Understanding face familiarity

The state of modelling face processing in humans with deep learning

Deep learning models trained for facial recognition now surpass the highest performing human participants. Recent evidence suggests that they also model some qualitative aspects of face processing in humans. This review compares the current understanding of deep learning models with psychological models of the face processing system. Psychological models consist of two components that operate on the information encoded when people perceive a face, which we refer to here as 'face codes'. The first component, the core system, extracts face codes from retinal input that encode invariant and changeable properties. The second component, the extended system, links face codes to personal information about a person and their social context. Studies of face codes in existing deep learning models reveal some surprising results. For example, face codes in networks designed for identity recognition also encode expression information, which contrasts with psychological models that separate invariant and changeable properties. Deep learning can also be used to implement candidate models of the face processing system, for example to compare alternative cognitive architectures and codes that might support interchange between core and extended face processing systems. We conclude by summarizing seven key lessons from this research and outlining three open questions for future study.

Challenges of face identification with varied mask coverage in the post COVID-19 era

Introduction

Recent studies have shown that wearing masks can influence face recognition abilities. During the COVID-19 pandemic, people became increasingly familiar with seeing masked faces, leading to a reduced familiarity with fully uncovered faces. With Taiwan now transitioning to a post-COVID-19 phase and the removal of mask mandates, this study investigates how varying levels of mask coverage affect face identification accuracy and response times.

Methods

We examined three levels of mask coverage-full coverage (FC), coverage up to the middle of the nose bridge (MB), and coverage up to the bottom of the nose bridge (BB)-to determine their effects on identification performance. A computer-based simulation was conducted with 100 university students (50 men and 50 women), where participants completed 30 trials (5 trials for each mask coverage level across two target sexes). Each trial presented a masked target face corresponding to one of the three coverage levels, alongside four full-face images. Participants were instructed to choose the image that best matched the masked target face, with an option to select "None" if no match was found.

Results

The findings indicate that faces with FC were identified both faster and more accurately, while those with MB coverage were the most challenging and time-consuming to recognize, particularly for female targets. The performance with BB coverage was intermediate between the other two levels.

Conclusion

This study highlights a notable shift in face identification processes in the aftermath of the pandemic, with FC now leading to quicker and more accurate recognitions, suggesting a significant adaptability in human perceptual mechanisms. These results emphasize the importance of further research into face recognition as we continue to adapt to the pandemic's lasting effects on social interactions and identity verification.

A taxometric analysis of developmental prosopagnosia: Evidence for a categorically distinct impairment

Poor performance on cognitive assessment tasks may indicate a selective 'impairment'. However, it is unclear whether such difficulties separate the individual from the general population qualitatively (i.e., they form a discrete group) or quantitatively (i.e., they represent the lower end of a continuous distribution). Taxometric methods address this question but have rarely been applied to cognitive disorders. This study examined the latent structure of developmental prosopagnosia (DP) - a relatively selective deficit in face recognition that occurs in the absence of neurological injury. Multiple taxometric procedures were applied to dominant diagnostic indices of face recognition ability across two independent datasets. All analyses supported a categorical outcome, even for mild cases of DP, suggesting that it is a qualitatively distinct condition. This finding has significant implications for our understanding of DP given it has traditionally been viewed as a continuous impairment. In particular, existing (arbitrary) diagnostic cut-offs may be too conservative, underestimating prevalence rates and prohibiting big-data approaches to theoretical study. More broadly, these conclusions support application of the taxometric method to many other cognitive processes where weaknesses are predominantly assumed to reside on a continuous distribution.

The neural dynamics of familiar face recognition

Humans are highly efficient at recognising familiar faces. However, previous EEG/ERP research has given a partial and fragmented account of the neural basis of this remarkable ability. We argue that this is related to insufficient consideration of fundamental characteristics of familiar face recognition. These include image-independence (recognition across different pictures), levels of familiarity (familiar faces vary hugely in duration and intensity of our exposure to them), automaticity (we cannot voluntarily withhold from recognising a familiar face), and domain-selectivity (the degree to which face familiarity effects are selective). We review recent EEG/ERP work, combining uni- and multivariate methods, that has systematically targeted these shortcomings. We present a theoretical account of familiar face recognition, dividing it into early visual, domain-sensitive and domain-general phases, and integrating image-independence and levels of familiarity. Our account incorporates classic and more recent concepts, such as multi-dimensional face representation and course-to-fine processing. While several questions remain to be addressed, this new account represents a major step forward in our understanding of the neurophysiological basis of familiar face recognition.

The temporal dynamics of familiar face recognition: Event-related brain potentials reveal the efficient activation of facial identity representations

While it is widely known that humans are typically highly accurate at recognizing familiar faces, it is less clear how efficiently recognition is achieved. In a series of three experiments, we used event-related brain potentials (ERP) in a repetition priming paradigm to examine the efficiency of familiar face recognition. Specifically, we varied the presentation time of the prime stimulus between 500 ms and 33 ms (Experiments 1 and 2), and additionally used backward masks (Experiment 3) to prevent the potential occurrence of visual aftereffects. Crucially, to test for the recognition of facial identity rather than a specific picture, we used different images of the same facial identities in repetition conditions. We observed clear ERP repetition priming effects between 300 and 500 ms after target onset at all prime durations, which suggests that the prime stimulus was sufficiently well processed to allow for facilitated recognition of the target in all conditions. This finding held true even in severely restricted viewing conditions including very brief prime durations and backward masks. We conclude that the facial recognition system is both highly effective and efficient, thus allowing for our impressive ability to recognise the faces that we know.

The importance of conceptual knowledge when becoming familiar with faces during naturalistic viewing

Although the ability to recognise familiar faces is a critical part of everyday life, the process by which a face becomes familiar in the real world is not fully understood. Previous research has focussed on the importance of perceptual experience. However, in natural viewing, perceptual experience with faces is accompanied by increased knowledge about the person and the context in which they are encountered. Although conceptual information is known to be crucial for the formation of new episodic memories, it requires a period of consolidation. It is unclear, however, whether a similar process occurs when we learn new faces. Using a natural viewing paradigm, we investigated how the context in which events are presented influences our understanding of those events and whether, after a period of consolidation, this has a subsequent effect on face recognition. The context was manipulated by presenting events in 1) the original sequence, or 2) a scrambled sequence. Although this manipulation was predicted to have a significant effect on conceptual understanding of events, it had no effect on overall visual experience with the faces. Our prediction was that this contextual manipulation would affect face recognition after the information has been consolidated into memory. We found that understanding of the narrative was greater for participants who viewed the movie in the original sequence compared to those that viewed the movie in a scrambled order. To determine if the context in which the movie was viewed had an effect on face recognition, we compared recognition in the original and scrambled condition. We found an overall effect of conceptual knowledge on face recognition. That is, participants who viewed the original sequence had higher face recognition compared to participants who viewed the scrambled sequence. However, our planned comparisons did not reveal a greater effect of conceptual knowledge on face recognition after consolidation. In an exploratory analysis, we found that overlap in conceptual knowledge between participants was significantly correlated with the overlap in face recognition. We also found that this relationship was greater after a period of consolidation. Together, these findings provide new insights into the role of non-visual, conceptual knowledge for face recognition during natural viewing.

An identity-irrelevant discrimination task reveals familiarity-advantage in face perception and no self-advantage in voice perception

Whether or not self-face and self-voice are processed more accurately than others' remains inconclusive. Most previous studies asked participants to judge the presented stimulus as their own or as others', and compared response accuracy to discuss self-advantage. However, it is possible that participants responded correctly in the "other" trials not by identifying "other" but rather by rejecting "self." The present study employed an identity-irrelevant discrimination task, in which participants detected the odd stimulus among the three sequentially presented stimuli. We measured the discrimination thresholds for the self, friend, and stranger conditions. In Experiment 1 (face), the discrimination thresholds for self and friends' faces were lower than those for strangers' faces. This suggests that self-face may not be perceived as special or unique, and facial representation may become more accurate due to increased familiarity through repetitive exposure. Whereas, in Experiment 2 (voice), the discrimination thresholds did not differ between the three conditions, suggesting that the sensitivity to changes is the same regardless of identity. Overall, we found no evidence for self-advantage in identification accuracy, as we observed a familiarity-advantage rather than self-advantage in face processing and a null difference in voice processing.

Using deep neural networks to disentangle visual and semantic information in human perception and memory

Mental representations of familiar categories are composed of visual and semantic information. Disentangling the contributions of visual and semantic information in humans is challenging because they are intermixed in mental representations. Deep neural networks that are trained either on images or on text or by pairing images and text enable us now to disentangle human mental representations into their visual, visual-semantic and semantic components. Here we used these deep neural networks to uncover the content of human mental representations of familiar faces and objects when they are viewed or recalled from memory. The results show a larger visual than semantic contribution when images are viewed and a reversed pattern when they are recalled. We further reveal a previously unknown unique contribution of an integrated visual-semantic representation in both perception and memory. We propose a new framework in which visual and semantic information contribute independently and interactively to mental representations in perception and memory.

The generality of the attentional boost effect for famous, unfamiliar, and inverted faces

Familiarity and face inversion not only affect face recognition and memory but also influence attention. Face processing is less attention-demanding for familiar than for unfamiliar faces and for upright than for inverted faces. The automaticity raises the question of how face processing interacts with an increase in attention. Using a dual-task paradigm, we tested the interaction between attention and face familiarity and orientation. Participants encoded a series of faces to memory while simultaneously monitoring a stream of colored squares, pressing the space bar for target-colored squares and making no response to distractor-colored squares. Replicating previous findings of the attentional boost effect (ABE), we found that faces encoded with target squares were better remembered than faces encoded with distractor squares. If the automatic nature of familiar (or upright) face processing makes attention unnecessary, then the attentional boost should be attenuated for familiar relative to unfamiliar faces and for upright relative to inverted faces. Data from three experiments showed, however, that the ABE was the same for all types of faces. These results suggest that target detection did not simply elevate attention in an early encoding phase. Rather, selecting targets and rejecting distractors in the color task may have led to yoked temporal selection of target-concurrent faces for entry into memory.

How neural representations of newly learnt faces change over time: Event-related brain potential evidence for overnight consolidation

Previous experiments have shown that a brief encounter with a previously unfamiliar person leads to the establishment of new facial representations, which can be activated by completely novel pictures of the newly learnt face. The present study examined how stable such novel neural representations are over time, and, specifically, how they become consolidated within the first 24 h after learning. Using event-related brain potentials (ERPs) in a between-participants design, we demonstrate that clear face familiarity effects in the occipito-temporal N250 are evident immediately after learning. These effects then undergo change, with a nearly complete absence of familiarity-related ERP differences 4 h after the initial encounter. Critically, 24 h after learning, the original familiarity effect re-emerges. These findings suggest that the neural correlates of novel face representations are not stable over time but change during the first day after learning. The resulting pattern of change is consistent with a process of consolidation.

Neural mechanisms of face familiarity and learning in the human amygdala and hippocampus

Recognizing familiar faces and learning new faces play an important role in social cognition. However, the underlying neural computational mechanisms remain unclear. Here, we record from single neurons in the human amygdala and hippocampus and find a greater neuronal representational distance between pairs of familiar faces than unfamiliar faces, suggesting that neural representations for familiar faces are more distinct. Representational distance increases with exposures to the same identity, suggesting that neural face representations are sharpened with learning and familiarization. Furthermore, representational distance is positively correlated with visual dissimilarity between faces, and exposure to visually similar faces increases representational distance, thus sharpening neural representations. Finally, we construct a computational model that demonstrates an increase in the representational distance of artificial units with training. Together, our results suggest that the neuronal population geometry, quantified by the representational distance, encodes face familiarity, similarity, and learning, forming the basis of face recognition and memory.

Personal familiarity of faces, animals, objects, and scenes: Distinct perceptual and overlapping conceptual representations

While face, object, and scene recognition are often studied at a basic categorization level (e.g. "a face", "a car", "a kitchen"), we frequently recognise individual items of these categories as unique entities (e.g. "my mother", "my car", "my kitchen"). This recognition of individual identity is essential to appropriate behaviour in our world. However, relatively little is known about how we recognise individually familiar visual stimuli. Using event-related brain potentials, the present study examined whether and to what extent the underlying neural representations of personally familiar items are similar or different across different categories. In three experiments, we examined the recognition of personally highly familiar faces, animals, indoor scenes, and objects. We observed relatively distinct familiarity effects in an early time window (200-400 ms), with a clearly right-lateralized occipito-temporal scalp distribution for human faces and more bilateral and posterior distributions for other stimulus categories, presumably reflecting access to at least partly discrete visual long-term representations. In contrast, we found clearly overlapping familiarity effects in a later time window (starting 400 to 500 ms after stimulus onset), again with a mainly right occipito-temporal scalp distribution, for all stimulus categories. These later effects appear to reflect the sustained activation of conceptual properties relevant to any potential interaction. We conclude that familiarity for items from the various visual stimulus categories tested here is represented differently at the perceptual level, while relatively overlapping conceptual mechanisms allow for the preparation of impending potential interaction with the environment.

A narrow band of image dimensions is critical for face recognition

A key challenge in human and computer face recognition is to differentiate information that is diagnostic for identity from other sources of image variation. Here, we used a combined computational and behavioural approach to reveal critical image dimensions for face recognition. Behavioural data were collected using a sorting and matching task with unfamiliar faces and a recognition task with familiar faces. Principal components analysis was used to reveal the dimensions across which the shape and texture of faces in these tasks varied. We then asked which image dimensions were able to predict behavioural performance across these tasks. We found that the ability to predict behavioural responses in the unfamiliar face tasks increased when the early PCA dimensions (i.e. those accounting for most variance) of shape and texture were removed from the analysis. Image similarity also predicted the output of a computer model of face recognition, but again only when the early image dimensions were removed from the analysis. Finally, we found that recognition of familiar faces increased when the early image dimensions were removed, decreased when intermediate dimensions were removed, but then returned to baseline recognition when only later dimensions were removed. Together, these findings suggest that early image dimensions reflect ambient changes, such as changes in viewpoint or lighting, that do not contribute to face recognition. However, there is a narrow band of image dimensions for shape and texture that are critical for the recognition of identity in humans and computer models of face recognition.

Modeling Biological Face Recognition with Deep Convolutional Neural Networks

Deep convolutional neural networks (DCNNs) have become the state-of-the-art computational models of biological object recognition. Their remarkable success has helped vision science break new ground, and recent efforts have started to transfer this achievement to research on biological face recognition. In this regard, face detection can be investigated by comparing face-selective biological neurons and brain areas to artificial neurons and model layers. Similarly, face identification can be examined by comparing in vivo and in silico multidimensional "face spaces." In this review, we summarize the first studies that use DCNNs to model biological face recognition. On the basis of a broad spectrum of behavioral and computational evidence, we conclude that DCNNs are useful models that closely resemble the general hierarchical organization of face recognition in the ventral visual pathway and the core face network. In two exemplary spotlights, we emphasize the unique scientific contributions of these models. First, studies on face detection in DCNNs indicate that elementary face selectivity emerges automatically through feedforward processing even in the absence of visual experience. Second, studies on face identification in DCNNs suggest that identity-specific experience and generative mechanisms facilitate this particular challenge. Taken together, as this novel modeling approach enables close control of predisposition (i.e., architecture) and experience (i.e., training data), it may be suited to inform long-standing debates on the substrates of biological face recognition.

The impact of stability in appearance on the development of facial representations

The way faces become familiar and what information is represented as familiarity develops has puzzled researchers in the field of human face recognition for decades. In this paper, we present three experiments serving as proof of concept for a cost-efficient mechanism of face learning describing how facial representations form over time and accounting for recognition errors. We propose that the encoding of facial information is dynamic and modulated by the intrinsic stability in individual faces' appearance. We drew on a robust and ecological method using a proxy of exposure to famous faces in the real world and manipulated test images to assess the prediction that recognition of famous faces is affected by their relative stability in appearance. We consistently show that stable facial appearances (like Tom Cruise's) facilitate recognition in early stages of familiarisation but that performance does not improve much over time. In contrast, variations in appearance (like Jared Leto's) hinder recognition at first but improve performance with further media exposure. This pattern of results is consistent with the proposed cost-efficient face learning mechanism whereby facial representations build on a foundation of large-scale diagnostic information and refine over time if needed. When coarse information loses its diagnostic value through the experience of variations in appearance across encounters, diagnostic facial details and/or their spatial relationships must receive more weights, leading to refined representations that are more discriminative and reliable than representations of stable faces.

A neural marker of the human face identity familiarity effect

Human adults associate different views of an identity much better for familiar than for unfamiliar faces. However, a robust and consistent neural index of this behavioral face identity familiarity effect (FIFE)-not found in non-human primate species-is lacking. Here we provide such a neural FIFE index, measured implicitly and with one fixation per face. Fourteen participants viewed 70 s stimulation sequences of a large set (n = 40) of widely variable natural images of a face identity at a rate of 6 images/second (6 Hz). Different face identities appeared every 5th image (1.2 Hz). In a sequence, face images were either familiar (i.e., famous) or unfamiliar, participants performing a non-periodic task unrelated to face recognition. The face identity recognition response identified at 1.2 Hz over occipital-temporal regions in the frequency-domain electroencephalogram was 3.4 times larger for familiar than unfamiliar faces. The neural response to familiar faces-which emerged at about 180 ms following face onset-was significant in each individual but a case of prosopdysgnosia. Besides potential clinical and forensic applications to implicitly measure one's knowledge of a face identity, these findings open new perspectives to clarify the neurofunctional source of the FIFE and understand the nature of human face identity recognition.

Developing familiarity during the first eight months of knowing a person: A longitudinal EEG study on face and identity learning

It is well-established that familiar and unfamiliar faces are processed differently, but surprisingly little is known about how familiarity builds up over time and how novel faces gradually become represented in the brain. Here, we used event-related brain potentials (ERPs) in a pre-registered, longitudinal study to examine the neural processes accompanying face and identity learning during the first eight months of knowing a person. Specifically, we examined how increasing real-life familiarity affects visual recognition (N250 Familiarity Effect) and the integration of person-related knowledge (Sustained Familiarity Effect, SFE). Sixteen first-year undergraduates were tested in three sessions, approximately one, five, and eight months after the start of the academic year, with highly variable "ambient" images of a new friend they had met at university and of an unfamiliar person. We observed clear ERP familiarity effects for the new friend after one month of familiarity. While there was an increase in the N250 effect over the course of the study, no change in the SFE was observed. These results suggest that visual face representations develop faster relative to the integration of identity-specific knowledge.

Understanding face matching

Many security settings rely on the identity matching of unfamiliar people, which has led this task to be studied extensively in Cognitive Psychology. In these experiments, observers typically decide whether pairs of faces depict one person (an identity match) or two different people (an identity mismatch). The visual similarity of the to-be-compared faces must play a primary role in how observers accurately resolve this task, but the nature of this similarity-accuracy relationship is unclear. The current study investigated the association between accuracy and facial similarity at the level of individual items (Experiments 1 and 2) and facial features (Experiments 3 and 4). All experiments demonstrate a strong link between similarity and matching accuracy, indicating that this forms the basis of identification decisions. At a feature level, however, similarity exhibited distinct relationships with match and mismatch accuracy. In matches, similarity information was generally shared across the features of a face pair under comparison, with greater similarity linked to higher accuracy. Conversely, features within mismatching face pairs exhibited greater variation in similarity information. This indicates that identity matches and mismatches are characterised by different similarity profiles, which present distinct challenges to the cognitive system. We propose that these identification decisions can be resolved through the accumulation of convergent featural information in matches and the evaluation of divergent featural information in mismatches.

How quickly do we learn new faces in everyday life? Neurophysiological evidence for face identity learning after a brief real-life encounter

Faces learnt in a single experimental session elicit a familiarity effect in event-related brain potentials (ERPs), with more negative amplitudes for newly learnt relative to unfamiliar faces in the N250 component. However, no ERP study has examined face learning following a brief real-life encounter, and it is not clear how long it takes to learn new faces in such ecologically more valid conditions. To investigate these questions, the present study examined whether robust image-independent representations, as reflected in the N250 familiarity effect, could be established after a brief unconstrained social interaction by analysing the ERPs elicited by highly variable images of the newly learnt identity and an unfamiliar person. Significant N250 familiarity effects were observed after a 30-min (Experiment 1) and a 10-min (Experiment 2) encounter, and a trend was observed after 5 min of learning (Experiment 3), demonstrating that 5-10 min of exposure were sufficient for the initial establishment of image-independent representations. Additionally, the magnitude of the effects reported after 10 and 30 min was comparable suggesting that the first 10 min of a social encounter might be crucial, with extra 20 min from the same encounter not adding further benefit for the initial formation of robust face representations.

From Pictures to the People in Them: Averaging Within-Person Variability Leads to Face Familiarization

Familiar faces can be confidently recognized despite sometimes radical changes in their appearance. Exposure to within-person variability-differences in facial characteristics over successive encounters-contributes to face familiarization. Research also suggests that viewers create mental averages of the different views of faces they encounter while learning them. Averaging over within-person variability is thus a promising mechanism for face familiarization. In Experiment 1, 153 Canadian undergraduates (88 female; age: M = 21 years, SD = 5.24) learned six target identities from eight different photos of each target interspersed among 32 distractor identities. Face-matching accuracy improved similarly irrespective of awareness of the target's identity, confirming that target faces presented among distractors can be learned incidentally. In Experiment 2, 170 Canadian undergraduates (125 female; age: M = 22.6 years, SD = 6.02) were tested using a novel indirect measure of learning. The results show that viewers update a mental average of a person's face as it becomes learned. Our findings are the first to show how averaging within-person variability over time leads to face familiarization.

Effect of mask coverage on face identification in Taiwanese men and women

Mask wearing is the easiest and most effective way to avoid COVID-19 infection; however, it affects interpersonal activities, especially face identification. This study examined the effects of three mask coverage levels (full coverage, FC; coverage up to the middle [MB] or bottom of the nose bridge [BB]) on face identification accuracy and time. A total of 115 university students (60 men and 55 women) were recruited to conduct a computer-based simulation test consisting of 30 questions (10 questions [five face images each of men and women] for the three mask coverage levels). One unmasked target face and four face images with a specified mask coverage level were designed for each question, and the participants were requested to select the same face from the four covered face images on the basis of the target face. The ANOVA results indicated that identification accuracy was significantly affected by sex (p < 0.01) and the mask coverage level (p < 0.001), whereas identification time was only influenced by sex (p < 0.05). The multiple comparison results indicated that the identification accuracy rate for faces wearing a mask with FC (90.3%) was significantly lower than for those wearing masks with coverage up to the MB (93.7%) and BB (94.9%) positions; however, no difference in identification accuracy rate was observed between the MB and BB levels. Women exhibited a higher identification accuracy rate than men (94.1% vs. 91.9%) in identifying unfamiliar faces, even though they may spend less time identifying the images. A smaller mask coverage level (i.e., the BB level) does not facilitate face identification. The findings can be served as a reference for people to trade-off between wearing a mask and interpersonal interaction in their daily activities.

Perceptual similarity modulates effects of learning from variability on face recognition

Face recognition is a challenging classification task that humans perform effortlessly for familiar faces. Recent studies have emphasized the importance of exposure to high variability appearances of the same identity to perform this task. However, these studies did not explicitly measure the perceptual similarity between the learned images and the images presented at test, which may account for the advantage of learning from high variability. Particularly, randomly selected test images are more likely to be perceptually similar to learned high variability images, and dissimilar to learned low variability images. Here we dissociated effects of learning from variability and study-test perceptual similarity, by collecting human similarity ratings for the study and test images. Using these measures, we independently manipulated the variability between the learning images and their perceptual similarity to the test images. Different groups of participants learned face identities from a low or high variability set of images. The learning phase was followed by a face matching test (Experiment 1) or a face recognition task (Experiment 2) that presented novel images of the learned identities that were perceptually dissimilar or similar to the learned images. Results of both experiments show that perceptual similarity between study and test, rather than image variability at learning per se, predicts face recognition. We conclude that learning from high variability improves face recognition for perceptually similar but not for perceptually dissimilar images. These findings may not be specific to faces and should be similarly evaluated for other domains.

Top-down modulation and cortical-AMG/HPC interaction in familiar face processing

Humans can accurately recognize familiar faces in only a few hundred milliseconds, but the underlying neural mechanism remains unclear. Here, we recorded intracranial electrophysiological signals from ventral temporal cortex (VTC), superior/middle temporal cortex (STC/MTC), medial parietal cortex (MPC), and amygdala/hippocampus (AMG/HPC) in 20 epilepsy patients while they viewed faces of famous people and strangers as well as common objects. In posterior VTC and MPC, familiarity-sensitive responses emerged significantly later than initial face-selective responses, suggesting that familiarity enhances face representations after they are first being extracted. Moreover, viewing famous faces increased the coupling between cortical areas and AMG/HPC in multiple frequency bands. These findings advance our understanding of the neural basis of familiar face perception by identifying the top-down modulation in local face-selective response and interactions between cortical face areas and AMG/HPC.

What happens to our representation of identity as familiar faces age? Evidence from priming and identity aftereffects

Matching identity in images of unfamiliar faces is error prone, but we can easily recognize highly variable images of familiar faces - even images taken decades apart. Recent theoretical development based on computational modelling can account for how we recognize extremely variable instances of the same identity. We provide complementary behavioural data by examining older adults' representation of older celebrities who were also famous when young. In Experiment 1, participants completed a long-lag repetition priming task in which primes and test stimuli were the same age or different ages. In Experiment 2, participants completed an identity after effects task in which the adapting stimulus was an older or young photograph of one celebrity and the test stimulus was a morph between the adapting identity and a different celebrity; the adapting stimulus was the same age as the test stimulus on some trials (e.g., both old) or a different age (e.g., adapter young, test stimulus old). The magnitude of priming and identity after effects were not influenced by whether the prime and adapting stimulus were the same age or different age as the test face. Collectively, our findings suggest that humans have one common mental representation for a familiar face (e.g., Paul McCartney) that incorporates visual changes across decades, rather than multiple age-specific representations. These findings make novel predictions for state-of-the-art algorithms (e.g., Deep Convolutional Neural Networks).

Asymmetric generalizability of multimodal brain-behavior associations across age-groups

Machine learning methods have increasingly been used to map out brain-behavior associations (BBA), and to predict out-of-scanner behavior of unseen subjects. Given the brain changes that occur in the context of aging, the accuracy of these predictions is likely to depend on how similar the training and testing data sets are in terms of age. To this end, we examined how well BBAs derived from an age-group generalize to other age-groups. We partitioned the CAM-CAN data set (N = 550) into the young, middle, and old age-groups, then used the young and old age-groups to construct prediction models for 11 behavioral outcomes using multimodal neuroimaging features (i.e., structural and resting-state functional connectivity, and gray matter volume/cortical thickness). These models were then applied to all three age-groups to predict their behavioral scores. When the young-derived models were used, a graded pattern of age-generalization was generally observed across most behavioral outcomes-predictions are the most accurate in the young subjects in the testing data set, followed by the middle and then old-aged subjects. Conversely, when the old-derived models were used, the disparity in the predictive accuracy across age-groups was mostly negligible. These findings hold across different imaging modalities. These results suggest the asymmetric age-generalization of BBAs-old-derived BBAs generalized well to all age-groups, however young-derived BBAs generalized poorly beyond their own age-group.

Children's ability to recognize their parent's face improves with age

Adults are experts at recognizing familiar faces across images that incorporate natural within-person variability in appearance (i.e., ambient images). Little is known about children's ability to do so. In the current study, we investigated whether 4- to 7-year-olds (n = 56) could recognize images of their own parent-a person with whom children have had abundant exposure in a variety of different contexts. Children were asked to identify images of their parent that were intermixed with images of other people. We included images of each parent taken both before and after their child was born to manipulate how close the images were to the child's own experience. When viewing before-birth images, 4- and 5-year-olds were less sensitive to identity than were older children; sensitivity did not differ when viewing images taken after the child was born. These findings suggest that with even the most familiar face, 4- and 5-year-olds have difficulty recognizing instances that go beyond their direct experience. We discuss two factors that may contribute to the prolonged development of familiar face recognition.

The influence of familiarity on memory for faces and mask wearing

During the COVID-19 pandemic, the wearing of face masks became mandatory in public areas or at workplaces in many countries. While offering protection, the coverage of large parts of our face (nose, mouth and chin) may have consequences for face recognition. This seems especially important in the context of contact tracing which can require memory of familiar and unfamiliar contacts and whether they were wearing a mask. In this study, we tested how well participants perform at remembering faces and whether they wore a mask, and if this depends on familiarity. Our results show that: (a) Participants remembered familiar faces better than unfamiliar ones, regardless of mask wearing. (b) Participants remembered unmasked faces better than masked faces, regardless of familiarity. (c) Participants were significantly worse at remembering whether an unfamiliar face was wearing a mask or not-even if they indicated remembering the face. (d) Participants showed a bias to indicate no memory of unfamiliar faces. (e) Participants showed a bias to indicate that unfamiliar faces wore a mask, even if they did not. In sum, it was harder to remember both, the identity of unfamiliar faces and whether they wore a mask. These findings have practical relevance for contact tracing and epidemic control.

Face learning via brief real-world social interactions includes changes in face-selective brain areas and hippocampus

Making new acquaintances requires learning to recognise previously unfamiliar faces. In the current study, we investigated this process by staging real-world social interactions between actors and the participants. Participants completed a face-matching behavioural task in which they matched photographs of the actors (whom they had yet to meet), or faces similar to the actors (henceforth called foils). Participants were then scanned using functional magnetic resonance imaging (fMRI) while viewing photographs of actors and foils. Immediately after exiting the scanner, participants met the actors for the first time and interacted with them for 10 min. On subsequent days, participants completed a second behavioural experiment and then a second fMRI scan. Prior to each session, actors again interacted with the participants for 10 min. Behavioural results showed that social interactions improved performance accuracy when matching actor photographs, but not foil photographs. The fMRI analysis revealed a difference in the neural response to actor photographs and foil photographs across all regions of interest (ROIs) only after social interactions had occurred. Our results demonstrate that short social interactions were sufficient to learn and discriminate previously unfamiliar individuals. Moreover, these learning effects were present in brain areas involved in face processing and memory.

Partitioning natural face image variability emphasises within-identity over between-identity representation for understanding accurate recognition

Accurately recognising faces enables social interactions. In recent years it has become clear that people's accuracy differs markedly depending on viewer's familiarity with a face and their individual skill, but the cognitive and neural bases of these accuracy differences are not understood. We examined cognitive representations underlying these accuracy differences by measuring similarity ratings to natural facial image variation. Natural variation was sampled from uncontrolled images on the internet to reflect the appearance of faces as they are encountered in daily life. Using image averaging, and inspired by the computation of Analysis of Variance, we partitioned this variation into differences between faces (between-identity variation) and differences between photos of the same face (within-identity variation). This allowed us to compare modulation of these two sources of variation attributable to: (i) a person's familiarity with a face and, (ii) their face recognition ability. Contrary to prevailing accounts of human face recognition and perceptual learning, we found that modulation of within-identity variation - rather than between-identity variation - was associated with high accuracy. First, familiarity modulated similarity ratings to within-identity variation more than to between-face variation. Second, viewers that are extremely accurate in face recognition - 'super-recognisers' - differed from typical perceivers mostly in their ratings of within-identity variation, compared to between-identity variation. In a final computational analysis, we found evidence that transformations of between- and within-identity variation make separable contributions to perceptual expertise in face recognition. We conclude that inter- and intra-individual accuracy differences primarily arise from differences in the representation of within-identity image variation.

Detecting a viewer's familiarity with a face: Evidence from event-related brain potentials and classifier analyses

Human observers recognize the faces of people they know efficiently and without apparent effort. Consequently, recognizing a familiar face is often assumed to be an automatic process beyond voluntary control. However, there are circumstances in which a person might seek to hide their recognition of a particular face. The present study therefore used event-related potentials (ERPs) and a classifier based on logistic regression to determine if it is possible to detect whether a viewer is familiar with a particular face, regardless of whether the participant is willing to acknowledge it or not. In three experiments, participants were presented with highly variable "ambient" images of personally familiar and unfamiliar faces, while performing an incidental butterfly detection task (Experiment 1), an explicit familiarity judgment task (Experiment 2), and a concealed familiarity task in which they were asked to deny familiarity with one truly known facial identity while acknowledging familiarity with a second known identity (Experiment 3). In all three experiments, we observed substantially more negative ERP amplitudes at occipito-temporal electrodes for familiar relative to unfamiliar faces starting approximately 200 ms after stimulus onset. Both the earlier N250 familiarity effect, reflecting visual recognition of a known face, and the later sustained familiarity effect, reflecting the integration of visual with additional identity-specific information, were similar across experiments and thus independent of task demands. These results were further supported by the classifier analysis. We conclude that ERP correlates of familiar face recognition are largely independent of voluntary control and discuss potential applications in forensic settings.

3D visualisation of psychometric estimates for the ideal male body

Psychological concerns are frequently indexed by psychometric questionnaires but the mental representations that they seek to quantify are difficult to visualise. We used a set of questionnaires designed to measure men's concept of their bodies including: the Drive for Muscularity Scale (DMS; McCreary & Sasse, 2000), the Perceived Sociocultural Pressures Scale (PSPS; Stice, Nemeroff, & Shaw, 1996a), the Body Appreciation Scale (BAS-2; Tylka & Wood-Barcalow, 2015), and the Sociocultural Attitudes Towards Appearance Questionnaire-3 (SATAQ-3; Thompson, van den Berg, Roehrig, Guarda, & Heinberg, 2004). We combined their use with an interactive 3D modelling programme to allow men to create computer-generated representations of their ideal bodies. We used a principal component analysis to extract those shape components of our participants' CGI ideal bodies that were predicted by the questionnaires and reconstructed the body shapes that these questionnaires were capturing. Moving from the lowest to the highest score on both the DMS and SATAQ corresponded with changes in muscularity, particularly muscle mass and definition. This approach allows us to demonstrate the actual body features that are being captured by a particular questionnaire.

The sibling familiarity effect: Is within-person facial variability shared across siblings?

Humans are experts at familiar face recognition, but poor at unfamiliar face recognition. Familiarity is created when a face is encountered across varied conditions, but the way in which a person's appearance varies is identity-specific, so familiarity with one identity does not benefit recognition of other individuals. However, the faces of biological siblings share structural similarities, so we explored whether the benefits of familiarity are shared across siblings. Results show that familiarity with one half of a sibling pair improves kin detection (experiment 1), and that unfamiliar face matching is more accurate when targets are the siblings of familiar versus unfamiliar individuals (experiment 2). PCA applied to facial images of celebrities and their siblings demonstrates that faces are generally better reconstructed in the principal components of a same-sex sibling than those of an unrelated individual. When we encounter the unfamiliar sibling of someone we already know, our pre-existing representation of their familiar relation may usefully inform processing of the unfamiliar face. This can benefit both kin detection and identity processing, but the benefits are constrained by the degree to which facial variability is shared.

When experience does not promote expertise: security professionals fail to detect low prevalence fake IDs

Professional screeners frequently verify photograph IDs in such industries as professional security, bar tending, and sales of age-restricted materials. Moreover, security screening is a vital tool for law enforcement in the search for missing or wanted persons. Nevertheless, previous research demonstrates that novice participants fail to spot fake IDs when they are rare (i.e., the low prevalence effect; LPE). To address whether this phenomenon also occurs with professional screeners, we conducted three experiments. Experiment 1 compared security professional and non-professionals. Experiment 2 compared bar-security professionals, access-security professionals, and non-professionals. Finally, Experiment 3 added a newly created Professional Identity Training Questionnaire to determine whether and how aspects of professionals’ employment predict ID-matching accuracy. Across all three experiments, all participants were susceptible to the LPE regardless of professional status. Neither length/type of professional experience nor length/type of training experience affected ID verification performance. We discuss task performance and survey responses with aims to acknowledge and address this potential problem in real-world screening scenarios.

The effect of expertise, target usefulness and image structure on visual search

Experts outperform novices on many cognitive and perceptual tasks. Extensive training has tuned experts to the most relevant information in their specific domain, allowing them to make decisions quickly and accurately. We compared a group of fingerprint examiners to a group of novices on their ability to search for information in fingerprints across two experiments-one where participants searched for target features within a single fingerprint and another where they searched for points of difference between two fingerprints. In both experiments, we also varied how useful the target feature was and whether participants searched for these targets in a typical fingerprint or one that had been scrambled. Experts more efficiently located targets when searching for them in intact but not scrambled fingerprints. In Experiment 1, we also found that experts more efficiently located target features classified as more useful compared to novices, but this expert-novice difference was not present when the target feature was classified as less useful. The usefulness of the target may therefore have influenced the search strategies that participants used, and the visual search advantages that experts display appear to depend on their vast experience with visual regularity in fingerprints. These results align with a domain-specific account of expertise and suggest that perceptual training ought to involve learning to attend to task-critical features.

Perceptual difficulty modulates the direction of information flow in familiar face recognition

Humans are fast and accurate when they recognize familiar faces. Previous neurophysiological studies have shown enhanced representations for the dichotomy of familiar vs. unfamiliar faces. As familiarity is a spectrum, however, any neural correlate should reflect graded representations for more vs. less familiar faces along the spectrum. By systematically varying familiarity across stimuli, we show a neural familiarity spectrum using electroencephalography. We then evaluated the spatiotemporal dynamics of familiar face recognition across the brain. Specifically, we developed a novel informational connectivity method to test whether peri-frontal brain areas contribute to familiar face recognition. Results showed that feed-forward flow dominates for the most familiar faces and top-down flow was only dominant when sensory evidence was insufficient to support face recognition. These results demonstrate that perceptual difficulty and the level of familiarity influence the neural representation of familiar faces and the degree to which peri-frontal neural networks contribute to familiar face recognition.

Insights from computational models of face recognition: A reply to Blauch, Behrmann and Plaut

We agree with Blauch, Behrmann, and Plaut (2020) on a number of points, and are reassured that their data bear out our previous findings. We discuss differences in modelling style, and the usefulness of different types of model for supporting psychological understanding. We emphasise the role that within-person variability plays in recognising familiar faces and clarify the range over which it is idiosyncratic. The combination of image analysis with top-down support to cohere different images of the same person seems to be an important characteristic of successful models.

Multiple-image arrays in face matching tasks with and without memory

Previous research has shown that exposure to within-person variability facilitates face learning. A different body of work has examined potential benefits of providing multiple images in face matching tasks. Viewers are asked to judge whether a target face matches a single face image (as when checking photo-ID) or multiple face images of the same person. The evidence here is less clear, with some studies finding a small multiple-image benefit, and others finding no advantage. In four experiments, we address this discrepancy in the benefits of multiple images from learning and matching studies. We show that multiple-image arrays only facilitate face matching when arrays precede targets. Unlike simultaneous face matching tasks, sequential matching and learning tasks involve memory and require abstraction of a stable representation of the face from the array, for subsequent comparison with a target. Our results show that benefits from multiple-image arrays occur only when this abstraction is required, and not when array and target images are available at once. These studies reconcile apparent differences between face learning and face matching and provide a theoretical framework for the study of within-person variability in face perception.

Seeing through disguise: Getting to know you with a deep convolutional neural network

People use disguise to look unlike themselves (evasion) or to look like someone else (impersonation). Evasion disguise challenges human ability to see an identity across variable images; Impersonation challenges human ability to tell people apart. Personal familiarity with an individual face helps humans to see through disguise. Here we propose a model of familiarity based on high-level visual learning mechanisms that we tested using a deep convolutional neural network (DCNN) trained for face identification. DCNNs generate a face space in which identities and images co-exist in a unified computational framework, that is categorically structured around identity, rather than retinotopy. This allows for simultaneous manipulation of mechanisms that contrast identities and cluster images. In Experiment 1, we measured the DCNN's baseline accuracy (unfamiliar condition) for identification of faces in no disguise and disguise conditions. Disguise affected DCNN performance in much the same way it affects human performance for unfamiliar faces in disguise (cf. Noyes & Jenkins, 2019). In Experiment 2, we simulated familiarity for individual identities by averaging the DCNN-generated representations from multiple images of each identity. Averaging improved DCNN recognition of faces in evasion disguise, but reduced the ability of the DCNN to differentiate identities of similar appearance. In Experiment 3, we implemented a contrast learning technique to simultaneously teach the DCNN appearance variation and identity contrasts between different individuals. This facilitated identification with both evasion and impersonation disguise. Familiar face recognition requires an ability to group images of the same identity together and separate different identities. The deep network provides a high-level visual representation for face recognition that supports both of these mechanisms of face learning simultaneously.

Deep learning of shared perceptual representations for familiar and unfamiliar faces: Reply to commentaries

We recently argued that human unfamiliar face identity perception reflects substantial perceptual expertise, and that the advantage for familiar over unfamiliar face identity matching reflects a learned mapping between generic high-level perceptual features and a unique identity representation of each individual (Blauch, Behrmann and Plaut, 2020). Here we respond to two commentaries by Young and Burton (2020) and Yovel and Abudarham (2020), clarifying and elaborating our stance on various theoretical issues, and discussing topics for future research in human face recognition and the learning of perceptual representations.

Contrast Negation Impairs Sorting Unfamiliar Faces by Identity: A Comparison With Original (Contrast-Positive) and Stretched Images

Recognition of unfamiliar faces is difficult in part due to variations in expressions, angles, and image quality. Studies suggest shape and surface properties play varied roles in face learning, and identification of unfamiliar faces uses diagnostic pigmentation/surface reflectance relative to shape information. Here, participants sorted photo-cards of unfamiliar faces by identity, which were shown in their original, stretched, and contrast-negated forms, to examine the utility of diagnostic shape and surface properties in sorting unfamiliar faces by identity. In four experiments, we varied the presentation order of conditions (contrast-negated first or original first with stretched second across experiments) and whether the same or different photo-cards were seen across conditions. Stretching the images did not impair performance in any measures relative to other conditions. Contrast negation generally exacerbated poor sorting by identity compared with the other conditions. However, seeing the contrast-negated photo-cards last mitigated some of the effects of contrast negation. Together, results suggest an important role for surface properties such as pigmentation and reflectance for sorting by identity and add to literatures on informational content and appearance variability in discrimination of facial identity.

Getting to Know Someone: Familiarity, Person Recognition, and Identification in the Human Brain

In our everyday life, we continuously get to know people, dominantly through their faces. Several neuroscientific experiments showed that familiarization changes the behavioral processing and underlying neural representation of faces of others. Here, we propose a model of the process of how we actually get to know someone. First, the purely visual familiarization of unfamiliar faces occurs. Second, the accumulation of associated, nonsensory information refines person representation, and finally, one reaches a stage where the effortless identification of very well-known persons occurs. We offer here an overview of neuroimaging studies, first evaluating how and in what ways the processing of unfamiliar and familiar faces differs and, second, by analyzing the fMRI adaptation and multivariate pattern analysis results we estimate where identity-specific representation is found in the brain. The available neuroimaging data suggest that different aspects of the information emerge gradually as one gets more and more familiar with a person within the same network. We propose a novel model of familiarity and identity processing, where the differential activation of long-term memory and emotion processing areas is essential for correct identification.

Surgical face masks impair human face matching performance for familiar and unfamiliar faces

In response to the COVID-19 pandemic, many governments around the world now recommend, or require, that their citizens cover the lower half of their face in public. Consequently, many people now wear surgical face masks in public. We investigated whether surgical face masks affected the performance of human observers, and a state-of-the-art face recognition system, on tasks of perceptual face matching. Participants judged whether two simultaneously presented face photographs showed the same person or two different people. We superimposed images of surgical masks over the faces, creating three different mask conditions: control (no masks), mixed (one face wearing a mask), and masked (both faces wearing masks). We found that surgical face masks have a large detrimental effect on human face matching performance, and that the degree of impairment is the same regardless of whether one or both faces in each pair are masked. Surprisingly, this impairment is similar in size for both familiar and unfamiliar faces. When matching masked faces, human observers are biased to reject unfamiliar faces as "mismatches" and to accept familiar faces as "matches". Finally, the face recognition system showed very high classification accuracy for control and masked stimuli, even though it had not been trained to recognise masked faces. However, accuracy fell markedly when one face was masked and the other was not. Our findings demonstrate that surgical face masks impair the ability of humans, and naïve face recognition systems, to perform perceptual face matching tasks. Identification decisions for masked faces should be treated with caution.

From concepts to percepts in human and machine face recognition: A reply to Blauch, Behrmann & Plaut

Intact recognition of familiar faces is critical for appropriate social interactions. Thus, the human face processing system should be optimized for familiar face recognition. Blauch et al. (2020) used face recognition deep convolutional neural networks (DCNNs) that are trained to maximize recognition of the trained (familiar) identities, to model human unfamiliar and familiar face recognition. In line with this model, we discuss behavioral, neuroimaging and computational findings that indicate that human face recognition develops from the generation of identity-specific concepts of familiar faces that are learned in a supervised manner, to the generation of view-invariant identity-general perceptual representations. Face-trained DCNNs seem to share some fundamental similarities with this framework.

Face Familiarity and Image-Specific Memory

Face familiarity produces advantages for both memory and matching. By developing an internal representation through repeated experience, viewers extract identity-specific information that aids subsequent recognition. However, researchers have recently argued that this process may also result in a familiarity disadvantage, whereby specific instances of the face are more difficult to remember, perhaps due to this process of prioritising identity- over image-specific information. Although previous experiments found no evidence of this disadvantage in working memory, initial research has demonstrated an effect in longer term storage. Here, we attempted to replicate this finding by focussing on the ability to learn images of a single (un)familiar identity. Our results failed to demonstrate a familiarity disadvantage while replicating the finding that familiarity influences response bias. As researchers continue to investigate how familiarity alters both internal representations and associated processes, it is important to establish which processes may or may not be affected.