A cognitive template for human face detection.

Faces are highly informative social stimuli, yet before any information can be accessed, the face must first be detected in the visual field. A detection template that serves this purpose must be able to accommodate the wide variety of face images we encounter, but how this generality could be achieved remains unknown. In this study, we investigate whether statistical averages of previously encountered faces can form the basis of a general face detection template. We provide converging evidence from a range of methods-human similarity judgements and PCA-based image analysis of face averages (Experiment 1-3), human detection behaviour for faces embedded in complex scenes (Experiment 4 and 5), and simulations with a template-matching algorithm (Experiment 6 and 7)-to examine the formation, stability and robustness of statistical image averages as cognitive templates for human face detection. We integrate these findings with existing knowledge of face identification, ensemble coding, and the development of face perception.

Penumbral thoughts: Contents of consciousness upon waking.

Thoughts shape our experience, choice, and behaviour throughout the day. Yet the content of 'penumbral thoughts'-first thoughts upon waking-has received very little research attention. Across seven independent samples (total N = 829), we used recall and reflection methods, solicited the same day, to understand what individuals think as they regain consciousness. These penumbral thoughts show remarkable thematic consistency: individuals were most likely to reflect on their somatic or psychological state, focus on temporal orientation, and prioritise waking actions. Survey results demonstrate that temporal and spatial orientation are dominated by the current time and the day ahead, rather than the past or other future timescales. Our results provide some insight into the order of priority in consciousness. We conclude that establishing one's temporal position is important to the daily process of 'rebooting' conscious awareness.

Weakened weekdays: lockdown disrupts the weekly cycle of risk tolerance.

Risk tolerance decreases from Monday to Thursday and increases on Friday. Antecedents of this weekly risk cycle are difficult to investigate experimentally as manipulating the seven-day cycle is impractical. Here we used temporal disorientation during the UK COVID-19 lockdown to conduct a natural experiment. In two studies, we measured responses to risk in participants with either a strong or weak sense of weekday, after either a short or long period of disruption to their weekly routine by lockdown. In Study 1 (N = 864), the weekly risk cycle was consistent in risk attitude measures specifically to participants who reported a strong sense of weekday. In Study 2 (N = 829), the weekly risk cycle was abolished, even for participants who retained a strong sense of weekday. We propose that two factors sustain the weekly risk cycle. If the sense of weekday is lacking, then weekday will have little effect because the current day is not salient. If weekday associations decay, then weekday will have little effect because the current day is not meaningful. The weekly risk cycle is strong and consistent when (i) sense of weekday is robust and (ii) weekday associations are maintained.

An Honest Joker reveals stereotypical beliefs about the face of deception.

Research on deception detection has mainly focused on Simple Deception, in which false information is presented as true. Relatively few studies have examined Sophisticated Deception, in which true information is presented as false. Because Sophisticated Deception incentivizes the appearance of dishonesty, it provides a window onto stereotypical beliefs about cues to deception. Here, we adapted the popular Joker Game to elicit spontaneous facial expressions under Simple Deception, Sophisticated Deception, and Plain Truth conditions, comparing facial behaviors in static, dynamic nonspeaking, and dynamic speaking presentations. Facial behaviors were analysed via machine learning using the Facial Action Coding System. Facial activations were more intense and longer lasting in the Sophisticated Deception condition than in the Simple Deception and Plain Truth conditions. More facial action units intensified in the static condition than in the dynamic speaking condition. Simple Deception involved leaked facial behaviors of which deceivers were unaware. In contrast, Sophisticated Deception involved deliberately leaked facial cues, including stereotypical cues to lying (e.g., gaze aversion). These stereotypes were inaccurate in the sense that they diverged from cues in the Simple Deception condition-the actual appearance of deception in this task. Our findings show that different modes of deception can be distinguished via facial action analysis. They also show that stereotypical beliefs concerning cues to deception can inform behavior. To facilitate future research on these topics, the multimodal stimuli developed in this study are available free for scientific use.

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.

A weekday intervention to reduce missed appointments.

The burden of missed healthcare appointments is so great that even small reductions in Did Not Attend (DNA) rate can secure tangible benefits. Previous studies have identified demographic factors that predict DNA rate. However, it is not obvious that these insights can be used to improve attendance, as healthcare providers do not control patient demographics. One factor that providers do control is appointment scheduling. We previously reported that appointments at the beginning of the week are more likely to be missed than appointments at the end of the week. This observation suggests a simple intervention to reduce DNA rate: schedule appointments for later in the week. Using data from a UK mental health hospital, we compared attendance rates for 12-months before and 12-months after the intervention began (916 appointments in total). Overall DNA rate fell from 34.2% pre-intervention to 23.4% post-intervention [χ2 (1, N = 916) = 13.01, p < 0.001; Relative Risk Reduction, 31.6%]. This effect was carried mainly by female patients, for whom more appointments could be moved to later in the week. Our findings confirm that DNA rate can be significantly reduced by loading appointments onto high-attendance days.

Capacity limits in face detection.

Face detection is a prerequisite for further face processing, such as extracting identity or semantic information. Those later processes appear to be subject to strict capacity limits, but the location of the bottleneck is unclear. In particular, it is not known whether the bottleneck occurs before or after face detection. Here we present a novel test of capacity limits in face detection. Across four behavioural experiments, we assessed detection of multiple faces via observers' ability to differentiate between two types of display. Fixed displays comprised items of the same type (all faces or all non-faces). Mixed displays combined faces and non-faces. Critically, a 'fixed' response requires all items to be processed. We found that additional faces could be detected with no cost to efficiency, and that this capacity-free performance was contingent on visual context. The observed pattern was not specific to faces, but detection was more efficient for faces overall. Our findings suggest that strict capacity limits in face perception occur after the detection step.

The roles of shape and texture in the recognition of familiar faces.

The surface texture of the face is proposed to be the dominant cue in face recognition. In this study, we investigated the role of shape information in face recognition. We compared the roles of shape and surface texture in the recognition of face identity using familiar and unfamiliar hybrid faces in which the average shape from one facial identity was combined with the average texture of a different identity. In the first experiment (n = 53), participants had to match the name of a familiar person to one of eight hybrid face images. In texture trials, all images had the correct shape, but only one image had the correct texture. In shape trials, all images had the correct texture, but only one image had the correct shape. Importantly, neither task could be performed by perceptual matching. Although performance was lower for the shape trials (81%) compared to texture trials (99%), both were significantly above chance (12.5%). In the second experiment (n = 110), participants had to name hybrid faces. There were two potentially correct answers for each face image: one based on the texture and one based on the shape. Participants reported the correct name based on the texture on 61% of trials and the correct name based on the shape on 12% of trials. In the third experiment (n = 19), fMR-adaptation was used to measure the neural sensitivity to changes in the shape or texture. The core face-selective regions showed a similar sensitivity to shape and texture. These findings confirm that texture is the dominant cue for face recognition, but also show that shape plays an important role in the recognition and neural response to familiar faces.

GFMT2: A psychometric measure of face matching ability.

We present an expanded version of a widely used measure of unfamiliar face matching ability, the Glasgow Face Matching Test (GFMT). The GFMT2 is created using the same source database as the original test but makes five key improvements. First, the test items include variation in head angle, pose, expression and subject-to-camera distance, making the new test more difficult and more representative of challenges in everyday face identification tasks. Second, short and long versions of the test each contain two forms that are calibrated to be of equal difficulty, allowing repeat tests to be performed to examine effects of training interventions. Third, the short-form tests contain no repeating face identities, thereby removing any confounding effects of familiarity that may have been present in the original test. Fourth, separate short versions are created to target exceptionally high performing or exceptionally low performing individuals using established psychometric principles. Fifth, all tests are implemented in an executable program, allowing them to be administered automatically. All tests are available free for scientific use via www.gfmt2.org .

Face-evoked thoughts.

The thoughts that come to mind when viewing a face depend partly on the face and partly on the viewer. This basic interaction raises the question of how much common ground there is in face-evoked thoughts, and how this compares to viewers' expectations. Previous analyses have focused on early perceptual stages of face processing. Here we take a more expansive approach that encompasses later associative stages. In Experiment 1 (free association), participants exhibited strong egocentric bias, greatly overestimating the extent to which other people's thoughts resembled their own. In Experiment 2, we show that viewers' familiarity with a face can be decoded from their face-evoked thoughts. In Experiment 3 (person association), participants reported who came to mind when viewing a face-a task that emphasises connections in a social network rather than nodes. Here again, viewers' estimates of common ground exceeded actual common ground by a large margin. We assume that a face elicits much the same thoughts in other people as it does in us, but that is a mistake. In this respect, we are more isolated than we think.

The Medusa effect reveals levels of mind perception in pictures.

Throughout our species history, humans have created pictures. The resulting picture record reveals an overwhelming preference for depicting things with minds. This preference suggests that pictures capture something of the mind that is significant to us, albeit at reduced potency. Here, we show that abstraction dims the perceived mind, even within the same picture. In a series of experiments, people were perceived as more real, and higher in both Agency (ability to do) and Experience (ability to feel), when they were presented as pictures than when they were presented as pictures of pictures. This pattern persisted across different tasks and even when comparators were matched for identity and image size. Viewers spontaneously discriminated between different levels of abstraction during eye tracking and were less willing to share money with a more abstracted person in a dictator game. Given that mind perception underpins moral judgement, our findings suggest that depicted persons will receive greater or lesser ethical consideration, depending on the level of abstraction.

Two Factors in Face Recognition: Whether You Know the Person's Face and Whether You Share the Person's Race.

One of the best-known phenomena in face recognition is the other-race effect, the observation that own-race faces are better remembered than other-race faces. However, previous studies have not put the magnitude of other-race effect in the context of other influences on face recognition. Here, we compared the effects of (a) a race manipulation (own-race/other-race face) and (b) a familiarity manipulation (familiar/unfamiliar face) in a 2 × 2 factorial design. We found that the familiarity effect was several times larger than the race effect in all performance measures. However, participants expected race to have a larger effect on others than it actually did. Face recognition accuracy depends much more on whether you know the person's face than whether you share the same race.

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

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

Dunning-Kruger effects in face perception.

The Dunning-Kruger Effect refers to a common failure of metacognitive insight in which people who are incompetent in a given domain are unaware of their incompetence. This effect has been found in a wide range of tasks, raising the question of whether there is any 'special' domain in which it is not found. One plausible candidate is face perception, which has sometimes been thought to be 'special'. To test this possibility, we assessed participants' insight into their own face perception abilities (self-estimates) and those of other people (peer estimates). We found classic Dunning-Kruger Effects in matching tasks for unfamiliar identity, familiar identity, gaze direction, and emotional expression. Low performers overestimated themselves, and high performers underestimated themselves. Interestingly, participants' self-estimates were more stable across tasks than their actual performance. In addition, peer estimates revealed a consistent egocentric bias. High performers attributed higher accuracy to other people than did low performers. We conclude that metacognitive insight into face perception abilities is limited and subject to systematic biases. Our findings urge caution when interpreting self-report measures of face perception ability. They also reveal a fundamental source of uncertainty in social interactions.

Facial identity across the lifespan.

We can recognise people that we know across their lifespan. We see family members age, and we can recognise celebrities across long careers. How is this possible, despite the very large facial changes that occur as people get older? Here we analyse the statistical properties of faces as they age, sampling photos of the same people from their 20s to their 70s. Across a number of simulations, we observe that individuals' faces retain some idiosyncratic physical properties across the adult lifespan that can be used to support moderate levels of age-independent recognition. However, we found that models based exclusively on image-similarity only achieved limited success in recognising faces across age. In contrast, more robust recognition was achieved with the introduction of a minimal top-down familiarisation procedure. Such models can incorporate the within-person variability associated with a particular individual to show a surprisingly high level of generalisation, even across the lifespan. The analysis of this variability reveals a powerful statistical tool for understanding recognition, and demonstrates how visual representations may support operations typically thought to require conceptual properties.

More human than human: a Turing test for photographed faces.

BACKGROUND: Recent experimental work has shown that hyper-realistic face masks can pass for real faces during live viewing. However, live viewing embeds the perceptual task (mask detection) in a powerful social context that may influence respondents' behaviour. To remove this social context, we assessed viewers' ability to distinguish photos of hyper-realistic masks from photos of real faces in a computerised two-alternative forced choice (2AFC) procedure. RESULTS: In experiment 1 (N = 120), we observed an error rate of 33% when viewing time was restricted to 500 ms. In experiment 2 (N = 120), we observed an error rate of 20% when viewing time was unlimited. In both experiments we saw a significant performance cost for other-race comparisons relative to own-race comparisons. CONCLUSIONS: We conclude that viewers could not reliably distinguish hyper-realistic face masks from real faces in photographic presentations. As well as its theoretical interest, failure to detect synthetic faces has important implications for security and crime prevention, which often rely on facial appearance and personal identity being related.

Deliberate disguise in face identification.

Facial image comparison is difficult for unfamiliar faces and easy for familiar faces. Those conclusions are robust, but they arise from situations in which the people being identified cooperate with the effort to identify them. In forensic and security settings, people are often motivated to subvert identification by manipulating their appearance, yet little is known about deliberate disguise and its effectiveness. We distinguish two forms of disguise-Evasion (trying not to look like oneself) and Impersonation (trying to look like another person). We present a new set of disguised face images (the FAÇADE image set), in which models altered their appearance to induce specific identification errors. In Experiment 1, unfamiliar observers were less accurate matching disguise items, especially evasion items, than matching undisguised items. A similar pattern held in Experiment 2, in which participants were informed about the disguise manipulations. In Experiment 3, familiar observers saw through impersonation disguise, but accuracy was lower for evasion disguise. Quantifying the performance cost of disguise reveals distinct performance profiles for impersonation and evasion. Evasion disguise was especially effective and reduced identification performance for familiar observers as well as for unfamiliar observers. We subsume these findings under a statistical framework of face learning. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

I recognise your name but I can't remember your face: An advantage for names in recognition memory.

Forgetting someone's name is a common failure of memory, and often occurs despite being able to recognise that person's face. This gives rise to the widespread view that memory for names is generally worse than memory for faces. However, this everyday error confounds stimulus class (faces vs. names) with memory task: recognition versus recall. Here we compare memory for faces and names when both are tested in the same recognition memory framework. Contrary to the common view, we find a clear advantage for names over faces. Across three experiments, we show that recognition of previously unfamiliar names exceeds recognition of previously unfamiliar faces. This advantage persists, even when the same face pictures are repeated at learning and test-a picture-memory task known to produce high levels of performance. Differential performance between names and faces disappears in recognition memory for familiar people. The results are discussed with reference to representational complexity and everyday memory errors.

Anchoring Effects in Facial Attractiveness.

First impressions from faces emerge quickly and shape subsequent behaviour. Given that different pictures of the same face evoke different impressions, we asked whether presentation order affects the overall impression of the person. In three experiments, we presented naturally varying photos of a person's face in ascending (low-to-high) or descending (high-to-low) order of attractiveness. We found that attractiveness ratings for a subsequent test item were higher for the descending condition than for the ascending condition (Experiment 1), consistent with anchoring effects. In Experiment 2, we ruled out contrast between the final item and the test item as the cause of the effect by demonstrating anchoring within the sequence itself. In Experiment 3, we found that order of image presentation also affected dating decisions. Our findings demonstrate that first impressions from faces depend not only on visual information but also on the order in which that information is received. We suggest that models of impression formation and learning of individual faces could be improved by considering temporal order of encounters.