External Facial Features Research Articles

Improving Detection of DeepFakes through Facial Region Analysis in Images

In the evolving landscape of digital media, the discipline of media forensics, which encompasses the critical examination and authentication of digital images, videos, and audio recordings, has emerged as an area of paramount importance. This heightened significance is predominantly attributed to the burgeoning concerns surrounding the proliferation of DeepFakes, which are highly realistic and manipulated media content, often created using advanced artificial intelligence techniques. Such developments necessitate a profound understanding and advancement in media forensics to ensure the integrity of digital media in various domains. Current research endeavours are primarily directed towards addressing a common challenge observed in DeepFake datasets, which pertains to the issue of overfitting. Many suggested remedies centre around the application of data augmentation methods, with a frequently adopted strategy being the incorporation of random erasure or cutout. This method entails the random removal of sections from an image to introduce diversity and mitigate overfitting. Generating disparities between the altered and unaltered images serves to inhibit the model from excessively adapting itself to individual samples, thus leading to more favourable results. Nonetheless, the stochastic nature of this approach may inadvertently obscure facial regions that harbour vital information necessary for DeepFake detection. Due to the lack of guidelines on specific regions for cutout, most studies use a randomised approach. However, in recent research, face landmarks have been integrated to designate specific facial areas for removal, even though the selection remains somewhat random. Therefore, there is a need to acquire a more comprehensive insight into facial features and identify which regions hold more crucial data for the identification of DeepFakes. In this study, the investigation delves into the data conveyed by various facial components through the excision of distinct facial regions during the training of the model. The goal is to offer valuable insights to enhance forthcoming face removal techniques within DeepFake datasets, fostering a deeper comprehension among researchers and advancing the realm of DeepFake detection. Our study presents a novel method that uses face cutout techniques to improve understanding of key facial features crucial in DeepFake detection. Moreover, the method combats overfitting in DeepFake datasets by generating diverse images with these techniques, thereby enhancing model robustness. The developed methodology is validated against publicly available datasets like FF++ and Celeb-DFv2. Both face cutout groups surpassed the Baseline, indicating cutouts improve DeepFake detection. Face Cutout Group 2 excelled, with 91% accuracy on Celeb-DF and 86% on the compound dataset, suggesting external facial features’ significance in detection. The study found that eyes are most impactful and the nose is least in model performance. Future research could explore the augmentation policy’s effect on video-based DeepFake detection.

Understanding face detection with visual arrays and real-world scenes

ABSTRACT Face detection has been studied by presenting faces in blank displays, object arrays, and real-world scenes. This study investigated whether these display contexts differ in what they can reveal about detection, by comparing frontal-view faces with those shown in profile (Experiment 1), rotated by 90° (Experiment 2), or turned upside-down (Experiment 3). In blank displays, performance for all face conditions was equivalent, whereas upright frontal faces showed a consistent detection advantage in arrays and scenes. Experiment 4 examined which facial characteristics drive this detection advantage by rotating either the internal or external facial features by 90° while the other features remained upright. Faces with rotated internal features were detected as efficiently as their intact frontal counterparts, whereas detection was impaired when external features were rotated. Finally, Experiment 5 applied Voronoi transformations to scenes to confirm that complexity of stimulus displays modulates the detection advantage for upright faces. These experiments demonstrate that context influences what can be learned about the face detection process. In complex visual arrays and natural scenes, detection proceeds more effectively when external facial features are preserved in an upright orientation. These findings are consistent with a cognitive detection template that focuses on general face-shape information.

Тургеневская традиция в ранней прозе А. П. Чехова (на примере повести «Цветы запоздалые», 1882)

The paper examines how Turgenev’s tradition is reinterpreted in the story “Late-Blooming Flowers,” belonging to the early period of Chekhov’s creative work. Attention is paid to the system of images and techniques of their depiction, the manor world, and reverence for the word. The analysis of Chekhov’s and Turgenev’s landscapes revealed that the ways of depict- ing familiar images were transformed into clichés. Psychological parallelism in Turgenev’s work contributes to the understanding of the characters’ inner states. Chekhov describes na- ture in cursory strokes by combining heterogeneous concepts in one row, leading to a comic effect. He ironically refers to the technique of creating the main characters with external facial features not indicative of the hero’s inner state. The change of literary eras also affects the type of hero, demanding a different artistic challenge. The change of values allows Chekhov to express his attitude to the modern era: the moral values of the past are consigned to oblivi- on. In Chekhov’s story, the manor world is destroyed, with the honor and dignity of the noble word replaced by drunkenness and riotous life. To summarize, Chekhov reinterpreted Turge- nev’s tradition not by criticizing the writing style but by expressing his attitude toward the writing process. For it was no longer possible to write as one did before, the literature of Che- khov’s time required new ways of imagery. It is through rethinking the Turgenev tradition that Chekhov assesses the moral foundation of the modern era, revealing the impossibility of the former way of life to exist.

The importance of internal and external features in recognizing faces that vary in familiarity and race.

Familiar and unfamiliar faces are recognized in fundamentally different ways. One way in which recognition differs is in terms of the features that facilitate recognition: previous studies have shown that familiar face recognition depends more on internal facial features (i.e., eyes, nose and mouth), whereas unfamiliar face recognition depends more on external facial features (i.e., hair, ears and contour). However, very few studies have examined the recognition of faces that vary in both familiarity and race, and the reliance on different facial features, whilst also using faces that incorporate natural within-person variability. In the current study, we used an online version of the card sorting task to assess adults’ (n = 258) recognition of faces that varied in familiarity and race when presented with either the whole face, internal features only, or external features only. Adults better recognized familiar faces than unfamiliar faces in both the whole face and the internal features only conditions, but not in the external features only condition. Reasons why adults did not show an own-race advantage in recognition are discussed.

Eyewitness Memory for Person Identification: Predicting Mugbook Recognition Accuracy According to Person Description Abilities and Subjective Confidence of Witnesses.

Mugbook searches are conducted in case a suspect is not known and to assess if a previously convicted person might be recognized as a potential culprit. The goal of the two experiments reported here was to analyze if prior statements and information about the suspect can aid in the evaluation if such a mugbook search is subsequently advised or not. In experiment 1, memory accuracy for person descriptors was tested in order to analyze, which attributes could be chosen to down-scale the mugbook prior to testing. Results showed that age was the most accurate descriptor, followed by ethnicity and height. At the same time self-assessed low subjective accuracy of culprit descriptions by the witness seemed to be divergent to the objective actual performance accuracy. In experiment 2, a mugbook search was conducted after participants viewed a video of a staged crime and gave a description of the culprit. Results showed that accuracy in mugbook searches correlated positively with the total number of person descriptors given by the witness as well as with witness’ description of external facial features. Predictive confidence (i.e., subjective rating of own performance in the subsequent mugbook search), however did not show any relation to the identification accuracy in the actual mugbook search. These results highlight the notion that mugbooks should not be conducted according to the subjective estimation of the witness’ performance but more according to the actual statements and descriptions that the witness can give about the culprit.

External (Versus Internal) Facial Features Contribute Most to Repetition Priming in Facial Recognition: ERP Evidence.

Previous event-related potential (ERP) research demonstrated four successive ERP components in the repetition priming of human face recognition: P100, N170, N250r, and N400. While these components correspond, respectively, to the four stages proposed by the interactive activation and competition (IAC) model, there has been no emphasis in past research on how internal and external facial features affect repetition priming and the sensitivity of these ERP components to item interval. This study was designed to address these issues. We used faces of celebrities as targets, including completely familiar faces, familiar internal feature faces, and familiar external feature faces. We displayed a target face either immediately following its prime (immediate repetition) or after a delay with interference from a presentation of two other faces (delayed repetition). ERP differences at P100 and N170 were nearly statistically non-significant; familiar faces and familiar external feature faces were associated with reliable ERP signals of N250r and N400 in the immediate repetition condition. For delayed repetition, however, N250r and N400 signals were only preserved for the familiar external feature faces. The differences of these ERP components suggest that, compared with internal facial features, external features of a previously presented face contribute more to brain-based facial repetition priming, particularly during the last two stages of the IAC model.

Unitization of internal and external features contributes to associative recognition for faces: Evidence from modulations of the FN400

Associative recognition requires discriminating between old items and conjunction lures constructed by recombining elements from two different study items. This task can be solved not only by recollection but also by familiarity if the to-be-remembered stimuli are perceived as a unitized representation. In two event-related potential (ERP) studies, we provide evidence for the integration of internal and external facial features by showing that the early frontal old-new effect (considered a correlate of familiarity) is modulated by the specific combination of facial features. Participants studied faces consisting of internal features (eyes, eyebrows, nose, and mouth) paired with external features (hair, head shape, and ears). During the testing phase, intact, recombined, and new faces were presented. Recombined faces consisted of internal and external features taken from two different studied faces. The results showed that at the frontal sites, during the time window from 300 to 500 ms, ERPs to intact faces were more positive than those to new and recombined faces; the latter two did not differ from one another. The late parietal effect was observed only after a more extended study phase in Experiment 2. We take the modulation of the early frontal old-new effect as evidence for the contribution of familiarity to associative recognition for combinations of internal and external facial features.

On the Face of It: No Differential Sensitivity to Internal Facial Features in the Dog Brain.

Dogs are looking at and gaining information from human faces in a variety of contexts. Next to behavioral studies investigating the topic, recent fMRI studies reported face sensitive brain areas in dogs' temporal cortex. However, these studies used whole heads as stimuli which contain both internal (eyes, nose, mouth) and external facial features (hair, chin, face-outline). Behavioral studies reported that (1) recognition of human faces by dogs requires visibility of head contour and that (2) dogs are less successful in recognizing their owners from 2D pictures than from real human heads. In contrast, face perception in humans heavily depends on internal features and generalizes to 2D images. Whether putative face sensitive regions in dogs have comparable properties to those of humans has not been tested so far. In two fMRI experiments, we investigated (1) the location of putative face sensitive areas presenting only internal features of a real human face vs. a mono-colored control surface and (2) whether these regions show higher activity toward live human faces and/or static images of those faces compared to scrambled face images, all with the same outline. In Study 1 (n = 13) we found strong activity for faces in multiple regions, including the previously described temporo-parietal and occipital regions when the control was a mono-colored, homogeneous surface. These differences disappeared in Study 2 (n = 11) when we compared faces to scrambled faces, controlling for low-level visual cues. Our results do not support the assumption that dogs rely on a specialized brain region for processing internal facial characteristics, which is in line with the behavioral findings regarding dogs inability to recognize human faces based on these features.

Lay People Esthetic Evaluation of Primary Surgical Repair on Three-Dimensional Images of Cleft Lip and Palate Patients.

Background and Objectives: Previous literature has disclosed that facial attractiveness affects the esthetic evaluation of nose and lip deformity on frontal and lateral photographs. However, it has never been debated if the removal of the external facial features on three-dimensional (3D) models (“cropped assessment bias”) could provide a considerable usefulness in the interpretation and comparison of the results. Additionally, it has been assumed on two-dimensional (2D) studies that esthetic assessment biases with respect to observer gender, and it is not acknowledged if and to the extent that “gender assessment bias” may be influenced by a three-dimensional layout. The aim of this study is to investigate if facial traits and observers’ gender may affect the esthetic ratings of unilateral cleft lip and palate (UCLP) patients after soft tissue reconstruction. Materials and Methods: Three-dimensional images of ten UCLP patients’ images were acquired before the intervention (T0), one-month (T1) and six-months (T2) postoperative. Geomagic® software (version 2014; 3D Systems, Rock Hill, SC, USA) was used to remove the external facial features of 3D surface models. Five-point scale developed by Asher-McDade et al. was used to rate both nasolabial attractiveness and impairment for full-face (FF) and cropped-face (CF) 3D images. Forty-three judges (21 males, 22 females) were enrolled for the esthetic evaluation. Intraclass correlation coefficient (ICC) was used to test intra- and inter-examiner reliability; a value of 0.7 was set as the minimum acceptable level of reliability. Results: When comparing the 2 sets of observations (FF and CF), the ICC ranged from 0.654 to 0.823. Concerning gender assessment bias, the ICC ranged from 0.438 to 0.686 and from 0.722 to 0.788 for males and females, respectively. Concerning inter-examiner reliability, ICC for questions 2–7 ranged from 0.448 to 0.644 and from 0.659 to 0.817 at T0 and T2, respectively. Conclusions: The removal of external facial features provides subtle differences on the esthetic assessment of UCLP patients. Moreover, based on our data, examiners’ gender differences may affect esthetic assessment of UCLP patients. Despite the subjectivity of esthetic judgments, a reliable, validated and reproducible scoring protocol should consider the influence of gender differences on 3D esthetic assessment of UCLP patients.

From individual features to full faces: Combining aspects of face information.

We investigated how information from face features is combined by comparing sensitivity to individual features with that for external (head shape, hairline) and internal (nose, mouth, eyes, eyebrows) feature compounds. Discrimination thresholds were measured for synthetic faces under the following conditions: (a) full-faces; (b) individual features (e.g., nose); and (c) feature compounds (either external or internal). Individual features and feature compounds were presented both in isolation and embedded within a fixed, task irrelevant face context. Relative to the full-face baseline, threshold elevations for the internal feature compound (2.41x) were comparable to those for the most sensitive individual feature (nose = 2.12x). External features demonstrated the same pattern. A model that incorporated all available feature information within a single channel in an efficient way overestimated sensitivity to feature compounds. Embedding individual features within a task-irrelevant context reduced discrimination sensitivity, relative to isolated presentation. Sensitivity to feature compounds, however, was unaffected by embedding. A loss of sensitivity when embedding features within a fixed-face context is consistent with holistic processing, which limits access to information about individual features. However, holistic combination of information across face features is not efficient: Sensitivity to feature compounds is no better than sensitivity to the best individual feature. No effect of embedding internal feature compounds within task-irrelevant external face features (or vice versa) suggests that external and internal features are processed independently.

Accuracy of capturing oncology facial defects with multimodal image fusion versus laser scanning

Statement of problemFabrication of conventional facial prostheses is a labor-intensive process which traditionally requires an impression of the facial defect and surrounding tissues. Inaccuracies occur during the facial moulage because of soft-tissue compression, the patient's reflex movements, or the lack of support for the impression material. A variety of 3D imaging techniques have been introduced during the production of facial prostheses. However, the accuracy of the different imaging techniques has not been evaluated sufficiently in this clinical context. PurposeThe purpose of this in vitro study was to compare the difference in accuracy of capturing oncology facial defects with multimodal image fusion and laser scanning against a cone beam computed tomography (CBCT) reference scan. Material and methodsTen gypsum casts of oncology facial defects were acquired. To produce reference models, a 3D volumetric scan was obtained using a CBCT scanner and converted into surface data using open-source medical segmentation software. This model was cropped to produce a CBCT mask using an open-source system for editing meshes. The multimodal image fusion model was created using stereophotogrammetry to capture the external facial features and a custom optical structured light scanner to record the defect. The gypsum casts were also scanned using a commercial 3D laser scanner to create the laser-scanned model. Analysis of the best fit of each experimental model to the CBCT mask was performed in MeshLab. The unsigned mean distance was used to measure the absolute deviation of each model from the CBCT mask. A paired-samples t test was conducted to compare the mean global deviation of the 2 imaging modalities from the CBCT masks (α=.05). ResultsA statistically significant difference was found in the mean global deviation between the multimodal imaging model (220 ±50 μm) and the laser-scanned model (170 ±70 μm); (t(9)=2.56, P=.031). The color error maps illustrated that the greatest error was located at sites distant to the prosthesis margins. ConclusionsThe laser-scanned models were more accurate; however, the mean difference of 50 μm is unlikely to be clinically significant. The laser scanner had limited viewing angles and a longer scan time which may limit its transferability to maxillofacial practice.

The impact of external facial features on the construction of facial composites

Witnesses may construct a composite face of a perpetrator using a computerised interface. Police practitioners guide witnesses through this unusual process, the goal being to produce an identifiable image. However, any changes a perpetrator makes to their external facial-features may interfere with this process. In Experiment 1, participants constructed a composite using a holistic interface one day after target encoding. Target faces were unaltered, or had altered external-features: (i) changed hair, (ii) external-features removed or (iii) naturally-concealed external-features (hair, ears, face-shape occluded by a hooded top). These manipulations produced composites with more error-prone internal-features: participants’ familiar with a target’s unaltered appearance less often provided a correct name. Experiment 2 applied external-feature alterations to composites of unaltered targets; although whole-face composites contained less error-prone internal-features, identification was impaired. Experiment 3 replicated negative effects of changing target hair on construction and tested a practical solution: selectively concealing hair and eyes improved identification.Practitioner Summary: The research indicates that when a target identity disguises or changes hair, this can lead to a witness (or victim) constructing a composite that is less readily identified. We assess a practical method to overcome this forensic issue.Abbreviation: GEE: Generalized Estimating Equations

A face is more than just the eyes, nose, and mouth: fMRI evidence for the role of external face features in face recognition

A face is more than just the eyes, nose, and mouth: fMRI evidence for the role of external face features in face recognition

Do intoxicated witnesses produce poor facial composite images?

RationaleThe effect of alcohol intoxication on witness memory and performance has been the subject of research for some time, however, whether intoxication affects facial composite construction has not been investigated.ObjectivesIntoxication was predicted to adversely affect facial composite construction.MethodsThirty-two participants were allocated to one of four beverage conditions consisting of factorial combinations of alcohol or placebo at face encoding, and later construction. Participants viewed a video of a target person and constructed a composite of this target the following day. The resulting images were presented as a full face composite, or a part face consisting of either internal or external facial features to a second sample of participants who provided likeness ratings as a measure of facial composite quality.ResultsIntoxication at face encoding had a detrimental impact on the quality of facial composites produced the following day, suggesting that alcohol impaired the encoding of the target faces. The common finding that external compared to internal features are more accurately represented was demonstrated, even following alcohol at encoding. This finding was moderated by alcohol and target face gender such that alcohol at face encoding resulted in reduced likeness of external features for male composite faces only.ConclusionsModerate alcohol intoxication impairs the quality of facial composites, adding to existing literature demonstrating little effect of alcohol on line-up studies. The impact of intoxication on face perception mechanisms, and the apparent narrowing of processing to external face areas such as hair, is discussed in the context of alcohol myopia theory.

An in-vitro study to assess the feasibility, validity and precision of capturing oncology facial defects with multimodal image fusion

AimAssess the feasibility, validity and precision of multimodal image fusion to capture oncology facial defects based on plaster casts. MethodsTen casts of oncology facial defects were acquired. To create gold standard models, a 3D volumetric scan of each cast was obtained with a cone beam computed tomography (CBCT) scanner (NewTomVG). This was converted into surface data using open-source medical segmentation software and cropped to produce a CBCT mask using an open-source system for editing meshes. For the experimental model, the external facial features were captured using stereophotogrammetry (DI4D) and the defect was recorded with a custom optical structured light scanner. The two meshes were aligned, merged and resurfaced using MeshLab to produce a fused model. Analysis was performed in MeshLab on the best fit of the fused model to the CBCT mask. The unsigned mean distance was used to measure the absolute deviation of each model from the CBCT mask. To assess the precision of the technique, the process of producing the fused model was repeated to create five models each for the casts representing the best, middle and worst results. ResultsGlobal mean deviation was 0.22 mm (standard deviation 0.05 mm). The precision of the method appeared to be acceptable although there was variability in the location of the error for the worst cast. ConclusionThis method for merging two independent scans to produce a fused model shows strong potential as an accurate and repeatable method of capturing facial defects. Further research is required to explore its clinical use.

Effects of nasal drug delivery device and its orientation on sprayed particle deposition in a realistic human nasal cavity

In this study, the effects of nasal drug delivery device and the spray nozzle orientation on sprayed droplets deposition in a realistic human nasal cavity were numerically studied. Prior to performing the numerical investigation, an in-house designed automated actuation system representing mean adults actuation force was developed to produce realistic spray plume. Then, the spray plume development was filmed by high speed photography system, and spray characteristics such as spray cone angle, break-up length, and average droplet velocity were obtained through off-line image analysis. Continuing studies utilizing those experimental data as boundary conditions were applied in the following numerical spray simulations using a commercially available nasal spray device, which was inserted into a realistic adult nasal passage with external facial features. Through varying the particle releasing direction, the deposition fractions of selected particle sizes on the main nasal passage for targeted drug delivery were compared. The results demonstrated that the middle spray direction showed superior spray efficiency compared with upper or lower directions, and the 10µm agents were the most suitable particle size as the majority of sprayed agents can be delivered to the targeted area, the main passage. This study elaborates a comprehensive approach to better understand nasal spray mechanism and evaluate its performance for existing nasal delivery practices. Results of this study can assist the pharmaceutical industry to improve the current design of nasal drug delivery device and ultimately benefit more patients through optimized medications delivery.

Improving Unfamiliar Face Matching by Masking the External Facial Features

SummaryCounter‐terrorism and crime prevention often depend on our ability to match images of unfamiliar faces. For example, when issuing passports, staff must establish an applicant's identity by comparing the submitted photograph witht those in the database of current passports to ensure that multiple documents are not issued to the same person under different names. Previous research has shown that this is a difficult and error prone task. We suggest that this ‘passport problem’ may be due to an over‐reliance on the appearance of external facial features that can be unreliable cues to identity. Compatible with this explanation, we demonstrate that in difficult trials involving a change of appearance or attempted fraud involving a similar looking foil, participants are better able to determine whether two images are of the same person when shown only the internal features of the faces rather than whole images. This discovery has significant practical implications and could form the basis of a procedure to improve the detection of identity fraud.Copyright © 2016 John Wiley & Sons, Ltd.

Judgment of Nasolabial Esthetics in Cleft Lip and Palate Is Not Influenced by Overall Facial Attractiveness.

To determine whether judgment of nasolabial esthetics in cleft lip and palate (CLP) is influenced by overall facial attractiveness. Experimental study. University of Bern, Switzerland. Seventy-two fused images (36 of boys, 36 of girls) were constructed. Each image comprised (1) the nasolabial region of a treated child with complete unilateral CLP (UCLP) and (2) the external facial features, i.e., the face with masked nasolabial region, of a noncleft child. Photographs of the nasolabial region of six boys and six girls with UCLP representing a wide range of esthetic outcomes, i.e., from very good to very poor appearance, were randomly chosen from a sample of 60 consecutively treated patients in whom nasolabial esthetics had been rated in a previous study. Photographs of external facial features of six boys and six girls without UCLP with various esthetics were randomly selected from patients' files. Eight lay raters evaluated the fused images using a 100-mm visual analogue scale. Method reliability was assessed by reevaluation of fused images after >1 month. A regression model was used to analyze which elements of facial esthetics influenced the perception of nasolabial appearance. Method reliability was good. A regression analysis demonstrated that only the appearance of the nasolabial area affected the esthetic scores of fused images (coefficient = -11.44; P < .001; R(2) = 0.464). The appearance of the external facial features did not influence perceptions of fused images. Cropping facial images for assessment of nasolabial appearance in CLP seems unnecessary. Instead, esthetic evaluation can be performed on images of full faces.

The Role of External Features in Face Recognition with Central Vision Loss.

We evaluated how the performance of recognizing familiar face images depends on the internal (eyebrows, eyes, nose, mouth) and external face features (chin, outline of face, hairline) in individuals with central vision loss. In experiment 1, we measured eye movements for four observers with central vision loss to determine whether they fixated more often on the internal or the external features of face images while attempting to recognize the images. We then measured the accuracy for recognizing face images that contained only the internal, only the external, or both internal and external features (experiment 2) and for hybrid images where the internal and external features came from two different source images (experiment 3) for five observers with central vision loss and four age-matched control observers. When recognizing familiar face images, approximately 40% of the fixations of observers with central vision loss was centered on the external features of faces. The recognition accuracy was higher for images containing only external features (66.8 ± 3.3% correct) than for images containing only internal features (35.8 ± 15.0%), a finding contradicting that of control observers. For hybrid face images, observers with central vision loss responded more accurately to the external features (50.4 ± 17.8%) than to the internal features (9.3 ± 4.9%), whereas control observers did not show the same bias toward responding to the external features. Contrary to people with normal vision who rely more on the internal features of face images for recognizing familiar faces, individuals with central vision loss show a higher dependence on using external features of face images.

The Conjunction Effect and Feature Effect in Faces Are Modulated by Task Type

Previous research has suggested that faces can produce robust conjunction and feature effects at item recognition in the feature-conjunction paradigm. However, whether such two effects are affected when the task and contextual details are manipulated, has not been described yet. Also, the contribution of external facial features versus internal facial features to feature effect needs further exploration. Thus, the first goal of the present study was to investigate to what extent that the two effects were affected by task type(the two aspects of episodic memory, i.e., item recognition and source retrieval) for unfamiliar faces in the feature-conjunction paradigm. The second goal was to explore the interaction between the consistency of presentation location and task type. Third, we would study the possible difference between external facial features and internal facial features in the case of feature effect. Thirty-one subjects participated in this experiment. The formal stimuli were 720 headshots of unfamiliar individuals, half male and half female. In the study stage, subjects were asked to learn some unfamiliar faces that were either presented on the left side or on the right side of the screen. In the test stage, five types of faces were displayed, left or right: old faces, conjunction faces, old external feature faces, old internal feature faces, and totally new faces; and two types of tasks were randomly presented: the item recognition task during which subjects only had to make old/new judgments, and the source retrieval task which required an additional discrimination towards the location of old faces(i.e., subjects were instructed to differentiate the old faces that were displayed on the same location between study and test from the old faces on different locations and other faces). The results showed that both tasks elicited robust conjunction effect and feature effect for faces, but they were larger in source retrieval task. The interaction between location consistency and task type was significant in terms of memory performance: At recognition, location consistency facilitated both effects, while no such facilitation was observed at source retrieval. The feature effect in old external faces was similar to that in internal feature faces. These results strongly demonstrate that both conjunction effect and feature effect are sensitive to task type, and suggests an important role of high familiarity compared with recollection in such two effects, consistent with the dual-process model. The discrimination of conjunction and feature faces is regulated by the interaction of whether the presentation location is consistent or not between study and test and task type. Additionally, old external facial features and old internal facial features play similar roles in the contribution to feature effect.