The role of familiarity in three-dimensional view-transferability of face identity adaptation

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.

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).

How familiarity warps representation in the face space

Recognition of familiar as compared to unfamiliar faces is robust and resistant to marked image distortion or degradation. Here we tested the flexibility of familiar face recognition with a morphing paradigm where the appearance of a personally familiar face was mixed with the appearance of a stranger (Experiment 1) and the appearance of one's own face with the appearance of a familiar face and the appearance of a stranger (Experiment 2). The aim of the two experiments was to assess how categorical boundaries for recognition of identity are affected by familiarity. We found a narrower categorical boundary for the identity of personally familiar faces when they were mixed with unfamiliar identities as compared to the control condition, in which the appearance of two unfamiliar faces was mixed. Our results suggest that familiarity warps the representational geometry of face space, amplifying perceptual distances for small changes in the appearance of familiar faces that are inconsistent with the structural features that define their identities.

Size Effects in the Recognition of Blurry Faces

Spatial frequencies critical for recognition of faces are scale-dependent. Progressively coarser features of the face are utilized at smaller sizes, despite the availability of finer features. Blur removes fine details in an image, disrupting the finer features utilized for recognition at large sizes. At smaller sizes, observers utilize coarser features, and thus, recognition may be less impacted by blur. This coupling between size and critical spatial frequencies allows us to predict a regime in which observers tolerate blur better with decreasing image sizes within a range of moderate face sizes. We tested recognition of famous faces in four conditions: large-intact, small-intact, large-blurry, and small-blurry. Observers showed high recognition performance in both intact conditions. Blur significantly disrupted recognition, yet accuracy was significantly and consistently higher in the small-blurry compared to the large-blurry condition. These results are suggestive of an inherent scale-dependent mechanism that, at certain sizes, negatively impacts recognition of blurry images.

Contributions of shape and reflectance information to social judgments from faces

Face perception is based on both shape and reflectance information. However, we know little about the relative contribution of these kinds of information to social judgments of faces. In Experiment 1, we generated faces using validated computational models of attractiveness, competence, dominance, extroversion, and trustworthiness. Faces were manipulated orthogonally on five levels of shape and reflectance for each model. Both kinds of information had linear and additive effects on participants' social judgments. Shape information was more predictive of dominance, extroversion, and trustworthiness judgments, whereas reflectance information was more predictive of competence judgments. In Experiment 2, to test whether the amount of visual information alters the relative contribution of shape and reflectance information, we presented faces - varied on attractiveness, competence, and dominance - for five different durations (33-500 ms). For all judgments, the linear effect of both shape and reflectance increased as duration increased. Importantly, the relative contribution did not change across durations. These findings show that that the judged dimension is critical for which kind of information is weighted more heavily in judgments and that the relative contribution of shape and reflectance is stable across the amount of visual information available.

Face-Space: A Unifying Concept in Face Recognition Research

The concept of a multidimensional psychological space, in which faces can be represented according to their perceived properties, is fundamental to the modern theorist in face processing. Yet the idea was not clearly expressed until 1991. The background that led to the development of face-space is explained, and its continuing influence on theories of face processing is discussed. Research that has explored the properties of the face-space and sought to understand caricature, including facial adaptation paradigms, is reviewed. Face-space as a theoretical framework for understanding the effect of ethnicity and the development of face recognition is evaluated. Finally, two applications of face-space in the forensic setting are discussed. From initially being presented as a model to explain distinctiveness, inversion, and the effect of ethnicity, face-space has become a central pillar in many aspects of face processing. It is currently being developed to help us understand adaptation effects with faces. While being in principle a simple concept, face-space has shaped, and continues to shape, our understanding of face perception.

The Role of Familiarity on Viewpoint Adaptation for Self-Face and Other-Face Images

An adaptation method was used to investigate whether self-face processing is dissociable from general face processing. We explored the viewpoint aftereffect with face images having different degrees of familiarity (never-before-seen faces, recently familiarized faces, personally familiar faces, and the participant’s own face). A face viewpoint aftereffect occurs after prolonged viewing of a face viewed from one side, with the result that the perceived viewing direction of a subsequently presented face image shown near the frontal view is biased in a direction which is the opposite of the adapting orientation. We found that (1) the magnitude of the viewpoint aftereffect depends on the level of familiarity of the adapting and test faces, (2) a cross-identity transfer of the viewpoint aftereffect is found between all categories of faces, but not between an unfamiliar adaptor face and the self-face test, and (3) learning affects the processing of the self-face in greater measure than any other category of faces. These results highlight the importance of familiarity on the face aftereffects, but they also suggest the possibility of separate representations for the self-face, on the one side, and for highly familiar faces, on the other.

Effects of exposure to facial expression variation in face learning and recognition

Facial expression is a major source of image variation in face images. Linking numerous expressions to the same face can be a huge challenge for face learning and recognition. It remains largely unknown what level of exposure to this image variation is critical for expression-invariant face recognition. We examined this issue in a recognition memory task, where the number of facial expressions of each face being exposed during a training session was manipulated. Faces were either trained with multiple expressions or a single expression, and they were later tested in either the same or different expressions. We found that recognition performance after learning three emotional expressions had no improvement over learning a single emotional expression (Experiments 1 and 2). However, learning three emotional expressions improved recognition compared to learning a single neutral expression (Experiment 3). These findings reveal both the limitation and the benefit of multiple exposures to variations of emotional expression in achieving expression-invariant face recognition. The transfer of expression training to a new type of expression is likely to depend on a relatively extensive level of training and a certain degree of variation across the types of expressions.

Exemplar variance supports robust learning of facial identity

Differences in the visual processing of familiar and unfamiliar faces have prompted considerable interest in face learning, the process by which unfamiliar faces become familiar. Previous work indicates that face learning is determined in part by exposure duration; unsurprisingly, viewing faces for longer affords superior performance on subsequent recognition tests. However, there has been further speculation that exemplar variation, experience of different exemplars of the same facial identity, contributes to face learning independently of viewing time. Several leading accounts of face learning, including the averaging and pictorial coding models, predict an exemplar variation advantage. Nevertheless, the exemplar variation hypothesis currently lacks empirical support. The present study therefore sought to test this prediction by comparing the effects of unique exemplar face learning—a condition rich in exemplar variation—and repeated exemplar face learning—a condition that equates viewing time, but constrains exemplar variation. Crucially, observers who received unique exemplar learning displayed better recognition of novel exemplars of the learned identities at test, than observers in the repeated exemplar condition. These results have important theoretical and substantive implications for models of face learning and for approaches to face training in applied contexts.

Decision-dependent aftereffects for faces

Adaptation-related aftereffects (AEs) have been found in the perception of face identity, in that perception of an ambiguous face is typically biased away from the identity of a preceding unambiguous adaptor face. In previous studies, we could show that both perceptual ambiguity and physical similarity play a role in determining perceived face identity AEs, Cortex 49 (2013) 1963-1977, Plos One 8 (2013) e70525. Here, we tested further the role of ambiguity by manipulating participants' task such that the very same target stimuli were either ambiguous or unambiguous regarding stimulus classification. We created two partially overlapping continua spanning three unfamiliar face identities each, by morphing identity A via B to C, and B via C to D. In a first session, participants were familiarised with faces A and C and asked to classify faces of the A-B-C continuum as either identity A or C in an AE paradigm. Following adaptation to A or C, we observed contrastive AEs for the ambiguous identity B, but not for the unambiguous identities A or C. In a second session, the same participants were familiarised with faces B and D, followed by tests of AEs for the B-C-D continuum now involving a B-D classification task. We again observed contrastive AEs but only for target identity C (ambiguous for the decision) and not for B or D (unambiguous). Our results suggest that perceptual ambiguity, as given by the task-context, determines whether or not AEs are induced.

Repeated short presentations of morphed facial expressions change recognition and evaluation of facial expressions

This study investigated whether sensitivity to and evaluation of facial expressions varied with repeated exposure to non-prototypical facial expressions for a short presentation time. A morphed facial expression was presented for 500 ms repeatedly, and participants were required to indicate whether each facial expression was happy or angry. We manipulated the distribution of presentations of the morphed facial expressions for each facial stimulus. Some of the individuals depicted in the facial stimuli expressed anger frequently (i.e., anger-prone individuals), while the others expressed happiness frequently (i.e., happiness-prone individuals). After being exposed to the faces of anger-prone individuals, the participants became less sensitive to those individuals' angry faces. Further, after being exposed to the faces of happiness-prone individuals, the participants became less sensitive to those individuals' happy faces. We also found a relative increase in the social desirability of happiness-prone individuals after exposure to the facial stimuli.

View-adaptation reveals coding of face pose along image, not object, axes

High-level adaptation effects reveal important features of the neural coding of objects and faces. View-adaptation in particular is a highly useful means of characterizing how depth rotation of the face is represented and therefore, how view-invariant recognition of the face may be achieved. In the present study, we used view adaptation to determine the extent to which depth rotations of a face are represented in an image-based or object-based manner. Specifically, we dissociated object-based axes from image-based axes via a 90° planar rotation of the adapting face and observed that participants' responses pre- and post-adaptation are most consistent with an image-based representation of depth rotations of the face. We discuss our data in the context of previous results describing the impact of planar rotation on related aspects of face perception.

Face distortion aftereffects in personally familiar, famous, and unfamiliar faces

The internal face prototype is thought to be a construction of the average of every previously viewed face (Schwaninger et al., 2003). However, the influence of the most frequently encountered faces (i.e., personally familiar faces) has been generally understated. The current research explored the face distortion aftereffect in unfamiliar, famous, and personally familiar (each subject’s parent) faces. Forty-eight adult participants reported whether faces were distorted or not (distorted by shifting the eyes in the vertical axis) of a series of images that included unfamiliar, famous, and personally familiar faces. The number of faces perceived to be “odd” was measured pre- and post-adaptation to the most extreme distortion. Participants were adapted to either an unfamiliar, famous, or personally familiar face. The results indicate that adaptation transferred from unfamiliar faces to personally familiar faces more so than the converse and aftereffects did not transfer from famous faces to unfamiliar faces. These results are indicative of representation differences between unfamiliar, famous, and personally familiar faces, whereby personally familiar faces share representations of both unfamiliar and famous faces.

FIAEs in Famous Faces are Mediated by Type of Processing

An important question regarding face aftereffects is whether it is based on face-specific or lower-level mechanisms. One method for addressing this is to explore how adaptation in upright or inverted, photographic positive or negative faces transfers to test stimuli that are either upright or inverted and normal or negated. A series of studies are reported in which this is tested using a typical face identity aftereffect paradigm in unfamiliar and famous faces. Results showed that aftereffects were strongest when the adaptor matched the test stimuli. In addition, aftereffects did not transfer from upright adaptors to inverted test images, but did transfer from inverted adaptors to upright test images in famous faces. However, in unfamiliar faces, a different pattern was observed. The results are interpreted in terms of how identity adaptation interacts with low-level adaptation and highlight differences in the representation of famous and unfamiliar faces.

Shared or separate mechanisms for self-face and other-face processing? Evidence from adaptation

Evidence that self-face recognition is dissociable from general face recognition has important implications both for models of social cognition and for our understanding of face recognition. In two studies, we examine how adaptation affects the perception of personally familiar faces, and we use a visual adaptation paradigm to investigate whether the neural mechanisms underlying the recognition of one’s own and other faces are shared or separate. In Study 1 we show that the representation of personally familiar faces is rapidly updated by visual experience with unfamiliar faces, so that the perception of one’s own face and a friend’s face is altered by a brief period of adaptation to distorted unfamiliar faces. In Study 2, participants adapted to images of their own and a friend’s face distorted in opposite directions; the contingent aftereffects we observe are indicative of separate neural populations, but we suggest that these reflect coding of facial identity rather than of the categories “self” and “other.”

The effect of familiarity on face adaptation

Face aftereffects can provide information on how faces are stored by the human visual system (eg Leopold et al, 2001 Nature Neuroscience 4 89-94), but few studies have used robustly represented (highly familiar) faces. In this study we investigated the influence of facial familiarity on adaptation effects. Participants were adapted to a series of distorted faces (their own face, a famous face, or an unfamiliar face). In experiment 1, figural aftereffects were significantly smaller when participants were adapted to their own face than when they were adapted to the other faces (ie their own face appeared significantly less distorted than a famous or unfamiliar face). Experiment 2 showed that this 'own-face' effect did not occur when the same faces were used as adaptation stimuli for participants who were unfamiliar with them. Experiment 3 replicated experiment 1, but included a pre-adaptation baseline. The results highlight the importance of considering facial familiarity when conducting research on face aftereffects.

Visual adaptation and face perception

The appearance of faces can be strongly affected by the characteristics of faces viewed previously. These perceptual after-effects reflect processes of sensory adaptation that are found throughout the visual system, but which have been considered only relatively recently in the context of higher level perceptual judgements. In this review, we explore the consequences of adaptation for human face perception, and the implications of adaptation for understanding the neural-coding schemes underlying the visual representation of faces. The properties of face after-effects suggest that they, in part, reflect response changes at high and possibly face-specific levels of visual processing. Yet, the form of the after-effects and the norm-based codes that they point to show many parallels with the adaptations and functional organization that are thought to underlie the encoding of perceptual attributes like colour. The nature and basis for human colour vision have been studied extensively, and we draw on ideas and principles that have been developed to account for norms and normalization in colour vision to consider potential similarities and differences in the representation and adaptation of faces.

Identity-specific face adaptation effects: evidence for abstractive face representations

The effects of selective adaptation on familiar face perception were examined. After prolonged exposure to photographs of a celebrity, participants saw a series of ambiguous morphs that were varying mixtures between the face of that person and a different celebrity. Participants judged fewer of the morphs to resemble the celebrity to which they had been adapted, implying that they were now less sensitive to that particular face. Similar results were obtained when the adapting faces were highly dissimilar in viewpoint to the test morphs; when they were presented upside-down; or when they were vertically stretched to three times their normal height. These effects rule out explanations of adaptation effects solely in terms of low-level image-based adaptation. Instead they are consistent with the idea that relatively viewpoint-independent, person-specific adaptation occurred, at the level of either the "Face Recognition Units" or "Person Identity Nodes" in Burton, Bruce and Johnston's (1990) model of face recognition.

A novel face aftereffect based on recognition contrast thresholds

Previously, repulsive perceptual-shift face aftereffects have been reported. Here, we introduce a novel face adaptation method involving changes in contrast thresholds for face recognition. We find non-monotonic changes for adapted faces, with facilitation at short and suppression at long durations. Thresholds for unadapted faces were unaffected at short but elevated at long durations, more than those for the adapted face. A population-coding model showed that selective suppression of adapted representations cannot explain repulsive perceptual-bias aftereffects. The findings indicate greater complexity to adaptation, with facilitation, suppression, lateral inhibition of unadapted representations, and additional perceptual factors at long durations.

Three-dimensional information in face representations revealed by identity aftereffects

Representations of individual faces evolve with experience to support progressively more robust recognition. Knowledge of three-dimensional face structure is required to predict an image of a face as illumination and viewpoint change. Robust recognition across such transformations can be achieved with representations based on multiple two-dimensional views, three-dimensional structure, or both. We used face-identity adaptation in a familiarization paradigm to address a long-standing controversy concerning the role of two-dimensional versus three-dimensional information in face representations. We reasoned that if three-dimensional information is coded in the representations of familiar faces, then learning a new face using images generated by one three-dimensional transformation should enhance the robustness of the representation to another type of three-dimensional transformation. Familiarization with multiple views of faces enhanced the transfer of face-identity adaptation effects across changes in illumination by compensating for a generalization cost at a novel test viewpoint. This finding demonstrates a role for three-dimensional information in representations of familiar faces.

View-Contingent Aftereffects Suggest Joint Coding of Face Shape and View

While it is well established that different neural populations code different face views, behavioural evidence that these neurons also code other aspects of face shape is equivocal. For example, previous studies have interpreted the partial transfer of face aftereffects across different viewpoints as evidence for either view-specific coding of face shape or that the locus of adaptation is in face-coding mechanisms that are relatively robust to changes in face view. Here we show that it is possible to simultaneously induce aftereffects in opposite directions for 3/4 and front views of upright faces with manipulated mouth position (experiment 1). For example, simultaneous adaptation to 3/4 views with raised mouth position and front views with lowered mouth position caused raised mouth position to appear more normal for 3/4 views of novel faces, but less normal for front views. View-contingent adaptation did not occur for inverted faces, however (experiment 2). Dissociable aftereffects for different views of upright faces, but not for different views of inverted faces, suggest that neurons that code face view can also code other aspects of face shape.

Identity adaptation is mediated and moderated by visualisation ability

Depending on the previous research one reads, face aftereffects may or may not partially transfer across viewpoints. Two experiments are reported that explore this face-identity aftereffect by varying the adaptor type. Adaptation occurs when different views of the identity are used and even when the person's name is the adaptor. Brief exposure or nationality did not produce adaptation. In a third experiment, the role of visualisation was explored. Participants with higher visualisation scores showed greater adaptation to names than those with lower scores. These findings suggest non-facial identity cues can lead to a facial adaptation effect and that visualisation may be the mechanism behind this.

Turning the other cheek: the viewpoint dependence of facial expression after-effects

How do we visually encode facial expressions? Is this done by viewpoint-dependent mechanisms representing facial expressions as two-dimensional templates or do we build more complex viewpoint independent three-dimensional representations? Recent facial adaptation techniques offer a powerful way to address these questions. Prolonged viewing of a stimulus (adaptation) changes the perception of subsequently viewed stimuli (an after-effect). Adaptation to a particular attribute is believed to target those neural mechanisms encoding that attribute. We gathered images of facial expressions taken simultaneously from five different viewpoints evenly spread from the three-quarter leftward to the three-quarter rightward facing view. We measured the strength of expression after-effects as a function of the difference between adaptation and test viewpoints. Our data show that, although there is a decrease in after-effect over test viewpoint, there remains a substantial after-effect when adapt and test are at differing three-quarter views. We take these results to indicate that neural systems encoding facial expressions contain a mixture of viewpoint-dependent and viewpoint-independent elements. This accords with evidence from single cell recording studies in macaque and is consonant with a view in which viewpoint-independent expression encoding arises from a combination of view-dependent expression-sensitive responses.