Face Inversion Effect Research Articles

Modulating perceptual learning indexed by the face inversion effect: Simulating the application of transcranial direct current stimulation using the MKM model.

We report here two large studies investigating the effects of an established transcranial direct current stimulation (tDCS) procedure on perceptual learning as indexed by the face inversion effect. Experiments 1a and 1b (n = 128) examined the harmful generalization from Thatcherized faces to normal faces by directly comparing the size of the inversion effect for normal faces when presented intermixed with Thatcherized faces (Experiment 1a) versus that obtained when normal faces were presented intermixed with checkerboards (Experiment 1b). The results from the sham/control tDCS groups provide the first direct evidence in the literature showing how Thatcherized faces generalize onto normal ones producing a reduced inversion effect compared to when normal faces are presented with stimuli (e.g., checkerboards) that do not generalize significantly to normal faces. In the anodal tDCS groups, this effect was reversed, with a larger inversion effect recorded for normal faces in Experiment 1a versus that found in Experiment 1b. Further analyses within each experiment confirmed that the anodal tDCS procedure can enhance the inversion effect for normal faces in circumstances where harmful generalization would otherwise be produced by the Thatcherized faces (Experiment 1a). We also demonstrated our standard reduction in the inversion effect for normal faces consequent on the application of tDCS when presented intermixed with stimuli that do not generalize onto them. We interpret our results in terms of simulations using the MKM model of perceptual and associative learning. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Increased N250 elicited by facial familiarity: An ERP study including the face inversion effect and facial emotion processing

The present study aims to explore how familiarity modulates the neural processing of faces under different conditions: upright or inverted, neutral or emotional. To this purpose, 32 participants (25 female; age: M = 27.7 years, SD = 9.3) performed two face/emotion identification tasks during EEG recording. In the first task, to study facial processing, three different categories of facial stimuli were presented during a target detection task: famous familiar faces, faces of loved ones, and unfamiliar faces. To explore the face inversion effect according to each level of familiarity, these facial stimuli were also presented upside down. In the second task, to study emotional face processing, an emotional identification task on personally familiar and unfamiliar faces was conducted. The behavioural results showed an improved performance in the identification of facial expressions of emotion with the increase of facial familiarity, consistent with the previous literature. Regarding electrophysiological results, we found increased amplitudes of the P100, N170, and N250 for inverted compared to upright faces, independently of their degree of familiarity. Moreover, we did not find familiarity effects at the P100 and N170 time-windows, but we found that N250 amplitude was larger for personally familiar compared to unfamiliar faces. This result supports the reasoning that the facial familiarity increases the neural activity during the N250 time-window, which may be explained by the processing of additional information prompted by the viewing of our loved ones faces, in contrast to what happens with unfamiliar individuals.

The composite effect reveals that human (but not other primate) faces are special to humans.

Recognising faces is widely believed to be achieved using "special" neural and cognitive mechanisms that depend on "holistic" processing, which are not used when recognising other kinds of objects. An important, but largely unaddressed, question is how much like a Human face a stimulus needs to be to engage this "special" mechanism(s). In the current study, we attempted to answer this question in 3 ways. In Experiments 1 and 2 we examined the extent to which the disproportionate inversion effect for human faces extends to the faces of other species (including a range of other primates). Results suggested that the faces of other primates engage the mechanism responsible for the inversion effect approximately as well as that mechanism is engaged by Human faces, but that non-primate faces engage the mechanism less well. And so primate faces, in general, seem to produce a disproportionate inversion effect. In Experiment 3 we examined the extent to which the Composite effect extends to the faces of a range of other primates, and found no compelling evidence of a composite effect for the faces of any other primate. The composite effect was exclusive to Human faces. Because these data differ so dramatically from a previously reported study asking similar questions Taubert (2009), we also (in Experiment 4) ran an exact replication of Taubert's Experiment 2, which reported on both Inversion and Composite effects in a range of species. We were unable to reproduce the pattern of data reported by Taubert. Overall, the results suggest that the disproportionate inversion effect extends to all of the faces of the non-human primates tested, but that the composite effect is exclusive to Human faces.

Deep learning models challenge the prevailing assumption that face-like effects for objects of expertise support domain-general mechanisms.

The question of whether task performance is best achieved by domain-specific, or domain-general processing mechanisms is fundemental for both artificial and biological systems. This question has generated a fierce debate in the study of expert object recognition. Because humans are experts in face recognition, face-like neural and cognitive effects for objects of expertise were considered support for domain-general mechanisms. However, effects of domain, experience and level of categorization, are confounded in human studies, which may lead to erroneous inferences. To overcome these limitations, we trained deep learning algorithms on different domains (objects, faces, birds) and levels of categorization (basic, sub-ordinate, individual), matched for amount of experience. Like humans, the models generated a larger inversion effect for faces than for objects. Importantly, a face-like inversion effect was found for individual-based categorization of non-faces (birds) but only in a network specialized for that domain. Thus, contrary to prevalent assumptions, face-like effects for objects of expertise do not support domain-general mechanisms but may originate from domain-specific mechanisms. More generally, we show how deep learning algorithms can be used to dissociate factors that are inherently confounded in the natural environment of biological organisms to test hypotheses about their isolated contributions to cognition and behaviour.

Literacy acquisition facilitates inversion effects for faces with full-, low-, and high-spatial frequency: evidence from illiterate and literate adults.

Previous studies have found that literacy acquisition modulates configural face processing (i.e., holistic and second-order configural processing). However, it remains unclear how literacy acquisition impacts the configural processing indexed by the inversion effect of normal or filtered faces. We asked Chinese illiterate and literate adults to judge whether two sequentially-presented stimuli, including faces, houses (experiment 1), and high- or low-pass filtered faces (experiment 2) were identical. Literate adults outperformed illiterate controls in the upright face and house conditions (experiment 1) and the upright high- and low-pass filtered conditions (experiment 2) but not in the inverted conditions. Notably, the size of an inversion effect (i.e., subtracting inverted accuracy from upright accuracy) was greater among literate adults than that among illiterate adults in both experiments. These findings support that literacy acquisition promotes configural face processing.

Face processing in young adults with autism and ADHD: An event related potentials study.

Atypicalities in perception and interpretation of faces and emotional facial expressions have been reported in both autism and attention-deficit/hyperactivity disorder (ADHD) during childhood and adulthood. Investigation of face processing during young adulthood (18 to 25 years), a transition period to full-fledged adulthood, could provide important information on the adult outcomes of autism and ADHD. In this study, we investigated event-related potentials (ERPs) related to visual face processing in autism, ADHD, and co-occurring autism and ADHD in a large sample of young adults (N = 566). The groups were based on the Diagnostic Interview for ADHD in Adults 2.0 (DIVA-2) and the Autism Diagnostic Observation Schedule-2 (ADOS-2). We analyzed ERPs from two passive viewing tasks previously used in childhood investigations: (1) upright and inverted faces with direct or averted gaze; (2) faces expressing different emotions. Across both tasks, we consistently found lower amplitude and longer latency of N170 in participants with autism compared to those without. Longer P1 latencies and smaller P3 amplitudes in response to emotional expressions and longer P3 latencies for upright faces were also characteristic to the autistic group. Those with ADHD had longer N170 latencies, specific to the face-gaze task. Individuals with both autism and ADHD showed additional alterations in gaze modulation and a lack of the face inversion effect indexed by a delayed N170. Alterations in N170 for autistic young adults is largely consistent with studies on autistic adults, and some studies in autistic children. These findings suggest that there are identifiable and measurable socio-functional atypicalities in young adults with autism.

The face inversion effect does not provide a pure measure of holistic face processing.

It is widely held that upright faces are processed more holistically than inverted faces and that this difference is reflected in the face inversion effect. It is not clear, however, how the inversion effect can best be measured, whether it is task specific, or even whether it specifically correlates with processing of upright faces. We examined these questions in a large sample (N = 420) who provided data on processing of upright and inverted stimuli in two different tasks with faces and one with objects. We find that the inversion effects are task dependent, and that they do not correlate better among face processing tasks than they do across face and object processing tasks. These findings were obtained regardless of whether inversion effects were measured by means of difference scores or regression. In comparison, only inversion effects based on regression predicted performance with upright faces in tasks other than those the inversion effects were derived from. Critically, however, inversion effects based on regression also predicted performance with inverted faces to a similar degree as they predicted performance with upright faces. Consequently, and contrary to what is commonly assumed, inversion effects do not seem to capture effects specific to holistic processing of upright faces. While the present findings do not bring us closer to an understanding of which changes in cognitive processing are induced by inversion, they do suggest that inversion effects do not reflect a unitary construct; an implicit assumption that seems to characterize much of the research regarding face processing.

Supplemental Material for Modulating Perceptual Learning Indexed by the Face Inversion Effect: Simulating the Application of Transcranial Direct Current Stimulation Using the MKM Model

Supplemental Material for Modulating Perceptual Learning Indexed by the Face Inversion Effect: Simulating the Application of Transcranial Direct Current Stimulation Using the MKM Model

Reducing the perceptual field of view does not cause the face inversion effect

Reducing the perceptual field of view does not cause the face inversion effect

Investigating the origins of the face inversion effect with an extraordinary participant

Investigating the origins of the face inversion effect with an extraordinary participant

The face inversion effect or the face upright effect?

The face inversion effect (FIE) refers to the observation that presenting stimuli upside-down impairs the processing of faces disproportionally more than other mono-oriented objects. This has been taken as evidence that processing of faces and objects differ qualitatively. However, nearly all FIE studies are based on comparing individuation of upright faces, which most people are rather good at, with individuation of objects most people are much less familiar with individuating (e.g., radios and airplanes). Consequently, the FIE may mainly reflect differences between categories in how they are processed prior to inversion, with within-category discrimination of upright faces being a much more familiar task than within-category discrimination among members belonging to other object classes. We tested this hypothesis by comparing inversion effects for faces and objects using object recognition tasks that do not require within-category discrimination (object decision and old/new recognition memory tasks). In all tasks (seven with objects and two with faces) we find credible inversion effects, but in no instance were these effects significantly larger for faces than for objects. This suggests that the FIE can be a product of familiarity with the type of identification process required in the upright conditions rather than some process that is selectively affected for faces when stimuli are inverted.

Face inversion does not affect the reversed congruency effect of gaze.

In a spatial Stroop task, the eye-gaze target produces the reversed congruency effect-responses become shorter when the gaze direction and its location are incongruent than when they are congruent. The present study examined the face inversion effect on the gaze spatial Stroop task to clarify whether the holistic face processing or part-based processing of the eyes is responsible for the reversed congruency effect. In Experiment 1, participants judged the gaze direction of the upright or inverted face with a neutral expression presented either in the left or right visual field. In Experiment 2, we examined whether face inversion interacted with facial expressions (i.e., angry, happy, neutral, and sad). Face inversion disrupted holistic face processing, slowing down the overall performance relative to the performance with upright faces. However, face inversion did not affect the reversed congruency effect. These results further support the parts-based processing account and suggest that while faces are processed holistically, the reversed congruency effect, relying on the extracted local features (i.e., eyes), may be processed in a part-based manner.

Robots engage face-processing less strongly than humans.

Robot faces often differ from human faces in terms of their facial features (e.g., lack of eyebrows) and spatial relationships between these features (e.g., disproportionately large eyes), which can influence the degree to which social brain [i.e., Fusiform Face Area (FFA), Superior Temporal Sulcus (STS); Haxby et al., 2000] areas process them as social individuals that can be discriminated from other agents in terms of their perceptual features and person attributes. Of interest in this work is whether robot stimuli are processed in a less social manner than human stimuli. If true, this could undermine human-robot interactions (HRIs) because human partners could potentially fail to perceive robots as individual agents with unique features and capabilities-a phenomenon known as outgroup homogeneity-potentially leading to miscalibration of trust and errors in allocation of task responsibilities. In this experiment, we use the face inversion paradigm (as a proxy for neural activation in social brain areas) to examine whether face processing differs between human and robot face stimuli: if robot faces are perceived as less face-like than human-faces, the difference in recognition performance for faces presented upright compared to upside down (i.e., inversion effect) should be less pronounced for robot faces than human faces. The results demonstrate a reduced face inversion effect with robot vs. human faces, supporting the hypothesis that robot faces are processed in a less face-like manner. This suggests that roboticists should attend carefully to the design of robot faces and evaluate them based on their ability to engage face-typical processes. Specific design recommendations on how to accomplish this goal are provided in the discussion.

Transcranial direct current stimulation (tDCS) eliminates the other-race effect (ORE) indexed by the face inversion effect for own versus other-race faces

We investigate here individuals’ reduced ability to recognise faces from other racial backgrounds, a robust phenomenon named the other-race effect (ORE). In this literature the term “race” is used to refer to visually distinct ethnic groups. In our study, we will refer to two of such groups: Western Caucasian (also known as White European) and East Asian e.g., Chinese, Japanese, Korean. This study applied the tDCS procedure (double-blind, 10 min duration, 1.5 mA intensity, targeting Fp3 location), developed in the perceptual learning literature, specifically used to remove the expertise component of the face inversion effect (FIE), which consists of higher recognition performance for upright than inverted faces. In the tDCS-sham condition (N = 48) we find a robust ORE i.e., significantly larger FIE for own versus other-race faces due to higher performance for upright own-race faces. Critically, in the anodal-tDCS condition (N = 48) the FIE for own-race faces was significantly reduced compared to sham due to impaired performance for upright faces thus eliminating the cross-race interaction index of the ORE. Our results support the major role that perceptual expertise, manifesting through perceptual learning, has in determining the ORE indexed by the FIE.

Comparing Inverted Faces to Upright Faces Using Similarity or Mental Rotation

Abstract A major interest of research in face recognition lies in explaining the face inversion effect, in which the recognition of an inverted face is less successful than that of an upright face. However, prior research has devoted little effort to examining how the cognitive system handles comparison between inverted and upright faces. The results of a preparatory experiment and two following experiments support the conclusion that the comparison is based more on visual similarity than on a mental rotation of the inverted face to an upright face. Visual similarity is based on certain elements mutual to the two faces, which resist the transformation of inversion. These elements are symmetric or salient components of the face, such as round eyes or thick lips.

Seeing faces where there are none: Pareidolia correlates with age but not autism traits

Previous research has found that individuals with autism spectrum disorder experience difficulties when visually processing face stimuli compared to developmentally typical individuals. Whether, in the typically-developing population, face detection depends on autism-like traits (ALTs) is less clear. In this report, we aimed to develop an experimental design that is more sensitive to any individual differences in face detection than previous reports. We employed pareidolia, that is, cases where non-face stimuli are perceived to be faces, assuming this is more difficult than detection of ‘real’ faces, decreasing changes of ceiling performance. We also show multiple faces per trial, allowing for a more graded assessment of face detection ability. Participants were 263 individuals aged between 18 and 82 years of age. Pareidolia was investigated in two online experiments, with different types of stimuli: objects that could be perceived as faces (i.e., embedded faces task) and Mooney faces (Mooney face task). In the latter condition, we also investigated the face inversion effect. We found that neither detection ability or the inversion effect depended on ALTs. We did find a dependence of age for both measures, and a complex dependence on gender for Mooney faces. Our data suggest that face detection (and specifically pareidolia) does not depend on ALTs, but does depend on the age of the observer. The dependence on age appears to be different between the two experiments, suggesting that the underlying mechanisms necessary for face detection in our two experiments mature and decline at different rates.

The Face Inversion Effect in Deep Convolutional Neural Networks.

The face inversion effect (FIE) is a behavioral marker of face-specific processing that the recognition of inverted faces is disproportionately disrupted than that of inverted non-face objects. One hypothesis is that while upright faces are represented by face-specific mechanism, inverted faces are processed as objects. However, evidence from neuroimaging studies is inconclusive, possibly because the face system, such as the fusiform face area, is interacted with the object system, and therefore the observation from the face system may indirectly reflect influences from the object system. Here we examined the FIE in an artificial face system, visual geometry group network-face (VGG-Face), a deep convolutional neural network (DCNN) specialized for identifying faces. In line with neuroimaging studies on humans, a stronger FIE was found in VGG-Face than that in DCNN pretrained for processing objects. Critically, further classification error analysis revealed that in VGG-Face, inverted faces were miscategorized as objects behaviorally, and the analysis on internal representations revealed that VGG-Face represented inverted faces in a similar fashion as objects. In short, our study supported the hypothesis that inverted faces are represented as objects in a pure face system.

Inversion Effect of Hand Postures: Effect of Visual Experience Over Long and Short Term.

Some researchers argue that holistic processing is unique to face recognition supported by the face inversion effect. However, findings such as the body inversion effect challenge the face processing-specificity hypothesis, thus supporting the expertise hypothesis. Few studies have explored a possible hand inversion effect which could involve special processing similar to the face and body. We conducted four experiments to investigate the time course and flexibility of the hand posture inversion effect. We utilized a same/different discrimination task (Experiments 1 and 2), an identification task (Experiment 3), and a training paradigm involving the exposure of different hand orientations (Experiment 4). The results show the hand posture inversion effect (with fingers up as upright orientation) was not initially observed during the early phase of testing, but occurred in later phases. This suggests that both lifetime experience and recent exposure affect the hand posture inversion effect. We also found the hand posture inversion effect, once established, was stable across days and remained consistent across different tasks. In addition, the hand posture inversion effect for specific orientations could be obtained with short-term training of a given orientation, indicating the cognitive process is flexible.

Deep Convolutional Neural Network-Based Visual Stimuli Classification Using Electroencephalography Signals of Healthy and Alzheimer's Disease Subjects.

Visual perception is an important part of human life. In the context of facial recognition, it allows us to distinguish between emotions and important facial features that distinguish one person from another. However, subjects suffering from memory loss face significant facial processing problems. If the perception of facial features is affected by memory impairment, then it is possible to classify visual stimuli using brain activity data from the visual processing regions of the brain. This study differentiates the aspects of familiarity and emotion by the inversion effect of the face and uses convolutional neural network (CNN) models (EEGNet, EEGNet SSVEP (steady-state visual evoked potentials), and DeepConvNet) to learn discriminative features from raw electroencephalography (EEG) signals. Due to the limited number of available EEG data samples, Generative Adversarial Networks (GAN) and Variational Autoencoders (VAE) are introduced to generate synthetic EEG signals. The generated data are used to pretrain the models, and the learned weights are initialized to train them on the real EEG data. We investigate minor facial characteristics in brain signals and the ability of deep CNN models to learn them. The effect of face inversion was studied, and it was observed that the N170 component has a considerable and sustained delay. As a result, emotional and familiarity stimuli were divided into two categories based on the posture of the face. The categories of upright and inverted stimuli have the smallest incidences of confusion. The model’s ability to learn the face-inversion effect is demonstrated once more.

A multilevel Bayesian meta-analysis of the body inversion effect: Evaluating controversies over headless and sexualized bodies.

Face and body perception rely on specialized processing mechanisms to interpret social information efficiently. The body inversion effect (BIE), refers to an inversion effect for bodies, such that recognition of bodies is impaired by inversion. The BIE, like the face inversion effect (FIE), is particularly important because a disproportionate BIE relative to inversion effects for objects could be interpreted in much the same way as the disproportionate FIE has often been characterized; that is, as evidence of specialized, configural processing. However, research supporting the BIE is marked by methodological heterogeneity and mixed findings. Our multilevel Bayesian meta-analysis addresses inconsistencies in the literature by pooling data from numerous studies to estimate the magnitude of the BIE across various methodological and stimulus properties. We included 180 effect sizes from 41 empirical articles representing data from 2,274 participants. Overall, we found that the BIE was moderate-large in magnitude (Hedges' g = 0.75). Importantly, the inversion effect was larger for bodies than objects (b = 0.42); however, the inversion effect for faces was larger than for bodies (b = 0.34). We tested the role of discrimination dimension, stimulus type, face/head inclusion, stimulus sexualization, and sexualized stimulus sex as moderators of the BIE. We found that the BIE was moderated by discrimination dimension, stimulus type, stimulus sexualization, and sexualized stimulus sex. By synthesizing the existing literature, we provide a better theoretical understanding of how underlying visual processing mechanisms may differ for different types of social information (i.e., bodies vs. faces).