The resolution of face perception varies systematically across the visual field.

Visual abilities tend to vary predictably across the visual field-for simple low-level stimuli, visibility is better along the horizontal vs. vertical meridian and in the lower vs. upper visual field. In contrast, face perception abilities have been reported to show either distinct or entirely idiosyncratic patterns of variation in peripheral vision, suggesting a dissociation between the spatial properties of low- and higher-level vision. To assess this link more clearly, we extended methods used in low-level vision to develop an acuity test for face perception, measuring the smallest size at which facial gender can be reliably judged in peripheral vision. In 3 experiments, we show the characteristic inversion effect, with better acuity for upright faces than inverted, demonstrating the engagement of high-level face-selective processes in peripheral vision. We also observe a clear advantage for gender acuity on the horizontal vs. vertical meridian and a smaller-but-consistent lower- vs. upper-field advantage. These visual field variations match those of low-level vision, indicating that higher-level face processing abilities either inherit or actively maintain the characteristic patterns of spatial selectivity found in early vision. The commonality of these spatial variations throughout the visual hierarchy means that the location of faces in our visual field systematically influences our perception of them.

Radial bias in face identification.

Human vision in the periphery is most accurate for stimuli that point towards the fovea. This so-called radial bias has been linked with the organization and spatial selectivity of neurons at the lowest levels of the visual system, from retinal ganglion cells onwards. Despite evidence that the human visual system is radially biased, it is not yet known whether this bias persists at higher levels of processing, or whether high-level representations are invariant to this low-level orientation bias. We used the case of face identity recognition to address this question. The specialized high-level mechanisms that support efficient face recognition are highly dependent on horizontally oriented information, which convey the most useful identity cues in the fovea. We show that face selective mechanisms are more sensitive on the horizontal meridian (to the left and right of fixation) compared to the vertical meridian (above and below fixation), suggesting that the horizontal cues in the face are better extracted on the horizontal meridian, where they align with the radial bias. The results demonstrate that the radial bias is maintained at high-level recognition stages and emphasize the importance of accounting for the radial bias in future investigation of visual recognition processes in peripheral vision.

Contributions of low- and high-level contextual mechanisms to human face perception.

Contextual modulations at primary stages of visual processing depend on the strength of local input. Contextual modulations at high-level stages of (face) processing show a similar dependence to local input strength. Namely, the discriminability of a facial feature determines the amount of influence of the face context on that feature. How high-level contextual modulations emerge from primary mechanisms is unclear due to the scarcity of empirical research systematically addressing the functional link between the two. We tested (62) young adults' ability to process local input independent of the context using contrast detection and (upright and inverted) morphed facial feature matching tasks. We first investigated contextual modulation magnitudes across tasks to address their shared variance. A second analysis focused on the profile of performance across contextual conditions. In upright eye matching and contrast detection tasks, contextual modulations only correlated at the level of their profile (averaged Fisher-Z transformed r = 1.18, BF10 > 100), but not magnitude (r = .15, BF10 = .61), suggesting the functional independence but similar working principles of the mechanisms involved. Both the profile (averaged Fisher-Z transformed r = .32, BF10 = 9.7) and magnitude (r = .28, BF10 = 4.58) of the contextual modulations correlated between inverted eye matching and contrast detection tasks. Our results suggest that non-face-specialized high-level contextual mechanisms (inverted faces) work in connection to primary contextual mechanisms, but that the engagement of face-specialized mechanisms for upright faces obscures this connection. Such combined study of low- and high-level contextual modulations sheds new light on the functional relationship between different levels of the visual processing hierarchy, and thus on its functional organization.

Backward masking reveals coarse-to-fine dynamics in human V1.

Natural images exhibit luminance variations aligned across a broad spectrum of spatial frequencies (SFs). It has been proposed that, at early stages of processing, the coarse signals carried by the low SF (LSF) of the visual input are sent rapidly from primary visual cortex (V1) to ventral, dorsal and frontal regions to form a coarse representation of the input, which is later sent back to V1 to guide the processing of fine-grained high SFs (i.e., HSF). We used functional resonance imaging (fMRI) to investigate the role of human V1 in the coarse-to-fine integration of visual input. We disrupted the processing of the coarse and fine content of full-spectrum human face stimuli via backward masking of selective SF ranges (LSFs: <1.75cpd and HSFs: >1.75cpd) at specific times (50, 83, 100 or 150 ms). In line with coarse-to-fine proposals, we found that (1) the selective masking of stimulus LSF disrupted V1 activity in the earliest time window, and progressively decreased in influence, while (2) an opposite trend was observed for the masking of stimulus' HSF. This pattern of activity was found in V1, as well as in ventral (i.e. the Fusiform Face area, FFA), dorsal and orbitofrontal regions. We additionally presented subjects with contrast negated stimuli. While contrast negation significantly reduced response amplitudes in the FFA, as well as coupling between FFA and V1, coarse-to-fine dynamics were not affected by this manipulation. The fact that V1 response dynamics to strictly identical stimulus sets differed depending on the masked scale adds to growing evidence that V1 role goes beyond the early and quasi-passive transmission of visual information to the rest of the brain. It instead indicates that V1 may yield a 'spatially registered common forum' or 'blackboard' that integrates top-down inferences with incoming visual signals through its recurrent interaction with high-level regions located in the inferotemporal, dorsal and frontal regions.

Natural Contrast Statistics Facilitate Human Face Categorization.

The ability to detect faces in the environment is of utmost ecological importance for human social adaptation. While face categorization is efficient, fast and robust to sensory degradation, it is massively impaired when the facial stimulus does not match the natural contrast statistics of this visual category, i.e., the typically experienced ordered alternation of relatively darker and lighter regions of the face. To clarify this phenomenon, we characterized the contribution of natural contrast statistics to face categorization. Specifically, 31 human adults viewed various natural images of nonface categories at a rate of 12 Hz, with highly variable images of faces occurring every eight stimuli (1.5 Hz). As in previous studies, neural responses at 1.5 Hz as measured with high-density electroencephalography (EEG) provided an objective neural index of face categorization. Here, when face images were shown in their naturally experienced contrast statistics, the 1.5-Hz face categorization response emerged over occipito-temporal electrodes at very low contrast [5.1%, or 0.009 root-mean-square (RMS) contrast], quickly reaching optimal amplitude at 22.6% of contrast (i.e., RMS contrast of 0.041). Despite contrast negation preserving an image's spectral and geometrical properties, negative contrast images required twice as much contrast to trigger a face categorization response, and three times as much to reach optimum. These observations characterize how the internally stored natural contrast statistics of the face category facilitate visual processing for the sake of fast and efficient face categorization.

Coarse-to-Fine(r) Automatic Familiar Face Recognition in the Human Brain.

At what level of spatial resolution can the human brain recognize a familiar face in a crowd of strangers? Does it depend on whether one approaches or rather moves back from the crowd? To answer these questions, 16 observers viewed different unsegmented images of unfamiliar faces alternating at 6 Hz, with spatial frequency (SF) content progressively increasing (i.e., coarse-to-fine) or decreasing (fine-to-coarse) in different sequences. Variable natural images of celebrity faces every sixth stimulus generated an objective neural index of single-glanced automatic familiar face recognition (FFR) at 1 Hz in participants' electroencephalogram (EEG). For blurry images increasing in spatial resolution, the neural FFR response over occipitotemporal regions emerged abruptly with additional cues at about 6.3-8.7 cycles/head width, immediately reaching amplitude saturation. When the same images progressively decreased in resolution, the FFR response disappeared already below 12 cycles/head width, thus providing no support for a predictive coding hypothesis. Overall, these observations indicate that rapid automatic recognition of heterogenous natural views of familiar faces is achieved from coarser visual inputs than generally thought, and support a coarse-to-fine FFR dynamics in the human brain.

Information redundancy across spatial scales modulates early visual cortical processing.

Visual images contain redundant information across spatial scales where low spatial frequency contrast is informative towards the location and likely content of high spatial frequency detail. Previous research suggests that the visual system makes use of those redundancies to facilitate efficient processing. In this framework, a fast, initial analysis of low-spatial frequency (LSF) information guides the slower and later processing of high spatial frequency (HSF) detail. Here, we used multivariate classification as well as time-frequency analysis of MEG responses to the viewing of intact and phase scrambled images of human faces to demonstrate that the availability of redundant LSF information, as found in broadband intact images, correlates with a reduction in HSF representational dominance in both early and higher-level visual areas as well as a reduction of gamma-band power in early visual cortex. Our results indicate that the cross spatial frequency information redundancy that can be found in all natural images might be a driving factor in the efficient integration of fine image details.

Contrast versus identity encoding in the face image follow distinct orientation selectivity profiles.

Orientation selectivity is a fundamental property of primary visual encoding. High-level processing stages also show some form of orientation dependence, with face identification preferentially relying on horizontally-oriented information. How high-level orientation tuning emerges from primary orientation biases is unclear. In the same group of participants, we derived the orientation selectivity profile at primary and high-level visual processing stages using a contrast detection and an identity matching task. To capture the orientation selectivity profile, we calculated the difference in performance between all tested orientations (0, 45, 90, and 135°) for each task and for upright and inverted faces, separately. Primary orientation selectivity was characterized by higher sensitivity to oblique as compared to cardinal orientations. The orientation profile of face identification showed superior horizontal sensitivity to face identity. In each task, performance with upright and inverted faces projected onto qualitatively similar a priori models of orientation selectivity. Yet the fact that the orientation selectivity profiles of contrast detection in upright and inverted faces correlated significantly while such correlation was absent for identification indicates a progressive dissociation of orientation selectivity profiles from primary to high-level stages of orientation encoding. Bayesian analyses further indicate a lack of correlation between the orientation selectivity profiles in the contrast detection and face identification tasks, for upright and inverted faces. From these findings, we conclude that orientation selectivity shows distinct profiles at primary and high-level stages of face processing and that a transformation must occur from general cardinal attenuation when processing basic properties of the face image to horizontal tuning when encoding more complex properties such as identity.

Fixed or flexible? Orientation preference in identity and gaze processing in humans.

Vision begins with the encoding of contrast at specific orientations. Several works showed that humans identify their conspecifics best based on the horizontally-oriented information contained in the face image; this range conveys the main morphological features of the face. In contrast, the vertical structure of the eye region seems to deliver optimal cues to gaze direction. The present work investigates whether the human face processing system flexibly tunes to vertical information contained in the eye region when processing gaze direction. Alternatively, face processing may invariantly rely on the horizontal range, supporting the domain specificity of orientation tuning for faces and the gateway role of horizontal content to access any type of facial information. Participants judged the gaze direction of faces staring at a range of lateral positions. They additionally performed an identification task with upright and inverted face stimuli. Across tasks, stimuli were filtered to selectively reveal horizontal (H), vertical (V), or combined (HV) information. Most participants identified faces better based on horizontal than vertical information confirming the horizontal tuning of face identification. In contrast, they showed a vertically-tuned sensitivity to gaze direction. The logistic functions fitting the "left" and "right" response proportion as a function of gaze direction were indeed steeper when based on vertical than on horizontal information. The finding of a vertically-tuned processing of gaze direction favours the hypothesis that visual encoding of face information flexibly switches to the orientation channel carrying the cues most relevant to the task at hand. It suggests that horizontal structure, though predominant in the face stimulus, is not a mandatory gateway for efficient face processing. The present evidence may help better understand how visual signals travel the visual system to enable rich and complex representations of naturalistic stimuli such as faces.

Coarse-to-fine information integration in human vision.

Coarse-to-fine theories of vision propose that the coarse information carried by the low spatial frequencies (LSF) of visual input guides the integration of finer, high spatial frequency (HSF) detail. Whether and how LSF modulates HSF processing in naturalistic broad-band stimuli is still unclear. Here we used multivariate decoding of EEG signals to separate the respective contribution of LSF and HSF to the neural response evoked by broad-band images. Participants viewed images of human faces, monkey faces and phase-scrambled versions that were either broad-band or filtered to contain LSF or HSF. We trained classifiers on EEG scalp-patterns evoked by filtered scrambled stimuli and evaluated the derived models on broad-band scrambled and intact trials. We found reduced HSF contribution when LSF was informative towards image content, indicating that coarse information does guide the processing of fine detail, in line with coarse-to-fine theories. We discuss the potential cortical mechanisms underlying such coarse-to-fine feedback.

From coarse to fine: Interactive feature processing precedes local feature analysis in human face perception.

Face perception depends on a dynamic interplay of a "holistic" Interactive Feature Processing (IFP) and a Local Feature Processing (LFP) style. However, it is unclear whether features are processed locally before they are integrated into a holistic percept (Fine-to-Coarse strategy), or whether local feature processing occurs only after a holistic percept is established (Coarse-to-Fine strategy). The present Event-Related Potentials study investigates whether IFP precedes LFP (Coarse-to-Fine) or vice versa (Fine-to-Coarse). Participants matched target features within face pairs (here the eye region), in which distracter features (nose and mouth) called for the same or a different response (congruent and incongruent, respectively). Psychophysical results replicated previous findings. That is, dissimilar target features are locally processed (LFP), which minimizes interference from surrounding incongruent distracters. Conversely, an IFP mode is elicited when similar target features are embedded in dissimilar contexts. In IFP mode, incongruent distracters do interfere with the processing of similar target features, thereby deteriorating task performance. Face inversion, which preserves input properties but disrupts high-level face perception, annihilated these incongruency effects. Psychophysical observations were reflected at the neural level. The IFP and LFP modes of face perception elicited distinct time-courses in occipito-temporal cortex. IFP was affected by inversion as soon as 176 ms post-stimulus onset (coinciding with the N170 peak). In contrast, the first robust indications of LFP occurred 120 ms later, at 296 ms. Thus, the contribution of IFP to high-level face perception appears to temporally precede LFP. Moreover, results showed that the IFP and LFP modes did not only operate in distinct time intervals, but also in different brain areas: activity associated with the IFP mode was right-lateralized, whereas the LPF mode engaged the left hemisphere. In sum, interactive "holistic" encoding of facial features temporally precedes their local analysis. This agrees with models suggesting a Coarse-to-Fine strategy for face perception, in line with generic descriptions of visual perception in which global scene analysis precedes the examination of local details.

Crowding for faces is determined by visual (not holistic) similarity: Evidence from judgements of eye position.

Crowding (the disruption of object recognition in clutter) presents the fundamental limitation on peripheral vision. For simple objects, crowding is strong when target/flanker elements are similar and weak when they differ - a selectivity for target-flanker similarity. In contrast, the identification of upright holistically-processed face stimuli is more strongly impaired by upright than inverted flankers, whereas inverted face-targets are impaired by both - a pattern attributed to an additional stage of crowding selective for "holistic similarity" between faces. We propose instead that crowding is selective for target-flanker similarity in all stimuli, but that this selectivity is obscured by task difficulty with inverted face-targets. Using judgements of horizontal eye-position that are minimally affected by inversion, we find that crowding is strong when target-flanker orientations match and weak when they differ for both upright and inverted face-targets. By increasing task difficulty, we show that this selectivity for target-flanker similarity is obscured even for upright face-targets. We further demonstrate that this selectivity follows differences in the spatial order of facial features, rather than "holistic similarity" per se. There is consequently no need to invoke a distinct stage of holistic crowding for faces - crowding is selective for target-flanker similarity, even with complex stimuli such as faces.

Corrigendum: The orientation selectivity of face identification.

This corrects the article DOI: 10.1038/srep34204.

Ultra-coarse, single-glance human face detection in a dynamic visual stream.

Effective human interaction depends on our ability to rapidly detect faces in dynamic visual environments. Here we asked how basic units of visual information (spatial frequencies, or SF) contribute to this fundamental brain function. Human observers viewed initially blurry, unrecognizable natural object images presented at a fast 12 Hz rate and parametrically increasing in SF content over the course of 1 minute. By inserting highly variable natural face images as every 8th stimulus, we captured an objective neural index of face detection in participants' electroencephalogram (EEG) at exactly 1.5 Hz. This face-selective signal emerged over the right occipito-temporal cortex at <5 cycles/image, suggesting that the brain can - at a single glance - discriminate vastly different faces from multiple unsegmented object categories using only very coarse visual information. Local features (e.g., eyes) are not yet discernable at this threshold, indicating that fast face detection critically relies on global facial configuration. Interestingly, the face-selective neural response continued to increase with additional higher SF content until saturation around >50 cycles/image, potentially supporting higher-level recognition functions (e.g., facial identity recognition).

The orientation selectivity of face identification.

Recent work demonstrates that human face identification is most efficient when based on horizontal, rather than vertical, image structure. Because it is unclear how this specialization for upright (compared to inverted) face processing emerges in the visual system, the present study aimed to systematically characterize the orientation sensitivity profile for face identification. With upright faces, identification performance in a delayed match-to-sample task was highest for horizontally filtered images and declined sharply with oblique and vertically filtered images. Performance was well described by a Gaussian function with a bandwidth around 25°. Face inversion reshaped this sensitivity profile dramatically, with a downward shift of the entire tuning curve as well as a reduction in the amplitude of the horizontal peak and a doubling in bandwidth. The use of naturalistic outer contours (vs. a common outline mask) was also found to reshape this sensitivity profile by increasing sensitivity to oblique information in the near-horizontal range. Altogether, although face identification is sharply tuned to horizontal angles, both inversion and outline masking can profoundly reshape this orientation sensitivity profile. This combination of image- and observer-driven effects provides an insight into the functional relationship between orientation-selective processes within primary and high-level stages of the human brain.

Three-month-old infants' sensitivity to horizontal information within faces.

Horizontal information is crucial to face processing in adults. Yet the ontogeny of this preferential type of processing remains unknown. To clarify this issue, we tested 3-month-old infants' sensitivity to horizontal information within faces. Specifically, infants were exposed to the simultaneous presentation of a face and a car presented in upright or inverted orientation while their looking behavior was recorded. Face and car images were either broadband (UNF) or filtered to only reveal horizontal (H), vertical (V) or this combined information (HV). As expected, infants looked longer at upright faces than at upright cars, but critically, only when horizontal information was preserved in the stimulus (UNF, HV, H). These results first indicate that horizontal information already drives upright face processing at 3 months of age. They also recall the importance, for infants, of some facial features, arranged in a top-heavy configuration, particularly revealed by this band of information. © 2016 Wiley Periodicals, Inc. Dev Psychobiol 58: 536-542, 2016.

The impact of orientation filtering on face-selective neurons in monkey inferior temporal cortex.

Faces convey complex social signals to primates. These signals are tolerant of some image transformations (e.g. changes in size) but not others (e.g. picture-plane rotation). By filtering face stimuli for orientation content, studies of human behavior and brain responses have shown that face processing is tuned to selective orientation ranges. In the present study, for the first time, we recorded the responses of face-selective neurons in monkey inferior temporal (IT) cortex to intact and scrambled faces that were filtered to selectively preserve horizontal or vertical information. Guided by functional maps, we recorded neurons in the lateral middle patch (ML), the lateral anterior patch (AL), and an additional region located outside of the functionally defined face-patches (CONTROL). We found that neurons in ML preferred horizontal-passed faces over their vertical-passed counterparts. Neurons in AL, however, had a preference for vertical-passed faces, while neurons in CONTROL had no systematic preference. Importantly, orientation filtering did not modulate the firing rate of neurons to phase-scrambled face stimuli in any recording region. Together these results suggest that face-selective neurons found in the face-selective patches are differentially tuned to orientation content, with horizontal tuning in area ML and vertical tuning in area AL.

Horizontal tuning for faces originates in high-level Fusiform Face Area.

Recent work indicates that the specialization of face visual perception relies on the privileged processing of horizontal angles of facial information. This suggests that stimulus properties assumed to be fully resolved in primary visual cortex (V1; e.g., orientation) in fact determine human vision until high-level stages of processing. To address this hypothesis, the present fMRI study explored the orientation sensitivity of V1 and high-level face-specialized ventral regions such as the Occipital Face Area (OFA) and Fusiform Face Area (FFA) to different angles of face information. Participants viewed face images filtered to retain information at horizontal, vertical or oblique angles. Filtered images were viewed upright, inverted and (phase-)scrambled. FFA responded most strongly to the horizontal range of upright face information; its activation pattern reliably separated horizontal from oblique ranges, but only when faces were upright. Moreover, activation patterns induced in the right FFA and the OFA by upright and inverted faces could only be separated based on horizontal information. This indicates that the specialized processing of upright face information in the OFA and FFA essentially relies on the encoding of horizontal facial cues. This pattern was not passively inherited from V1, which was found to respond less strongly to horizontal than other orientations likely due to adaptive whitening. Moreover, we found that orientation decoding accuracy in V1 was impaired for stimuli containing no meaningful shape. By showing that primary coding in V1 is influenced by high-order stimulus structure and that high-level processing is tuned to selective ranges of primary information, the present work suggests that primary and high-level levels of the visual system interact in order to modulate the processing of certain ranges of primary information depending on their relevance with respect to the stimulus and task at hand.

Inhibition of return and attentional facilitation: Numbers can be counted in, letters tell a different story.

Prior research has provided strong evidence for spatial-numerical associations. Single digits can for instance act as attentional cues, orienting visuo-spatial attention to the left or right hemifield depending on the digit's magnitude, thus facilitating target detection in the cued hemifield (left/right hemifield after small/large digits, respectively). Studies using other types of behaviourally or biologically relevant central cues known to elicit automated symbolic attention orienting effects such as arrows or gaze have shown that the initial facilitation of cued target detection can turn into inhibition at longer stimulus onset asynchronies (SOAs). However, no studies so far investigated whether inhibition of return (IOR) is also observed using digits as uninformative central cues. To address this issue we designed an attentional cueing paradigm using SOAs ranging from 500 ms to 1650 ms. As expected, the results showed a facilitation effect at the relatively short 650 ms SOA, replicating previous findings. At the long 1650 ms SOA, however, participants were faster to detect targets in the uncued hemifield compared to the cued hemifield, showing an IOR effect. A control experiment with letters showed no such congruency effects at any SOA. These findings provide the first evidence that digits not only produce facilitation effects at shorter intervals, but also induce inhibitory effects at longer intervals, confirming that Arabic digits engage automated symbolic orienting of attention.

Selectivity of Face Perception to Horizontal Information over Lifespan (from 6 to 74 Year Old).

Face recognition in young human adults preferentially relies on the processing of horizontally-oriented visual information. We addressed whether the horizontal tuning of face perception is modulated by the extensive experience humans acquire with faces over the lifespan, or whether it reflects an invariable processing bias for this visual category. We tested 296 subjects aged from 6 to 74 years in a face matching task. Stimuli were upright and inverted faces filtered to preserve information in the horizontal or vertical orientation, or both (HV) ranges. The reliance on face-specific processing was inferred based on the face inversion effect (FIE). FIE size increased linearly until young adulthood in the horizontal but not the vertical orientation range of face information. These findings indicate that the protracted specialization of the face processing system relies on the extensive experience humans acquire at encoding the horizontal information conveyed by upright faces.