Face-Space: A Unifying Concept in Face Recognition Research

The Dimensions of dimensionality

Cognitive scientists often infer multidimensional representations from data. Whether the data involve text, neuroimaging, neural networks, or human judgments, researchers frequently infer and analyze latent representational spaces (i.e., embeddings). However, the properties of a latent representation (e.g., prediction performance, interpretability, compactness) depend on the inference procedure, which can vary widely across endeavors. For example, dimensions are not always globally interpretable and the dimensionality of different embeddings may not be readily comparable. Moreover, the dichotomy between multidimensional spaces and purportedly richer representational formats, such as graph representations, is misleading. We review what the different notions of dimension in cognitive science imply for how these latent representations should be used and interpreted.

The neuropsychological evaluation of face identity recognition

Facial identity recognition (FIR) is arguably the ultimate form of recognition for the adult human brain. Even if the term prosopagnosia is reserved for exceptionally rare brain-damaged cases with a category-specific abrupt loss of FIR at adulthood, subjective and objective impairments or difficulties of FIR are common in the neuropsychological population. Here we provide a critical overview of the evaluation of FIR both for clinicians and researchers in neuropsychology. FIR impairments occur following many causes that should be identified objectively by both general and specific, behavioral and neural examinations. We refute the commonly used dissociation between perceptual and memory deficits/tests for FIR, since even a task involving the discrimination of unfamiliar face images presented side-by-side relies on cortical memories of faces in the right-lateralized ventral occipito-temporal cortex. Another frequently encountered confusion is between specific deficits of the FIR function and a more general impairment of semantic memory (of people), the latter being most often encountered following anterior temporal lobe damage. Many computerized tests aimed at evaluating FIR have appeared over the last two decades, as reviewed here. However, despite undeniable strengths, they often suffer from ecological limitations, difficulties of instruction, as well as a lack of consideration for processing speed and qualitative information. Taking into account these issues, a recently developed behavioral test with natural images manipulating face familiarity, stimulus inversion, and correct response times as a key variable appears promising. The measurement of electroencephalographic (EEG) activity in the frequency domain from fast periodic visual stimulation also appears as a particularly promising tool to complete and enhance the neuropsychological assessment of FIR.

Investigating the neural effects of typicality and predictability for face and object stimuli

The brain calibrates itself based on the past stimulus diet, which makes frequently observed stimuli appear as typical (as opposed to uncommon stimuli, which appear as distinctive). Based on predictive processing theory, the brain should be more "prepared" for typical exemplars, because these contain information that has been encountered frequently, allowing it to economically represent items of that category. Thus, one could ask whether predictability and typicality of visual stimuli interact, or rather act in an additive manner. We adapted the design by Egner and colleagues (2010), who used cues to induce expectations about stimulus category (face vs. chair) occurrence during an orthogonal inversion detection task. We measured BOLD responses with fMRI in 35 participants. First, distinctive stimuli always elicited stronger responses than typical ones in all ROIs, and our whole-brain directional contrasts for the effects of typicality and distinctiveness converge with previous findings. Second and importantly, we could not replicate the interaction between category and predictability reported by Egner et al. (2010), which casts doubt on whether cueing designs are ideal to elicit reliable predictability effects. Third, likely as a consequence of the lack of predictability effects, we found no interaction between predictability and typicality in any of the four tested regions (bilateral fusiform face areas, lateral occipital complexes) when considering both categories, nor in the whole brain. We discuss the issue of replicability in neuroscience and sketch an agenda for how future studies might address the same question.

Designing for sensory adaptation: what you see depends on what you've been looking at - Recommendations, guidelines and standards should reflect this

Sensory systems continuously recalibrate their responses according to the current stimulus environment. As a result, perception is strongly affected by the current and recent context. These adaptative changes affect both sensitivity (e.g., habituating to noise, seeing better in the dark) and appearance (e.g. how things look, what catches attention) and adjust to many perceptual properties (e.g. from light level to the characteristics of someone's face). They therefore have a profound effect on most perceptual experiences, and on how and how well the senses work in different settings. Characterizing the properties of adaptation, how it manifests, and when it influences perception in modern environments can provide insights into the diversity of human experience. Adaptation could also be leveraged both to optimize perceptual abilities (e.g. in visual inspection tasks like radiology) and to mitigate unwanted consequences (e.g. exposure to potentially unhealthy stimulus environments).

Perceptual Observer Modeling Reveals Likely Mechanisms of Face Expression Recognition Deficits in Depression

BACKGROUND

Deficits in face emotion recognition are well documented in depression, but the underlying mechanisms are poorly understood. Psychophysical observer models provide a way to precisely characterize such mechanisms. Using model-based analyses, we tested 2 hypotheses about how depression may reduce sensitivity to detect face emotion: 1) via a change in selectivity for visual information diagnostic of emotion or 2) via a change in signal-to-noise ratio in the system performing emotion detection.

METHODS

Sixty adults, one half meeting criteria for major depressive disorder and the other half healthy control participants, identified sadness and happiness in noisy face stimuli, and their responses were used to estimate templates encoding the visual information used for emotion identification. We analyzed these templates using traditional and model-based analyses; in the latter, the match between templates and stimuli, representing sensory evidence for the information encoded in the template, was compared against behavioral data.

RESULTS

Estimated happiness templates produced sensory evidence that was less strongly correlated with response times in participants with depression than in control participants, suggesting that depression was associated with a reduced signal-to-noise ratio in the detection of happiness. The opposite results were found for the detection of sadness. We found little evidence that depression was accompanied by changes in selectivity (i.e., information used to detect emotion), but depression was associated with a stronger influence of face identity on selectivity.

CONCLUSIONS

Depression is more strongly associated with changes in signal-to-noise ratio during emotion recognition, suggesting that deficits in emotion detection are driven primarily by deprecated signal quality rather than suboptimal sampling of information used to detect emotion.

Differential effects of intrinsic properties of natural scenes and interference mechanisms on recognition processes in long-term visual memory

Humans display remarkable long-term visual memory (LTVM) processes. Even though images may be intrinsically memorable, the fidelity of their visual representations, and consequently the likelihood of successfully retrieving them, hinges on their similarity when concurrently held in LTVM. In this debate, it is still unclear whether intrinsic features of images (perceptual and semantic) may be mediated by mechanisms of interference generated at encoding, or during retrieval, and how these factors impinge on recognition processes. In the current study, participants (32) studied a stream of 120 natural scenes from 8 semantic categories, which varied in frequencies (4, 8, 16 or 32 exemplars per category) to generate different levels of category interference, in preparation for a recognition test. Then they were asked to indicate which of two images, presented side by side (i.e. two-alternative forced-choice), they remembered. The two images belonged to the same semantic category but varied in their perceptual similarity (similar or dissimilar). Participants also expressed their confidence (sure/not sure) about their recognition response, enabling us to tap into their metacognitive efficacy (meta-d'). Additionally, we extracted the activation of perceptual and semantic features in images (i.e. their informational richness) through deep neural network modelling and examined their impact on recognition processes. Corroborating previous literature, we found that category interference and perceptual similarity negatively impact recognition processes, as well as response times and metacognitive efficacy. Moreover, images semantically rich were less likely remembered, an effect that trumped a positive memorability boost coming from perceptual information. Critically, we did not observe any significant interaction between intrinsic features of images and interference generated either at encoding or during retrieval. All in all, our study calls for a more integrative understanding of the representational dynamics during encoding and recognition enabling us to form, maintain and access visual information.

Event-related potential correlates of implicit processing of own- and other-race faces in children

Human adults typically experience difficulties in recognizing and discriminating individual faces belonging to racial groups other than their own. The origin of this "other-race" effect is set in infancy, but the understanding of its developmental course is fragmented. We aimed to access the mechanisms of the other-race effect in childhood by unraveling the neural time course of own- and other-race face processing during a masked priming paradigm. White 6- and 7-year-old children (N = 19) categorized fully visible Asian (other-race) or White (own-race) target faces according to gender. Target faces were preceded by masked same-identity or different-identity prime faces, matching the target for race and gender. We showed an early priming effect on the N100 component, with larger amplitude to different-face pairs than to same-face pairs, and a later race effect on the N200 component, with larger amplitude to own-race face pairs than to other-race face pairs. Critically, race did not interact with priming at any processing stage (P100, N100, P200, N200, or P300). Our results suggest that race could have a temporally limited impact on face processing and that the implicit and unconscious identity processing of own- and other-race faces could be similar in 6- and 7-year-olds, depicting an immature other-race effect during childhood.

Information-theoretical analysis of the neural code for decoupled face representation

Processing faces accurately and efficiently is a key capability of humans and other animals that engage in sophisticated social tasks. Recent studies reported a decoupled coding for faces in the primate inferotemporal cortex, with two separate neural populations coding for the geometric position of (texture-free) facial landmarks and for the image texture at fixed landmark positions, respectively. Here, we formally assess the efficiency of this decoupled coding by appealing to the information-theoretic notion of description length, which quantifies the amount of information that is saved when encoding novel facial images, with a given precision. We show that despite decoupled coding describes the facial images in terms of two sets of principal components (of landmark shape and image texture), it is more efficient (i.e., yields more information compression) than the encoding in terms of the image principal components only, which corresponds to the widely used eigenface method. The advantage of decoupled coding over eigenface coding increases with image resolution and is especially prominent when coding variants of training set images that only differ in facial expressions. Moreover, we demonstrate that decoupled coding entails better performance in three different tasks: the representation of facial images, the (daydream) sampling of novel facial images, and the recognition of facial identities and gender. In summary, our study provides a first principle perspective on the efficiency and accuracy of the decoupled coding of facial stimuli reported in the primate inferotemporal cortex.

Race and statistics in facial recognition: Producing types, physical attributes, and genealogies

Principal component analysis (PCA) is a common statistical procedure. In forensics, it is used in facial recognition technologies and composite sketching systems. PCA is especially helpful in contexts with high facial diversity, which is often translated as racial diversity. In these settings, researchers use PCA to define a 'normal face' and organize the rest of the available facial diversity based on their resemblance to or difference from that norm. In this way, the use of PCA introduces an 'ontology of the normal' in which expectations about how a normal face should look are corroborated by statistical calculations of normality. I argue that the use of PCA can lead to a statistical reification of racial stereotypes that informs recognition practices. I discuss current and historical cases in which PCA is used: one of face perception theorization ('face space theory') and two of technology development (the 'eigenfaces' facial recognition algorithm and the 'EvoFIT' composite sketching system). In each, PCA aligns facial normality with racial expectations, and instrumentalizes race in specific ways: as a type, physical attribute, or genealogy. This analysis of PCA does two things. First, it opens the black box of facial recognition to uncover how stereotypes and intuitions about normality become part of theories and technologies of facial recognition. Second, it explains why racial categorizations remain central in contemporary identification technologies and other forensic practices.

AI Hyperrealism: Why AI Faces Are Perceived as More Real Than Human Ones

Recent evidence shows that AI-generated faces are now indistinguishable from human faces. However, algorithms are trained disproportionately on White faces, and thus White AI faces may appear especially realistic. In Experiment 1 (N = 124 adults), alongside our reanalysis of previously published data, we showed that White AI faces are judged as human more often than actual human faces-a phenomenon we term AI hyperrealism. Paradoxically, people who made the most errors in this task were the most confident (a Dunning-Kruger effect). In Experiment 2 (N = 610 adults), we used face-space theory and participant qualitative reports to identify key facial attributes that distinguish AI from human faces but were misinterpreted by participants, leading to AI hyperrealism. However, the attributes permitted high accuracy using machine learning. These findings illustrate how psychological theory can inform understanding of AI outputs and provide direction for debiasing AI algorithms, thereby promoting the ethical use of AI.

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.

Serial dependence in facial identity perception and visual working memory

Serial dependence (SD) refers to the effect in which a person's current perceptual judgment is attracted toward recent stimulus history. Perceptual and memory processes, as well as response and decisional biases, are thought to contribute to SD effects. The current study examined the processing stages of SD facial identity effects in the context of task-related decision processes and how such effects may differ from visual working memory (VWM) interactions. In two experiments, participants were shown a series of two sequentially presented face images. In Experiment 1, the two faces were separated by an interstimulus interval (ISI) of 1, 3, 6, or 10 s, and participants were instructed to reproduce the second face after a varying response delay of 0, 1, 3, 6, or 10 s. Results showed that SD effects occurred most consistently at ISI of 1 s and response delays of 1 and 6 s consistent with early and late stages of processing. In Experiment 2, the ISI was held constant at 1 s, and to separate SD from VWM interactions participants were post-cued to reproduce either the first or the second face. When the second face was the target, SD effects again occurred at response delays of 1 and 6 s, but not when the first face was the target. Together, the results demonstrates that SD facial identity effects occur independently of task-related processes in a distinct temporal fashion and suggest that SD and VWM interactions may rely on separate underlying mechanisms.

Explicit face memory abilities are positively related to the non-intentional encoding of faces: Behavioral and ERP evidence

Individual differences in face memory abilities have been shown to be related to individual differences in brain activity. The present study investigated brain-behavior relationships for the N250 component in event-related brain potentials, which is taken as a neural sign of face familiarity. We used a task in which a designated, typical target face and several (high- and low-distinctive) nontarget faces had to be distinguished during multiple presentations across a session. Separately, face memory/recognition abilities were measured with easy versus difficult tasks. We replicated an increase of the N250 amplitude to the target face across the session and observed a similar increase for the non-target faces, indicating the build-up of memory representations also for these faces. On the interindividual level, larger across-session N250 amplitude increases to low-distinctive non-target faces were related to faster face recognition as measured in an easy task. These findings indicate that non-intentional encoding of non-target faces into memory is associated with the swift recognition of explicitly learned faces; that is, there is shared variance of incidental and intentional face memory.

Varying sex and identity of faces affects face categorization differently in humans and computational models

Our faces display socially important sex and identity information. How perceptually independent are these facial characteristics? Here, we used a sex categorization task to investigate how changing faces in terms of either their sex or identity affects sex categorization of those faces, whether these manipulations affect sex categorization similarly when the original faces were personally familiar or unknown, and, whether computational models trained for sex classification respond similarly to human observers. Our results show that varying faces along either sex or identity dimension affects their sex categorization. When the sex was swapped (e.g., female faces became male looking, Experiment 1), sex categorization performance was different from that with the original unchanged faces, and significantly more so for people who were familiar with the original faces than those who were not. When the identity of the faces was manipulated by caricaturing or anti-caricaturing them (these manipulations either augment or diminish idiosyncratic facial information, Experiment 2), sex categorization performance to caricatured, original, and anti-caricatured faces increased in that order, independently of face familiarity. Moreover, our face manipulations showed different effects upon computational models trained for sex classification and elicited different patterns of responses in humans and computational models. These results not only support the notion that the sex and identity of faces are processed integratively by human observers but also demonstrate that computational models of face categorization may not capture key characteristics of human face categorization.

Building 3D Generative Models from Minimal Data

We propose a method for constructing generative models of 3D objects from a single 3D mesh and improving them through unsupervised low-shot learning from 2D images. Our method produces a 3D morphable model that represents shape and albedo in terms of Gaussian processes. Whereas previous approaches have typically built 3D morphable models from multiple high-quality 3D scans through principal component analysis, we build 3D morphable models from a single scan or template. As we demonstrate in the face domain, these models can be used to infer 3D reconstructions from 2D data (inverse graphics) or 3D data (registration). Specifically, we show that our approach can be used to perform face recognition using only a single 3D template (one scan total, not one per person). We extend our model to a preliminary unsupervised learning framework that enables the learning of the distribution of 3D faces using one 3D template and a small number of 2D images. Our approach is motivated as a potential model for the origins of face perception in human infants, who appear to start with an innate face template and subsequently develop a flexible system for perceiving the 3D structure of any novel face from experience with only 2D images of a relatively small number of familiar faces.

Similarity-based reasoning in conceptual spaces

Whereas the validity of deductive inferences can be characterized in terms of their logical form, this is not true for all inferences that appear pre-theoretically valid. Nonetheless, philosophers have argued that at least some of those inferences-sometimes called "similarity-based inferences" -can be given a formal treatment with the help of similarity spaces, which are mathematical spaces purporting to represent human similarity judgments. In these inferences, we conclude that a given property pertains to a category of items on the grounds that the same property pertains to a similar category of items. We look at a specific proposal according to which the strength of such inferences is a function of the distance, as measured in the appropriate similarity space, between the category referenced in the premise and the category referenced in the conclusion. We report the outcomes of three studies that all support the said proposal.

Differences between high and low performers in face recognition in electrophysiological correlates of face familiarity and distance-to-norm

Valentine's influential norm-based multidimensional face-space model (nMDFS) predicts that perceived distinctiveness of a face increases with its distance to the norm. Occipito-temporal event-related potentials (ERPs) have been recently shown to respond selectively to variations in distance-to-norm (P200) or familiarity (N250, late negativity), respectively (Wuttke & Schweinberger, 2019). Despite growing evidence on interindividual differences in face perception skills at the behavioral level, little research has focused on their electrophysiological correlates. To reveal potential interindividual differences in face spaces, we contrasted high and low performers in face recognition in regards to distance-to-norm (P200) and familiarity (N250). We replicated both the P200 distance-to-norm and the N250 familiarity effect. Importantly, we observed: i) reduced responses in low compared to high performers of face recognition, especially in terms of smaller distance-to-norm effects in the P200, possibly indicating less 'expanded' face spaces in low compared to high performers; ii) increased N250 responses to familiar original faces in high performers, suggesting more robust face identity representations. In summary, these findings suggest the contribution of both early norm-based face coding and robust face representations to individual face recognition skills, and indicate that ERPs can offer a promising route to understand individual differences in face perception and their neurocognitive correlates.

A model for person perception from familiar and unfamiliar voices

When hearing a voice, listeners can form a detailed impression of the person behind the voice. Existing models of voice processing focus primarily on one aspect of person perception - identity recognition from familiar voices - but do not account for the perception of other person characteristics (e.g., sex, age, personality traits). Here, we present a broader perspective, proposing that listeners have a common perceptual goal of perceiving who they are hearing, whether the voice is familiar or unfamiliar. We outline and discuss a model - the Person Perception from Voices (PPV) model - that achieves this goal via a common mechanism of recognising a familiar person, persona, or set of speaker characteristics. Our PPV model aims to provide a more comprehensive account of how listeners perceive the person they are listening to, using an approach that incorporates and builds on aspects of the hierarchical frameworks and prototype-based mechanisms proposed within existing models of voice identity recognition.

FamFac - A Database of Famous Faces for Psychology Experiments

Introduction

High variation in the low-level proprieties of visual stimuli and varying degrees of familiarity with famous faces may have caused a bias in the results of investigations that tried to disentangle the processes involved in familiar and unfamiliar face processing (e.g., temporal differences in the detection of the first event-related potentials specialized in face processing may have been caused by different methods of controlling variance in the low-level proprieties of visual stimuli).

Objective

To address these problems, we developed a freely available database of 183 famous faces whose low-level proprieties (brightness, size, resolution) have been homogenized and the level of familiarity established.

Method

The brightness of the stimuli was standardized by a custom-developed algorithm. The size and the resolution of the pictures were homogenized in Gimp. The familiarity level of the famous faces was established by a group of 48 Portuguese college students.

Results

Our results suggest that the brightness of each image did not differ significantly from the mean brightness value of the stimuli set, confirming the standardizing ability of the algorithm. Forty-one famous faces were classified as highly familiar.

Main findings and implications

This study provides two important resources, as both the algorithm and the database are freely available for research purposes. The homogenization of the low-level features and the control of the level of familiarity of the famous faces included in our database should ensure that they do not elicit confounding effects such as the ones verified in past studies.

Stable individual differences in unfamiliar face identification: Evidence from simultaneous and sequential matching tasks

Matching identity in images of unfamiliar faces is difficult: Images of the same person can look different and images of different people can look similar. Recent studies have capitalized on individual differences in the ability to distinguish match (same ID) vs. mismatch (different IDs) face pairs to inform models of face recognition. We addressed two significant gaps in the literature by examining the stability of individual differences in both sensitivity to identity and response bias. In Study 1, 210 participants completed a battery of four tasks in each of two sessions separated by one week. Tasks varied in protocol (same/different, lineup, sorting) and stimulus characteristics (low vs. high within-person variability in appearance). In Study 2, 148 participants completed a battery of three tasks in a single session. Stimuli were presented simultaneously on some trials and sequentially on others, introducing short-term memory demands. Principal components analysis revealed two components that were stable across time and tasks: sensitivity to identity and bias. Analyses of response times suggest that individual differences in bias reflect decision-making processes. We discuss the implications of our findings in applied settings and for models of face recognition.

Intracerebral Electrophysiological Recordings to Understand the Neural Basis of Human Face Recognition

Understanding how the human brain recognizes faces is a primary scientific goal in cognitive neuroscience. Given the limitations of the monkey model of human face recognition, a key approach in this endeavor is the recording of electrophysiological activity with electrodes implanted inside the brain of human epileptic patients. However, this approach faces a number of challenges that must be overcome for meaningful scientific knowledge to emerge. Here we synthesize a 10 year research program combining the recording of intracerebral activity (StereoElectroEncephaloGraphy, SEEG) in the ventral occipito-temporal cortex (VOTC) of large samples of participants and fast periodic visual stimulation (FPVS), to objectively define, quantify, and characterize the neural basis of human face recognition. These large-scale studies reconcile the wide distribution of neural face recognition activity with its (right) hemispheric and regional specialization and extend face-selectivity to anterior regions of the VOTC, including the ventral anterior temporal lobe (VATL) typically affected by magnetic susceptibility artifacts in functional magnetic resonance imaging (fMRI). Clear spatial dissociations in category-selectivity between faces and other meaningful stimuli such as landmarks (houses, medial VOTC regions) or written words (left lateralized VOTC) are found, confirming and extending neuroimaging observations while supporting the validity of the clinical population tested to inform about normal brain function. The recognition of face identity - arguably the ultimate form of recognition for the human brain - beyond mere differences in physical features is essentially supported by selective populations of neurons in the right inferior occipital gyrus and the lateral portion of the middle and anterior fusiform gyrus. In addition, low-frequency and high-frequency broadband iEEG signals of face recognition appear to be largely concordant in the human association cortex. We conclude by outlining the challenges of this research program to understand the neural basis of human face recognition in the next 10 years.

fMRI evidence that hyper-caricatured faces activate object-selective cortex

Many brain imaging studies have looked at the cortical responses to object categories and faces. A popular way to manipulate face stimuli is by using a "face space," a high dimensional representation of individual face images, with the average face located at the origin. However, how the brain responds to faces that deviate substantially from average has not been much explored. Increasing the distance from the average (leading to increased caricaturing) could increase neural responses in face-selective regions, an idea supported by results from non-human primates. Here, we used a face space based on principal component analysis (PCA) to generate faces ranging from average to heavily caricatured. Using functional magnetic resonance imaging (fMRI), we first independently defined face-, object- and scene-selective areas with a localiser scan and then measured responses to parametrically caricatured faces. We also included conditions in which the images of faces were inverted. Interestingly in the right fusiform face area (FFA), we found that the patterns of fMRI response were more consistent as caricaturing increased. However, we found no consistent effect of either caricature level or facial inversion on the average fMRI response in the FFA or face-selective regions more broadly. In contrast, object-selective regions showed an increase in both the consistency of response pattern and the average fMRI response with increasing caricature level. This shows that caricatured faces recruit processing from regions typically defined as object-selective, possibly through enhancing low-level properties that are characteristic of objects.

The other-race effect in facial expression processing: Behavioral and ERP evidence from a balanced cross-cultural study in women

Although evidence for cultural variants in facial expression decoding is accumulating, the other-race effect in facial expression processing and its neural correlates are still unclear. We investigated this question with a fully balanced design, in which a group of East Asian and a group of European Caucasian women categorized pictures of sad, happy, angry, and neutral facial expressions posed by individuals of their own-race and the other-race. Results revealed a disadvantage in categorizing expressions of anger in other-race faces in both samples, and for sad expressions in the European sample only. Partially consistent, East Asian participants showed longer latency of the N170 component in the event-related potential (ERP) and European Caucasian participants showed larger N170 amplitudes to other-race faces. The late positive complex in the ERP was less distinguishable among other-race facial expressions. Therefore, the present study observed an other-race effect in early and late stages of face processing, reflecting less efficient structural encoding and less elaborate processing for other-race than own-race faces.

Understanding the mechanisms underlying the other-'race' effect: An attempt at integrating different perspectives

Although different human races do not exist from the perspective of biology and genetics, ascribed 'race' influences psychological processing, such as memory and perception of faces. Research from this Special Issue, as well as a wealth of previous research, shows that other-'race' faces are more difficult to recognize compared to own-'race' faces, a phenomenon known as the other-'race' effect. Theories of expertise attribute the cause of the other-'race' effect to less efficient visual representations of other-'race' faces, which results from reduced visual expertise with other-'race' faces compared to own-'race' faces due to limited contact with individuals from other 'racial' groups. By contrast, social-cognitive accounts attribute the cause of the other-'race' effect to reduced motivation to individuate other-'race' faces compared to own-'race' faces. Evidence for both types of theories is still mixed, but progress in understanding the phenomenon has also been hampered by the fact that there has been little crosstalk between these accounts, which tend to be rooted in separate domains of experimental perception science and social psychology, respectively. To promote an integrative perspective on current knowledge on own- versus other-'race' face processing, the present Special Issue bridges different psychological subdisciplines, showcasing research using a large variety of methodological approaches and measures. In this guest editorial, we briefly highlight individual contributions to this Special Issue and offer what we see as important avenues for future research on the other-'race' effect.

Event-related brain potential correlates of the other-race effect: A review

People are better at remembering own-race relative to other-race faces. Here, we review event-related brain potential (ERP) correlates of this so-called other-'race' effect (ORE) by discussing three critical aspects that characterize the neural signature of this phenomenon. First, difficulties with other-race faces initially emerge during perceptual processing, which is indexed by an increased N170. Second, as evidenced by 'difference due to subsequent memory' effects, more effortful processing of other-race faces is needed for successful encoding into long-term memory. Third, ERP old/new effects reveal that a stronger engagement of processing resources is also required for successful retrieval of other-race faces from memory. The ERP evidence available to date thus suggests widespread ethnicity-related modulations during both perceptual and mnemonic processing stages. We further discuss how findings from the ORE compared with potentially related memory biases (e.g. other-gender or other-age effects) and how ERP findings inform the ongoing debate regarding the mechanisms underlying the ORE. Finally, we outline open questions and potential future directions with an emphasis on using multiple, ecologically more valid 'ambient' images for each face to assess the ORE in paradigms that capture identity rather than image recognition.

Impact of similarity on recognition of faces of Black and White targets

One reason for the persistence of racial inequality may be anticipated dissimilarity with racial outgroups. In the present research, we explored the impact of perceived similarity with White and Black targets on facial identity recognition accuracy. In two studies, participants first completed an ostensible personality survey. Next, in a Learning Phase, Black and White faces were presented on one of three background colours. Participants were led to believe that these colours indicated similarities between them and the target person in the image. Specifically, they were informed that the background colours were associated with the extent to which responses by the target person on the personality survey and their own responses overlapped. In actual fact, faces were randomly assigned to colour. In both studies, non-Black participants (Experiment 1) and White participants (Experiment 2) showed better recognition of White than Black faces. More importantly in the present context, a positive linear effect of similarity was found in both studies, with better recognition of increasingly similar Black and White targets. The independent effects for race of target and similarity, with no interaction, indicated that participants responded to Black and White faces according to category membership as well as on an interpersonal level related to similarity with specific targets. Together these findings suggest that while perceived similarity may enhance identity recognition accuracy for Black and White faces, it may not reduce differences in facial memory for these racial categories.

Ingroup and outgroup differences in face detection

Humans show improved recognition for faces from their own social group relative to faces from another social group. Yet before faces can be recognized, they must first be detected in the visual field. Here, we tested whether humans also show an ingroup bias at the earliest stage of face processing - the point at which the presence of a face is first detected. To this end, we measured viewers' ability to detect ingroup (Black and White) and outgroup faces (Asian, Black, and White) in everyday scenes. Ingroup faces were detected with greater speed and accuracy relative to outgroup faces (Experiment 1). Removing face hue impaired detection generally, but the ingroup detection advantage was undiminished (Experiment 2). This same pattern was replicated by a detection algorithm using face templates derived from human data (Experiment 3). These findings demonstrate that the established ingroup bias in face processing can extend to the early process of detection. This effect is 'colour blind', in the sense that group membership effects are independent of general effects of image hue. Moreover, it can be captured by tuning visual templates to reflect the statistics of observers' social experience. We conclude that group bias in face detection is both a visual and a social phenomenon.

Asymmetrical responding to male versus female other-race categories in 9- to 12-month-old infants

Faces can be categorized along various dimensions including gender or race, an ability developing in infancy. Infant categorization studies have focused on facial attributes in isolation, but the interaction between these attributes remains poorly understood. Experiment 1 examined gender categorization of other-race faces in 9- and 12-month-old White infants. Nine- and 12-month-olds were familiarized with Asian male or female faces, and tested with a novel exemplar from the familiarized category paired with a novel exemplar from a novel category. Both age groups showed novel category preferences for novel Asian female faces after familiarization with Asian male faces, but showed no novel category preference for novel Asian male faces after familiarization with Asian female faces. This categorization asymmetry was not due to a spontaneous preference hindering novel category reaction (Experiment 2), and both age groups displayed difficulty discriminating among male, but not female, other-race faces (Experiment 3). These results indicate that category formation for male other-race faces is mediated by categorical perception. Overall, the findings suggest that even by 12 months of age, infants are not fully able to form gender category representations of other-race faces, responding categorically to male, but not female, other-race faces.

Face dissimilarity judgments are predicted by representational distance in morphable and image-computable models

Discerning the subtle differences between individuals’ faces is crucial for social functioning. It requires us not only to solve general challenges of object recognition (e.g., invariant recognition over changes in view or lighting) but also to be attuned to the specific ways in which face structure varies. Three-dimensional morphable models based on principal component analyses of real faces provide descriptions of statistical differences between faces, as well as tools to generate novel faces. We rendered large sets of realistic face pairs from such a model and collected similarity and same/different identity judgments. The statistical model predicted human perception as well as state-of-the-art image-computable neural networks. Results underscore the statistical tuning of face encoding.

Human vision is attuned to the subtle differences between individual faces. Yet we lack a quantitative way of predicting how similar two face images look and whether they appear to show the same person. Principal component–based three-dimensional (3D) morphable models are widely used to generate stimuli in face perception research. These models capture the distribution of real human faces in terms of dimensions of physical shape and texture. How well does a “face space” based on these dimensions capture the similarity relationships humans perceive among faces? To answer this, we designed a behavioral task to collect dissimilarity and same/different identity judgments for 232 pairs of realistic faces. Stimuli sampled geometric relationships in a face space derived from principal components of 3D shape and texture (Basel face model [BFM]). We then compared a wide range of models in their ability to predict the data, including the BFM from which faces were generated, an active appearance model derived from face photographs, and image-computable models of visual perception. Euclidean distance in the BFM explained both dissimilarity and identity judgments surprisingly well. In a comparison against 16 diverse models, BFM distance was competitive with representational distances in state-of-the-art deep neural networks (DNNs), including novel DNNs trained on BFM synthetic identities or BFM latents. Models capturing the distribution of face shape and texture across individuals are not only useful tools for stimulus generation. They also capture important information about how faces are perceived, suggesting that human face representations are tuned to the statistical distribution of faces.

The cultural learning account of first impressions

Humans spontaneously attribute character traits to strangers based on their facial appearance. Although these 'first impressions' typically have no basis in reality, some authors have assumed that they have an innate origin. By contrast, the Trait Inference Mapping (TIM) account proposes that first impressions are products of culturally acquired associative mappings that allow activation to spread from representations of facial appearance to representations of trait profiles. According to TIM, cultural instruments, including propaganda, illustrated storybooks, art and iconography, ritual, film, and TV, expose many individuals within a community to common sources of correlated face-trait experience, yielding first impressions that are shared by many, but typically inaccurate. Here, we review emerging empirical findings, many of which accord with TIM, and argue that future work must distinguish first impressions based on invariant facial features (e.g., shape) from those based on facial behaviours (e.g., expressions).

Perceiving ingroup and outgroup faces within and across nations

The human face is arguably the most important of all social stimuli because it provides so much valuable information about others. Therefore, one critical factor for successful social communication is the ability to process faces. In general, a wide body of social cognitive research has demonstrated that perceivers are better at extracting information from their own-race compared to other-race faces and that these differences can be a barrier to positive cross-race relationships. The primary objective of the present paper was to provide an overview of how people process faces in diverse contexts, focusing on racial ingroup and outgroup members within one nation and across nations. To achieve this goal, we first broadly describe social cognitive research on categorization processes related to ingroups vs. outgroups. Next, we briefly examine two prominent mechanisms (experience and motivation) that have been used to explain differences in recognizing facial identities and identifying emotions when processing ingroup and outgroup racial faces within nations. Then, we explore research in this domain across nations and cultural explanations, such as norms and practices, that supplement the two proposed mechanisms. Finally, we propose future cross-cultural research that has the potential to help us better understand the role of these key mechanisms in processing ingroup and outgroup faces.

Caricatured facial movements enhance perception of emotional facial expressions

Although faces “in the wild” constantly undergo complicated movements, humans adeptly perceive facial identity and expression. Previous studies, focusing mainly on identity, used photographic caricature to show that distinctive form increases perceived dissimilarity. We tested whether distinctive facial movements showed similar effects, and we focussed on both perception of expression and identity. We caricatured the movements of an animated computer head, using physical motion metrics extracted from videos. We verified that these “ground truth” metrics showed the expected effects: Caricature increased physical dissimilarity between faces differing in expression and those differing in identity. Like the ground truth dissimilarity, participants’ dissimilarity perception was increased by caricature when faces differed in expression. We found these perceived dissimilarities to reflect the “representational geometry” of the ground truth. However, neither of these findings held for faces differing in identity. These findings replicated across two paradigms: pairwise ratings and multiarrangement. In a final study, motion caricature did not improve recognition memory for identity, whether manipulated at study or test. We report several forms of converging evidence for spatiotemporal caricature effects on dissimilarity perception of different expressions. However, more work needs to be done to discover what identity-specific movements can enhance face identification.

Familiarity, orientation, and realism increase face uncanniness by sensitizing to facial distortions

The uncanny valley predicts aversive reactions toward near-humanlike entities. Greater uncanniness is elicited by distortions in realistic than unrealistic faces, possibly due to familiarity. Experiment 1 investigated how familiarity and inversion affect uncanniness of facial distortions and the ability to detect differences between the distorted variants of the same face (distortion sensitivity). Familiar or unfamiliar celebrity faces were incrementally distorted and presented either upright or inverted. Uncanniness ratings increased across the distortion levels, and were stronger for familiar and upright faces. Distortion sensitivity increased with increasing distortion difference levels, again stronger for familiar and upright faces. Experiment 2 investigated how face realism, familiarity, and face orientation interacted for the increase of uncanniness across distortions. Realism increased the increase of uncanniness across the distortion levels, further enhanced by upright orientation and familiarity. The findings show that familiarity, upright orientation, and high face realism increase the sensitivity of uncanniness, likely by increasing distortion sensitivity. Finally, a moderated linear function of face realism and deviation level could explain the uncanniness of stimuli better than a quadratic function. A re-interpretation of the uncanny valley as sensitivity toward deviations from familiarized patterns is discussed.

Neurocognitive effects of a training program for poor face recognizers using shape and texture caricatures: A pilot investigation

Recent research suggested disproportional usage of shape information by people with poor face recognition, although texture information appears to be more important for familiar face recognition. Here, we tested a training program with faces that were selectively caricatured in either shape or texture parameters. Forty-eight young adults with poor face recognition skills (1 SD below the mean in at least 2/3 face processing tests: CFMT, GFMT, BFFT) were pseudo-randomly assigned to either one of two training groups or a control group (n = 16 each). Training comprised six sessions over three weeks. Per session, participants studied ten unfamiliar facial identities whose shape or texture characteristics were caricatured. Before and after training (or waiting in the control group), all participants completed EEG experiments on face learning and famous face recognition, and behavioral face processing tests. Results showed small but specific training-induced improvements: Whereas shape training improved face matching (training tasks, and to some extent GFMT), texture training elicited marked improvements in face learning (CFMT). Moreover, for the texture training group the N170 ERP was enhanced for novel faces post-training, suggesting training-induced changes in early markers of face processing. Although further research is necessary, this suggests that parameter-specific caricature training is a promising way to improve performance in people with poor face recognition skills.

I See Faces! A Review on Face Perception and Attractiveness with a Prosthodontic Peek at Cognitive Psychology

A human face contains a wealth of information about an individual, with which an observer can instinctively make a judgment on the attractiveness of the face. However, despite the profuse literature on facial and smile attractiveness, their origins, determinants, and perceptions remain controversial. The axiom in face processing research is that a face is perceived as an amalgamation of its features, and is referred to as "whole" or "holistic" perception. It is pertinent to the clinician involved in the provision of esthetic restorations to understand this holistic process of face recognition and perception of smile attractiveness. This review paper addresses face recognition and perception of attractiveness by reviewing the holistic perception of faces, including the multidimensional face-space model, and also reviews the smile and facial attractiveness according to the average, multiple motive, and secondary sex characteristics theories.

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.

Test a Witness's Memory of a Suspect Only Once

Eyewitness misidentifications are almost always made with high confidence in the courtroom. The courtroom is where eyewitnesses make their last identification of defendants suspected of (and charged with) committing a crime. But what did those same eyewitnesses do on the first identification test, conducted early in a police investigation? Despite testifying with high confidence in court, many eyewitnesses also testified that they had initially identified the suspect with low confidence or failed to identify the suspect at all. Presenting a lineup leaves the eyewitness with a memory trace of the faces in the lineup, including that of the suspect. As a result, the memory signal generated by the face of that suspect will be stronger on a later test involving the same witness, even if the suspect is innocent. In that sense, testing memory contaminates memory. These considerations underscore the importance of a newly proposed recommendation for conducting eyewitness identifications: Avoid repeated identification procedures with the same witness and suspect. This recommendation applies not only to additional tests conducted by police investigators but also to the final test conducted in the courtroom, in front of the judge and jury.

Symmetry in Emotional and Visual Similarity between Neutral and Negative Faces

Is Mr. Hyde more similar to his alter ego Dr. Jekyll, because of their physical identity, or to Jack the Ripper, because both evoke fear and loathing? The relative weight of emotional and visual dimensions in similarity judgements is still unclear. We expected an asymmetric effect of these dimensions on similarity perception, such that faces that express the same or similar feeling are judged as more similar than different emotional expressions of same person. We selected 10 male faces with different expressions. Each face posed one neutral expression and one emotional expression (five disgust, five fear). We paired these expressions, resulting in 190 pairs, varying either in emotional expressions, physical identity, or both. Twenty healthy participants rated the similarity of paired faces on a 7-point scale. We report a symmetric effect of emotional expression and identity on similarity judgements, suggesting that people may perceive Mr. Hyde to be just as similar to Dr. Jekyll (identity) as to Jack the Ripper (emotion). We also observed that emotional mismatch decreased perceived similarity, suggesting that emotions play a prominent role in similarity judgements. From an evolutionary perspective, poor discrimination between emotional stimuli might endanger the individual.

Face Adaptation-Investigating Nonconfigural Saturation Alterations

Recognizing familiar faces requires a comparison of the incoming perceptual information with mental face representations stored in memory. Mounting evidence indicates that these representations adapt quickly to recently perceived facial changes. This becomes apparent in face adaptation studies where exposure to a strongly manipulated face alters the perception of subsequent face stimuli: original, non-manipulated face images then appear to be manipulated, while images similar to the adaptor are perceived as "normal." The face adaptation paradigm serves as a good tool for investigating the information stored in facial memory. So far, most of the face adaptation studies focused on configural (second-order relationship) face information, mainly neglecting non-configural face information (i.e., that does not affect spatial face relations), such as color, although several (non-adaptation) studies were able to demonstrate the importance of color information in face perception and identification. The present study therefore focuses on adaptation effects on saturation color information and compares the results with previous findings on brightness. The study reveals differences in the effect pattern and robustness, indicating that adaptation effects vary considerably even within the same class of non-configural face information.

The effect of implicit racial bias on recognition of other-race faces

Previous research has established a possible link between recognition performance, individuation experience, and implicit racial bias of other-race faces. However, it remains unclear how implicit racial bias might influence other-race face processing in observers with relatively extensive experience with the other race. Here we examined how recognition of other-race faces might be modulated by observers’ implicit racial bias, in addition to the effects of experience and face recognition ability. Caucasian participants in a culturally diverse city completed a memory task for Asian and Caucasian faces, an implicit association test, a questionnaire assessing experience with Asians and Caucasians, and a face recognition ability test. As expected, recognition performance for Asian faces was positively predicted by increased face recognition ability, and experience with Asians. More importantly, it was also negatively predicted by increased positive bias towards Asians, which was modulated by an interaction between face recognition ability and implicit bias, with the effect of implicit bias observed predominantly in observers with high face recognition ability. Moreover, the positions of the first two fixations when participants learned the other-race faces were affected by different factors, with the first fixation modulated by the effect of experience and the second fixation modulated by the interaction between implicit bias and face recognition ability. Taken together, these findings suggest the complexity in understanding the perceptual and socio-cognitive influences on the other-race effect, and that observers with high face recognition ability may more likely evaluate racial features involuntarily when recognizing other-race faces.

Using genetic algorithms to uncover individual differences in how humans represent facial emotion

Emotional facial expressions critically impact social interactions and cognition. However, emotion research to date has generally relied on the assumption that people represent categorical emotions in the same way, using standardized stimulus sets and overlooking important individual differences. To resolve this problem, we developed and tested a task using genetic algorithms to derive assumption-free, participant-generated emotional expressions. One hundred and five participants generated a subjective representation of happy, angry, fearful and sad faces. Population-level consistency was observed for happy faces, but fearful and sad faces showed a high degree of variability. High test-retest reliability was observed across all emotions. A separate group of 108 individuals accurately identified happy and angry faces from the first study, while fearful and sad faces were commonly misidentified. These findings are an important first step towards understanding individual differences in emotion representation, with the potential to reconceptualize the way we study atypical emotion processing in future research.

The Importance of Internal and External Features in Matching Own and Other Race Faces

Research has shown that we are better at discriminating between faces that are our own race, and much less accurate with faces of another race. When the external features of faces were removed, this reduced the accuracy for recognizing other-races faces, more than own-race faces, suggesting that the external features (hair, face shape) are especially important for the recognition of other-race faces. The aim of the current study was to determine whether external features were more useful in matching other-race faces, and whether this was the case for Western and Eastern viewers. The current study employed a face matching task with Caucasian (U.K.) and Asian (Chinese) participants and found that responses were more accurate for own-race faces, and for whole faces when compared with faces where the internal or external features had been removed. Removing the external features of other-race faces increased the own-race bias for Chinese and U.K. participants, demonstrating the importance of viewing whole faces, including the external features when matching other-race faces.

Different measures of holistic face processing tap into distinct but partially overlapping mechanisms

Holistic processing, which includes the integration of facial features and analysis of their relations to one another, is a hallmark of what makes faces 'special'. Various experimental paradigms purport to measure holistic processing but these have often produced inconsistent results. This has led researchers to question the nature and structure of the mechanism(s) underlying holistic processing. Using an individual differences approach, researchers have examined relations between various measures of holistic processing in an attempt to resolve these questions. In keeping with this, we examined relationships between four commonly used measures of holistic face processing in a large group of participants (N = 223): (1) The Face Inversion Effect, (2) the Part Whole Effect (PWE), (3) the Composite Face Effect, and (4) the Configural Featural Detection Task (CFDT). Several novel methodological and analytical elements were introduced, including the use of factor analysis and the inclusion of control conditions to confirm the face specificity of all of the effects measured. The four indexes of holistic processing derived from each measure loaded onto two factors, one encompassing the PWE and the CFDT, and one encompassing the CE. The 16 conditions tested across the four tasks loaded onto four factors, each factor corresponding to a different measure. These results, together with those of other studies, suggest that holistic processing is a multifaceted construct and that different measures tap into distinct but partially overlapping elements of it.