to the external contour, spacing of internal features, or judgments of attractiveness.
Dobel et al. (2007)	Comparing six cases of congenital prosopagnosia to unimpaired participants using standardized test batteries, tailor-made experimental paradigms, and clinical questionnaires.	Bielefelder Famous Faces Test; Face and Name Learning Test I; Emotion Expression Judgment; Face and Name Learning Test II; BFRT; Delayed Matching to Sample of Faces and Eyeglasses.	Prosopagnosic individuals displayed deficits in recognizing famous faces and had a limited ability to retain new faces for short periods. Their judgment of emotional expression, speech reading, and memory for faces and names was only mildly impaired. No evidence was found for general visual deficits or social dysfunction.
Duchaine et al. (2007)	Investigating whether individuals with Developmental Prosopagnosia (DP) exhibit a specific pattern of deficits concerning global vs. local processing, as previously reported in studies.	Famous faces Test; Cambridge Face Perception Test (CFPT); CFMT; Eyes Test; Global-local Tasks;	Deficits in memory and perception were observed in individuals with DP. In the task that assessed global-local processing performance, individuals with DP displayed average performance similar to that of the control group. They exhibited a typical global advantage and typical global-to-local consistency effects, similar to the individuals in the control group.
Humphreys et al. (2007)	Investigating whether the ability to recognize facial expressions can be preserved in the absence of recognizing facial identity remains controversial.	Famous Face Recognition; Unfamiliar Face Discrimination; BFRT.	Facial expression processing in the group of three individuals with DP indicated that facial expression processing was at normal levels in all three of them. However, it was severely compromised in two individuals when compared individually with the participants with acquired Prosopagnosia (PAd). On the other hand, the two individuals with PAd showed marked difficulties with most expressions.
Bowles et al. (2009)	Assessing the extent to which norms for CFMT and CFPT must consider ageing, sex, and the testing country.	CFMT; CFPT.	The finding that the diagnostic threshold for prosopagnosia was influenced by age, participant ethnicity (within Caucasians), and sex for middle-aged and older adults in the CFPT.
McKone et al. (2011)	Developing the "CFMT-Australian" (CFMT-Aus), which complements the CFMT-original by using ethnicity better matched to a different European subpopulation.	CFMT - Aus; CFMT.	The reliability (0.88) and validity (convergent, divergent using trolleys, inversion effects) of the CFMT-Aus were confirmed. The ethnicity depicted by the face has subtle effects on facial processing, even in normal participants. The CFMT-Aus clarifies the diagnosis of prosopagnosia, and the study revealed that face memory at short intervals (3 minutes, 20 minutes, and 24 hours) leads to overlapping processes in normal participants.
Palermo et al. (2011)	Assessing the extent of holistic processing of facial expressions and facial identity in individuals with DP in comparison to healthy individuals.	Maccs Famous Face Test 2008 (MFFT-08); CFMT; CFPT; Ekman 60 Faces Test.	The clinical group exhibited weaker holistic processing for both facial expression and facial identity information. Therefore, normal facial expression recognition in individuals with DP may be a result of compensatory strategies.
Biotti & Cook (2016)	Describing three complementary experiments that examine emotion recognition in a sample of 17 developmental prosopagnosic individuals.	Twenty-Item Prosopagnosia Index questionnaire (PI20); CFPT; CFMT.	The first two experiments revealed a reduced ability in individuals with DP to classify facial expressions. After dividing the clinical sample into two subgroups, apperceptive and non-apperceptive, it was particularly evident that inaccurate expression categorization was more pronounced in those individuals exhibiting apperceptive profiles.
Esins et al. (2016)	Assessing various aspects of facial recognition as well as other aspects of object recognition.	CFMT; Surprise recognition test; Composite Face Test; Facial Motion Advantage Test.	Significant differences in performance were observed between individuals with prosopagnosia and those in the control group in all face tests, while both groups did not differ in object recognition tests. Prosopagnosic individuals exhibited significant deficits in face recognition tasks, including holistic processing and reduced processing of face configuration information. While the control group recognized dynamic faces better than static faces, this effect was not observed in the clinical group. The clinical group also showed a significantly reduced reliability coefficient in the CFMT.
Ulrich et al. (2016)	Investigating the possibility that some cases of DP are caused by memory impairment, with any perceptual complaints merely allied rather than causal.	CFMT; CFPT; The Mooney Face Task.	Six of the 11 individuals did not show evidence of perceptual impairment. In the remaining five individuals, no single perceptual deficit, or combination of deficits, was necessary or sufficient for poor recognition performance. These data suggest that some cases of DP are better explained by a memorial rather than a perceptual deficit and highlight the relevance of the apperceptive/associative distinction more commonly applied to the allied syndrome of acquired prosopagnosia.
Davies-Thompson et al. (2017)	Evaluating a perceptual learning program that incorporated variations in view and expression, which was aimed at training perceptual stages of face processing with an emphasis on ecological validity.	CFMT; CFPT; RMF; BFRT; Familiarity: Famous faces.	The control task did not affect perception, and training enhanced perceptual sensitivity for the trained faces, which generalized to new, untrained expressions of these faces. The improvement extended to new faces, and the benefits were sustained for three months. The training was more effective for individuals with more pronounced initial perceptual deficits.
Fisher et al. (2017)	Assessing whether the activation of visual working memory for individual faces is selectively impaired in DP.	Famous Faces Test; CFMT; CFPT.	Participants with DP exhibited impaired performance in the identity task but performed at the same level as controls in the expression task. The results indicate a specific deficit in facial visual memory related to identity in individuals with DP.
McKone et al. (2017)	Demonstrating the methodological suitability of CFMT - Chinese.	CFMT - Chinese.	The CFMT: Chinese is methodologically suitable for diagnosing prosopagnosia in Chinese participants, with high internal reliability (α = 0.86), an approximately normal distribution, and sample-standard scores well above chance.
White et al. (2017)	Testing face identification ability in six people with DP, who are severely impaired on face memory tasks, using tasks that do not rely on memory.	CFPT; GFMT.	Participants with DP showed a relatively unimpaired performance on a standard test of face matching ability, indicating that they can achieve normal levels of accuracy on the GFMT despite deficits in central face recognition ability. However, a pronounced deficit in face matching was observed, suggesting that individuals with DP have difficulty recognizing individuals when faced with natural everyday variations in a person's appearance.
Rossion & Michel (2018)	Providing up-to-date performance standards for the BFRT, including the time taken to complete the test, as assessed in an electronic version of this (BFRT-c).	Computerized Benton Facial Recognition Test (BFRT-c)	Although response reliability and accuracy were relatively low, this was primarily due to the significant variability in item difficulty. The response time measured in this version (with an average of 3 minutes) offers a reliable complementary measure.
Arizpe et al. (2019)	Evaluating the utility of the CFMQ in the identification of individuals with prosopagnosia.	Cambridge Face Memory Questionnaire (CFMQ); CFMT; Famous Faces Memory Test (FFMT); PI20.	The CFMQ demonstrated strong validity, with high correlations of 0.44 with CFMT and 0.52 with FFMT. However, cut-off analyses revealed that no CFMQ score produced a combination of clinical-grade sensitivity and positive predictive value in individuals sufficient to support its use as a standalone screening or diagnostic test.
Murray & Bate (2020)	Assessing the test-retest reliability of the Cambridge Face Memory Test (CFMT).	CFMT; CFMT-Aus; CFPT; Famous Face Test.	It was found that there were no significant differences in test-retest effects between participants who completed the tasks online and those who did so in the laboratory. Additionally, performance of the first 48 items of the CFMT was equally sensitive to the total score of all 72 items, suggesting that the test can be shortened for efficiency.
Matsuyoshi & Watanabe (2021)	Examining the relationship between the two representative self-report face recognition questionnaires (PI20 and HK questionnaire) and assessing the extent to which these questionnaires differ in their relationship with face recognition performance in normal populations, excluding predetermined extreme groups.	PI20; 15-item questionnaire in a Hong Kong population (HK questionnaire); Taiwanese Face Memory Test (TFMT).	The results indicated that both questionnaire scores had moderate correlations with face recognition performance (approximately r = 0.3), with the strongest correlation observed for HK11 as compared to PI2.
Stantic et al. (2022)	Developing the OFMT, establishing the range of performance on the measure to assess its utility in measuring individual differences, and comparing the OFMT scores with those of the CFMT and PI-20.	PI20; CFMT; Glasgow Face Matching Test (GFMT); Oxford Face Matching Test (OFMT).	The OFMT was found to be sensitive to individual differences in the typical population as well as in groups of prosopagnosic and over-recognizing individuals. The test-retest reliability of the task was on par with that of the CFMT and the GFMT.
Svart et al. (2022)	Exploring whether individuals with self-reported DP also report signs of other neurodevelopmental disorders, deficits, or conditions (developmental comorbidity).	PI20	Overall, 54% reported a developmental comorbidity, with the majority indicating specific cognitive impairments (e.g., memory) rather than diagnostic categories (e.g., attention deficit hyperactivity disorder).
Burns et al. (2022).	Enhancing the understanding of DP cases that are excluded by more traditionally employed tests, highlighting their limitations, and proposing a novel approach to PD diagnosis.	CFMT; Famous Faces Test; Cambridge Face Perception Test inverted scores (CFPTinv); Cambridge Face Perception Test upright scores (CFPTup); Cambridge Face Perception Test Holistic Perception measure; PI20.	The excluded PD cases were found to exhibit impairments in practically all aspects of facial processing. Furthermore, in some instances, they encountered difficulties comparable to those in classic cases, suggesting key shared characteristics across both subgroups of this condition. Additionally, a new approach was proposed for researchers to comprehensively diagnose and study prosopagnosia.
Bate et al. (2022).	Examining the effectiveness of a facial training program in adults and children with facial recognition difficulties.	CFMT; CFPT; Cambridge Car Memory Test; Cambridge Car Perception Test.	Improvements in facial memory were observed in three of the participants. While there was also improvement in facial perception, there was mixed evidence regarding the facial specificity of these gains. Collectively, these findings suggest plasticity in the human facial recognition system extending at least until mid-adulthood, and they also open the door for longer-term implementations of the face training program, likely to result in greater improvements for both adults and children.
Murray et al. (2022).	Assessing the validity of the BFRT-r in typical participants, providing normalization data for comparison in clinical cases, and evaluating the diagnostic utility of the test along with the BFRT-c in DP.	BFRT-c; Benton Facial Recognition Test- Revised (BFRT-r); CFMT;	The BFRT-r showed good internal reliability and content validity in a sample of typical participants. However, a subsequent assessment revealed that older participants took longer to complete it, underscoring the need for age-appropriate norms. Additionally, when administered to 32 individuals with developmental prosopagnosia, the BFRT-r exhibited greater sensitivity in diagnostic screening compared to the BFRT-c.
Little et al. (2023).	Investigating whether preferences for horizontal information in faces are associated with face recognition abilities in a typical sample and whether such preferences are absent in individuals with developmental prosopagnosia.	CFMT.	Overall, the study demonstrated that the preference for horizontal information in faces is associated with typical variations in facial recognition skills but is not impaired in individuals with DP.
Portch et al. (2023).	Developing an understanding of familiar face recognition in individuals with DP, including the reasons for both successful and unsuccessful performance.	FFMT.	The results suggest that individuals with DP can successfully and frequently employ compensatory strategies to recognize faces, and these strategies support most cases of preserved familial face recognition.

location_on

Instituto de Psicologia, Universidade de Brasília Instituto de Psicologia, Universidade de Brasília, 70910-900 - Brasília - DF - Brazil, Tel./Fax: (061) 274-6455 - Brasília - DF - Brazil
E-mail: revistaptp@gmail.com

rss_feed Stay informed of issues for this journal through your RSS reader

[TFN1] Source: figure prepared by the authors (2023).