Parallel discrimination of subjective contours defined by offset gratings

Perception of illusory contours in children and adults: An eye-tracking study

The eye-tracking study investigated the perception of subjective Kanizsa and Ehrenstein figures in adults and in children aged 3-4, 5-6, 7-8, and 9-11 years of age. More specifically, the distribution of looking at the inner stimulus part versus the inducing elements was measured for illusory figures, figures with real contours, and control displays. It was hypothesized that longer looking at the inner area of the illusory figures indicates global contour interpolation, whereas longer looking at the inducing elements indicates a local processing mode. According to the results, participants of all ages looked longer at the illusory Kanizsa and Ehrenstein contours than at the figures' inducing elements. However, performance was lowest in the children aged 3-4 years and increased during the preschool period. Moreover, the illusory contour displays elicited comparable visual responses as did the real contour displays. The use of the control displays that contained no contour information ensured that the participants' looking behavior was not driven by a spontaneous tendency to attend to the inner stimulus parts. The study confirms the view that sensitivity to illusory contours emerges very early in life.

Challenging deep learning models with image distortion based on the abutting grating illusion

Even state-of-the-art deep learning models lack fundamental abilities compared with humans. While many image distortions have been proposed to compare deep learning with humans, they depend on mathematical transformations instead of human cognitive functions. Here, we propose an image distortion based on the abutting grating illusion, which is a phenomenon discovered in humans and animals. The distortion generates illusory contour perception using line gratings abutting each other. We applied the method to MNIST, high-resolution MNIST, and "16-class-ImageNet" silhouettes. Many models, including models trained from scratch and 109 models pretrained with ImageNet or various data augmentation techniques, were tested. Our results show that abutting grating distortion is challenging even for state-of-the-art deep learning models. We discovered that DeepAugment models outperformed other pretrained models. Visualization of early layers indicates that better-performing models exhibit the endstopping property, which is consistent with neuroscience discoveries. Twenty-four human subjects classified distorted samples to validate the distortion.

Asymmetric visual representation of sex from human body shape

We efficiently infer others' states and traits from their appearance, and these inferences powerfully shape our social behaviour. One key trait is sex, which is strongly cued by the appearance of the body. What are the visual representations that link body shape to sex? Previous studies of visual sex judgment tasks find observers have a bias to report "male", particularly for ambiguous stimuli. This finding implies a representational asymmetry - that for the processes that generate a sex percept, the default output is "male", and "female" is determined by the presence of additional perceptual evidence. That is, female body shapes are positively coded by reference to a male default shape. This perspective makes a novel prediction in line with Treisman's studies of visual search asymmetries: female body targets should be more readily detected amongst male distractors than vice versa. Across 10 experiments (N = 32 each) we confirmed this prediction and ruled out alternative low-level explanations. The asymmetry was found with profile and frontal body silhouettes, frontal photographs, and schematised icons. Low-level confounds were controlled by balancing silhouette images for size and homogeneity, and by matching physical properties of photographs. The female advantage was nulled for inverted icons, but intact for inverted photographs, suggesting reliance on distinct cues to sex for different body depictions. Together, these findings demonstrate a principle of the perceptual coding that links bodily appearance with a significant social trait: the female body shape is coded as an extension of a male default. We conclude by offering a visual experience account of how these asymmetric representations arise in the first place.

Visual search does not always predict performance in tasks that require finding targets among distractors: The case of line-ending illusory contours

The standard visual search task is integral to the study of selective attention and in search tasks target present slopes are the primary index of attentional demand. However, there are times when similarities in slopes may obscure important differences between conditions. To demonstrate this point, we used the case of line-ending illusory contours, building on a study by Li, Cave, and Wolfe (2008) where orientation-based search for figures defined by line-ending illusory contours was compared to that for the corresponding real-contour controls. Consistent with Li et al. (2008), we found search to be efficient for both illusory contour figures and the corresponding real-contour controls, with no significant differences between them. However, major differences between illusory contours and the real-contour controls emerged in selective enumeration, a task where participants enumerated targets in a display of distractors, with the number of targets and distractors manipulated. When looking at the distractor slopes, the increase in RT to enumerate a single target as a function of the number of distractors (a direct analogue to target present trials, with identical displays), we found distractor costs for illusory contour figures to be over 100 ms/distractor higher than for the corresponding real-contour controls. Furthermore, the discrepancies in RT slope between 1-3 and 6-8 targets associated with subitizing were only seen in the real-contour controls. These results show that similarities in RT slopes in search may mask important differences between conditions that emerge in other tasks.

Global attention facilitates the planning, but not execution of goal-directed reaches

In daily life, humans interact with multiple objects in complex environments. A large body of literature demonstrates that target selection is biased toward recently attended features, such that reaches are faster and trajectory curvature is reduced when target features (i.e., color) are repeated (priming of pop-out). In the real world, however, objects are comprised of several features—some of which may be more suitable for action than others. When fetching a mug from the cupboard, for example, attention not only has to be allocated to the object, but also the handle. To date, no study has investigated the impact of hierarchical feature organization on target selection for action. Here, we employed a color-oddity search task in which targets were Pac-men (i.e., a circle with a triangle cut out) oriented to be either consistent or inconsistent with the percept of a global Kanizsa triangle. We found that reaches were initiated faster when a task-irrelevant illusory figure was present independent of color repetition. Additionally, consistent with priming of pop-out, both reach planning and execution were facilitated when local target colors were repeated, regardless of whether a global figure was present. We also demonstrated that figures defined by illusory, but not real contours, afforded an early target selection benefit. In sum, these findings suggest that when local targets are perceptually grouped to form an illusory surface, attention quickly spreads across the global figure and facilitates the early stage of reach planning, but not execution. In contrast, local color priming is evident throughout goal-directed reaching.

Access to Global and Local Properties in Visual Search for Compound Stimuli

The question of whether attention is drawn more easily to global or local aspects of a stimulus has been debated for more than 100 years We examined it anew, using the visual search task, which distinguishes sensory from attentional effects Subjects searched for a target feature (e g, triangle vs square), which was equally likely to occur in the local elements of a compound search item, in its global structure, or in both Element size and spacing were used to manipulate whether search was generally easier for local or global targets (e g, small size and dense spacing favor global detection) The novel result was that these factors had very little influence on search slopes for local targets, whereas they had large effects on search slopes for global targets This result suggests that a qualitatively different process underlies detection at the global level in traditional compound stimuli Our proposal that an attention-demanding grouping stage is involved was confirmed in a final experiment in which grouping was made selectively difficult at the local level

Texture discriminability in monkey inferotemporal cortex predicts human texture perception

Shape and texture are both important properties of visual objects, but texture is relatively less understood. Here, we characterized neuronal responses to discrete textures in monkey inferotemporal (IT) cortex and asked whether they can explain classic findings in human texture perception. We focused on three classic findings on texture discrimination: 1) it can be easy or hard depending on the constituent elements; 2) it can have asymmetries, and 3) it is reduced for textures with randomly oriented elements. We recorded neuronal activity from monkey inferotemporal (IT) cortex and measured texture perception in humans for a variety of textures. Our main findings are as follows: 1) IT neurons show congruent selectivity for textures across array size; 2) textures that were easy for humans to discriminate also elicited distinct patterns of neuronal activity in monkey IT; 3) texture pairs with asymmetries in humans also exhibited asymmetric variation in firing rate across monkey IT; and 4) neuronal responses to randomly oriented textures were explained by an average of responses to homogeneous textures, which rendered them less discriminable. The reduction in discriminability of monkey IT neurons predicted the reduced discriminability in humans during texture discrimination. Taken together, our results suggest that texture perception in humans is likely based on neuronal representations similar to those in monkey IT.

Guided Search 2.0 A revised model of visual search

An important component of routine visual behavior is the ability to find one item in a visual world filled with other, distracting items. This ability to performvisual search has been the subject of a large body of research in the past 15 years. This paper reviews the visual search literature and presents a model of human search behavior. Built upon the work of Neisser, Treisman, Julesz, and others, the model distinguishes between a preattentive, massively parallel stage that processes information about basic visual features (color, motion, various depth cues, etc.) across large portions of the visual field and a subsequent limited-capacity stage that performs other, more complex operations (e.g., face recognition, reading, object identification) over a limited portion of the visual field. The spatial deployment of the limited-capacity process is under attentional control. The heart of the guided search model is the idea that attentional deployment of limited resources isguided by the output of the earlier parallel processes. Guided Search 2.0 (GS2) is a revision of the model in which virtually all aspects of the model have been made more explicit and/or revised in light of new data. The paper is organized into four parts: Part 1 presents the model and the details of its computer simulation. Part 2 reviews the visual search literature on preattentive processing of basic features and shows how the GS2 simulation reproduces those results. Part 3 reviews the literature on the attentional deployment of limited-capacity processes in conjunction and serial searches and shows how the simulation handles those conditions. Finally, Part 4 deals with shortcomings of the model and unresolved issues.

Evidence of unlimited-capacity surface completion

Capacity limitations of perceptual surface completion were assessed using a simultaneous-sequential method. Observers searched among multiple surfaces requiring perceptual completion in front of other objects (modal completion) or behind other objects (amodal completion). In the simultaneous condition, all surfaces were presented at once, whereas in the sequential condition, they appeared in subsets of 2 at a time. For both modal and amodal surface completion, performance was as good in the simultaneous condition as in the sequential condition, indicating that surface completion unfolds independently for multiple surfaces across the visual field (i.e., has unlimited capacity). We confirmed this was due to the formation of surfaces defined by the pacmen inducers, and not simply to the detection of individual features of the pacmen inducers. These results provide evidence that surface-completion processes can be engaged and unfold independently for multiple surfaces across the visual field. In other words, surface completion can occur through unlimited-capacity processes. These results contribute to a developing understanding of capacity limitations in perceptual processing more generally.

How does our search engine "see" the world? The case of amodal completion

This article illustrates a dissociation between the perceived attributes of an object and the ability of those attributes to guide the deployment of attention in visual search. Orientation is an attribute that guides search. Thus, a vertical line will "pop out" amid horizontal distractors. Amodal completion can create perceptually convincing oriented stimuli when two elements appear to form a complete object partially hidden behind an occluder. Previous work (e.g., Rensink & Enns, Vision Research, 38, 2489-2505, 1998) has shown a preattentive role for amodal completion in search tasks. Here, we show that orientation based on perceptually compelling amodal completion may fail to guide attention. The broader conclusion is that introspection is a poor guide to the capabilities of our internal search engine.

Fur in the midst of the waters: visual search for material type is inefficient

A limited set of attributes can guide visual selective attention. Thus, it is possible to deploy attention to an item defined by an appropriate color, size, or orientation but not to a specific type of line intersection or a specific letter (assuming other attributes like orientation are controlled). What defines the set of guiding attributes? Perhaps it is the set of attributes of surfaces or materials in the world. L. Sharan, R. E. Rosenholtz, and E. H. Adelson (submitted for publication) have shown that observers are extremely adept at identifying materials. Are they equally adept at guiding attention to one type of material among distractors of another? A series of visual search experiments shows that the answer is "no." It may be easy to identify "fur" or "stone," but search for a patch of fur among the stones will be inefficient.

The Computation of Shape Orientation in Search for Kanizsa Figures

Previous studies of visual search for illusory figures have provided equivocal results, with efficient search for Kanizsa squares (eg Davis and Driver, 1994 Nature371 291–293) contrasting with inefficient search for Kanizsa triangles (eg Grabowecky and Treisman, 1989 Investigative Ophthalmology and Visual Science30 457). Here, we investigated whether shape orientation can explain these differences. The results from three experiments replicated previous findings: Kanizsa squares in experiment 1 could be detected more efficiently than Kanizsa triangles in experiment 2. In addition, when controlling for stimulus complexity in experiment 3, we found search for Kanizsa diamonds intermediate in efficiency. Taken together, these results suggest an oblique effect in search for Kanizsa figures with cardinal shape orientations leading to more efficient performance than oblique shape orientations. Our findings indicate that both shape orientation and stimulus complexity affect search for illusory figures.

The contrasting impact of global and local object attributes on Kanizsa figure detection

Studies on the involvement of object completions in search for illusory figures have so far reported equivocal results. We have addressed this issue by investigating at which level object attributes in Kanizsa figures influence search. Employing a paradigm that investigated global and local attributes in the composition of distractors with relation to target composition, we report a selective involvement of multilevel processing upon detection. Four experiments demonstrate that global surface information, but not the surrounding global contour, determines the speed of Kanizsa figure detection. By contrast, local inducer information is encoded far less efficiently in search than processes computing the global object. Our conclusions are that surface filling-in acts as a major determinant of search, but depends on the relevance of the particular hierarchical level (local or global) coding the target.

Orientation-selective adaptation to illusory contours in human visual cortex

Humans can perceive illusory or subjective contours in the absence of any real physical boundaries. We used an adaptation protocol to look for orientation-selective neural responses to illusory contours defined by phase-shifted abutting line gratings in the human visual cortex. We measured functional magnetic resonance imaging (fMRI) responses to illusory-contour test stimuli after adapting to an illusory-contour adapter stimulus that was oriented parallel or orthogonal to the test stimulus. We found orientation-selective adaptation to illusory contours in early (V1 and V2) and higher-tier visual areas (V3, hV4, VO1, V3A/B, V7, LO1, and LO2). That is, fMRI responses were smaller for test stimuli parallel to the adapter than for test stimuli orthogonal to the adapter. In two control experiments using spatially jittered and phase-randomized stimuli, we demonstrated that this adaptation was not just in response to differences in the distribution of spectral power in the stimuli. Orientation-selective adaptation to illusory contours increased from early to higher-tier visual areas. Thus, both early and higher-tier visual areas contain neurons selective for the orientation of this type of illusory contour.

Electrophysiological Correlates of Similarity-based Interference during Detection of Visual Forms

Illusory figure completion demonstrates the ability of the visual system to integrate information across gaps. Mechanisms that underlie figural emergence support the interpolation of contours and the filling-in of form information [Grossberg, S., & Mingolla, E. Neural dynamics of form perception: Boundary completion, illusory figures and neon colour spreading. Psychological Review, 92, 173–211, 1985]. Although both processes contribute to figure formation, visual search for an illusory target configuration has been shown to be susceptible to interfering form, but not contour, information [Conci, M., Müller, H. J., & Elliott, M. A. The contrasting impact of global and local object attributes on Kanizsa figure detection. Submitted]. Here, the physiological basis of form interference was investigated by recording event-related potentials elicited from contour- and surface-based distracter interactions with detection of a target Kanizsa figure. The results replicated the finding of form interference and revealed selection of the target and successful suppression of the irrelevant distracter to be reflected by amplitude differences in the N2pc component (240–340 msec). In conclusion, the observed component variations reflect processes of target selection on the basis of integrated form information resulting from figural completion processes.

Functional neuroimaging findings on the human perception of illusory contours

Illusory contours (IC) have attracted a considerable interest in recent years to derive models of how sensory information is processed and integrated within the visual system. In addition to various findings from neuropsychology, neurophysiology, and psychophysics, several recent studies have used functional neuroimaging to identify the cerebral substrates underlying human perception of IC (in particular Kanizsa figures). In this paper, we review the results from more than 20 neuroimaging studies on IC perception and highlight the great diversity of findings across these studies. We then provide a detailed discussion about the localization ('where' debate) and the timing ('when' debate) of IC processing as suggested by functional neuroimaging. Cortical responses involving visual areas as early as V1/V2 and latencies as rapid as 100 ms have been reported in several studies. Particular issues concerning the role of the right hemisphere and the retinotopic encoding of IC are also discussed. These different findings are tentatively brought together to propose different hypothetical cortical mechanisms that might be responsible for the visual formation of IC. Several remaining questions on IC processing that could potentially be explored with functional neuroimaging techniques are finally emphasized.

Recovering depth-order from orientation-defined junctions

This study investigated how visual systems recover depth-order from orientation-defined junctions. Stimuli were superimposed stripes defined by Gabor micro-patterns (Gabors). In one stripe (random stripe), Gabor orientation was randomly selected from a given range, while in the other (constant stripe) it was selected so as to be different from the mean orientation of the random stripe by 90 degrees . Observers reported which of the two stripes, the right- or left-tilted one, they perceived as "nearer" than the other. Observers frequently reported that the random stripe was nearer than the constant stripe. The results appeared to stem from detection of discontinuity of texture edges of the constant stripe due to masking by the random stripe at junctions. This idea was confirmed in the following experiments where discontinuity of the texture edges at junctions was introduced by changing the Gabor luminance contrast in one stripe but keeping it intact in the other. The results indicated that processing of texture edges at junctions can contribute to the perception of depth-order.

Kanizsa subjective figures capture visual spatial attention: evidence from electrophysiological and behavioral data

Figural binding and attention are two important processes that help to perceive the outside world. Binding is necessary to link together the different features of single objects which are represented in a distributed fashion in the brain. Attention serves to focus onto a small subset of incoming information. It is still not clear how exactly these two mechanisms operate and interact. We performed two experiments employing illusory Kanizsa figures (KFs) to investigate the temporal order of figural binding and spatial attention. In a visual search task, subjects had to detect the presence of a KF among distractor stimuli. We found only a slight increase of reaction times when increasing the number of distractors, indicating that KFs popped out and drew the perceiver's attention. In a further event-related potential (ERP) study, we used displays of the search task as non-informative cue for a subsequent target choice-reaction task. Enhanced contralateral negative amplitudes (starting at about 230 ms) over ventral occipital areas were found for cue displays which included a KF. For target stimuli, faster reaction times and enhanced ipsilateral N1 amplitudes over occipito-parietal areas were observed for validly (target presentation inside a KF) as compared to invalidly cued targets (target presentation outside a KF). Furthermore, enhanced contralateral N1 amplitudes were found for invalidly cued targets. It might be that interactions between perceptual closure processing of the ventral pathway and spatial target processing of the dorsal pathway contributed to the present result. We conclude that KFs automatically capture spatial attention when used as visual cues.

Surface segregation influences pre-attentive search in depth

While searching for an object in a cluttered scene, in some situations, the visual system adopts a pre-attentive parallel search, where the time taken is independent of the number of items in the scene. In others, the search is serial, time taken being a function of the set size. We show that detecting the number of targets (2, 3 or 4) that differ in depth from background items is a parallel process, but only when they are all in the same surface in depth. The search is serial if the targets are in different depth planes, but parallel even if the targets were on a surface tilted in depth, showing that surface segregation influences a parallel, apparently pre-attentive, stage.

Uniformity and asymmetry of rapid curved-line detection explained by parallel categorical coding of contour curvature

The aim of this work was to elucidate several characteristic phenomena associated with rapid curved-line detection in multi-element arrays and to provide a unified account of the underlying curvature-sensitive mechanisms. To this end, a parametric experiment was performed in which the detectability of a curved-line target in a briefly presented planar array of curved-line distractors was measured for a range of target and distractor curvatures and distractor numbers. For both vertically oriented and randomly oriented curved lines, it was found that (1) the dependence of target detectability on target curvature was independent of distractor number for small distractor curvatures but not for medium-to-large distractor curvatures; (2) an asymmetry in target detectability with respect to interchange of target and distractor curvatures occurred only with large distractor numbers; and (3) with small distractor numbers, target detectability depended only on the difference between target and distractor curvatures. These properties of spatial parallelism, asymmetry, and uniformity were explained quantitatively by a minimal model of rapid curved-line detection in which contour curvature was coded in terms of just two or three curvature categories, depending on curved-line orientation.

Magnetic responses of human visual cortex to illusory contours

To examine the neural mechanism underlying illusory-contour perception, we measured the magnetic responses of the human visual cortex to an abutting-line grating inducing illusory contours (test stimulus) and a non-abutting-line grating (control stimulus) using the technique of magnetoencephalography (MEG). In the initial latency period of 60-80 ms, the MEG response to the test stimulus was nearly identical with that to the control stimulus, but in the subsequent period of 80-150 ms, the former was larger than the latter. The origin of the peak MEG response to the test stimulus was estimated to be in the vicinity of striate cortex/extrastriate visual cortex for two of the four subjects. These results suggest that, in accord with those of the previous electrophysiological and functional magnetic resonance imaging studies, illusory-contour signals are generated in the very early stage(s) of processing in the primate visual cortex.

Modulations of the processing of line discontinuities under selective attention conditions?

We examined whether the processing of discontinuities involved in figure-ground segmentation, like line ends, can be modulated under selective attention conditions. Subjects decided whether a gap in collinear or parallel lines was located to the right or left. Two stimuli were displayed in immediate succession. When the gaps were on the same side, reaction times (RTs) for the second stimulus increased when collinear lines followed parallel lines, or the reverse, but only when the two stimuli shared the same orientation and location. The effect did not depend on the global form of the stimuli or on the relative orientation of the gaps. A frame drawn around collinear elements affected the results, suggesting a crucial role of the "amodal" orthogonal lines produced when line ends are aligned. Including several gaps in the first stimulus also eliminated RT variations. By contrast, RT variations remained stable across several experimental blocks and were significant for interstimulus intervals from 50 to 600 msec between the two stimuli. These results are interpreted in terms of a modulation of the processing of line ends or the production of amodal lines, arising when attention is selectively drawn to a gap.

Visual search asymmetries in motion and optic flow fields

In visual search, items defined by a unique feature are found easily and efficiently. Search for a moving target among stationary distractors is one such efficient search. Search for a stationary target among moving distractors is markedly more difficult. In the experiments reported here, we confirm this finding and further show that searches for a stationary target within a structured flow field are more efficient than searches for stationary targets among distractors moving in random directions. The structured motion fields tested included uniform direction of motion, a radial flow field simulating observer forward motion, and a deformation flow field inconsistent with observer motion. The results using optic flow stimuli were not significantly different from the results obtained with other structured fields of distractors. The results suggest that the local properties of the flow fields rather than global optic flow properties are important for determining the efficiency of search for a stationary target.

Neuronal correlates of real and illusory contour perception: functional anatomy with PET

Illusory contours provide a striking example of the visual system's ability to extract a meaningful representation of the surroundings from fragmented visual stimuli. Psychophysical and neurophysiological data suggest that illusory contours are processed in early visual cortical areas, and neuroimaging studies in humans have shown that Kanizsa-type illusory contours activate early retinotopic visual areas that are also activated by real contours. It is not known whether other types of illusory contours are processed by the same mechanisms, nor is it clear to what extent attentional effects may have influenced these results, as no attempt was made to match the salience of real and illusory stimuli in previous imaging studies. It therefore remains an open question whether there are any brain regions specifically involved in the perception of illusory contours. To address these questions, we have used 15O-butanol positron emission tomography (PET) and a novel kind of illusory contour stimulus that is induced only by aligned line ends. By employing a form discrimination task that was matched for attention and stimulus salience across conditions we were able to directly contrast perception of real and illusory contours. We found that the regions activated by illusory contour perception were the same as those activated by real contours. Only one region, located in the right fusiform gyrus, was significantly more strongly activated by perception of illusory contours than by real contours. In addition, a principal component analysis suggested that illusory contour perception is associated with a change in the correlation between V1 and V2. We conclude that different kinds of illusory contours are processed by the same cortical regions and that these regions overlap extensively with those involved in processing of real contours. At the regional level, perception of illusory contours thus appears to differ from perception of real contours by the degree of involvement of higher visual areas as well as by the nature of interaction between early visual areas.

Detection and discrimination of subjective contours defined by offset gratings

Three experiments were conducted to refine our understanding of the mechanisms that encode subjective contours. In Experiment 1, discrimination thresholds (stimulus onset asynchronies [SOAs] yielding 81% correct) were measured in a backward masking paradigm for subjective contours defined by offset gratings. For large apertures, thresholds increased as carrier frequency increased. For the smallest aperture, thresholds were a U-shaped function of carrier frequency. Experiment 2 showed that these threshold results were generally consistent with the rated strength of the subjective contours. Experiment 3 showed that detection thresholds (SOAs yielding 81% correct) again increased with carrier spatial frequency, increased for obliquely oriented carriers, and, for a particular frequency and orientation of the carrier, were lower when the subjective contour was orthogonal to the carrier. All of these results are well explained by a two-stage process in which a second-layer filter integrates the responses of end-stopped mechanisms to the terminators defining the subjective contour. In the model, the end-stopped mechanisms have low-pass sensitivity to carrier spatial frequency, and the sizes of the second-layer filters are proportional to the scale of the end-stopped mechanisms from which they draw their input.

Bilateral symmetry embedded in noise is detected accurately only at fixation

Bilateral or mirror symmetry is a ubiquitous feature of biological forms that the visual system could exploit for segmenting an object from a cluttered background. If this is so, the visual system may be prepared to detect symmetry at all retinal locations in parallel. Indeed, a biologically plausible model that responds optimally at axes of symmetry is quite easy to construct. Our data show, however, that if such a mechanism exists, it works with high efficiency only at the fovea. The detection of vertical bilateral symmetry embedded in random noise is very poor unless the axis of symmetry is very close to the point of fixation. This leads to the conclusion that symmetry does not play an important role in image segmentation and that it is important to the visual system only after it is fixated.

Illusory contours and spatial neglect

It has been controversial whether the perception of illusory contours arise from higher level cognitive mechanisms that require attention or from early preattentive visual processes. We studied three patients with left spatial neglect who were unable to detect the left inducers of Kanizsa illusory figures in a same/different judgment task but nonetheless showed implicit perception of the figures in a midpoint judgment, in that they made identical bisection for figures with illusory or real contours but very different bisection for other spatially discontinuous figures that did not yield illusory filling-in. Grouping and filling-in mechanisms can thus occur without explicit detection of, or attention to, the inducing features, consistently with the hypothesis that they involve preattentive visual processes.

Retention of local information in generation of subjective contours

Temporal integration characteristics of subjective contour perception was investigated, using the sequential presentation of two pairs of disks with a sector removed. In the first experiment, by matching the contrast of a "real" stimulus, the perceived contrast of the subjective contours was measured as a function of the stimulus onset asynchrony (SOA) between the two pairs of the disks. With increasing SOA, the perceived contrast decreased gradually and levelled off at the SOA of ca 372 msec (1 SD = 119 msec). In the second experiment, the perceived contrast of the inducing disks was measured as a function of SOA using the matching method. The time limit of the additive effect for the contrast perception of the inducing disks was much shorter than that for subjective contours; the critical SOA was ca 65 msec (1 SD = 36 msec). The remarkable difference of the integration time was explained by a hierarchical process; the local spatial filtering, the retention of local information, and the completion of gaps by multiplicative or AND operation.

Illusory contour-motion arising from translating terminators

Periodic grating patterns were created by phase shifting or eliminating vertical columns of a fine line carrier grating oriented 45 deg. Motion was created by translating the patterns parallel to the carrier grating. This veridical motion was seen when terminators (i) were created in low-frequency carriers; (ii) terminated short lines; and (iii) moved slowly. In the complementary conditions an illusory contour-motion was seen perpendicular to the orientation of the terminator-defined contours. A model involving a competition between second-layer filters (encoding the orientation and motions of the terminator defined contours) and double endstopped mechanisms (signalling the presence of terminators) was developed and found to be in quantitative agreement with these data. Experiments with plaids composed of two such patterns were generally consistent with the results of the one-dimensional cases. Coherent "subjective contour plaid" motion was almost always seen when the two subjective contours had the same orientation and were perfectly phase aligned.

Preattentive object files: shapeless bundles of basic features

A considerable body of recent evidence shows that preattentive processes can carve visual input into candidate objects. Borrowing and modifying terminology from Kahneman & Treisman (1984), this paper investigates the properties of these preattentive object files. Experiments 1-3 show that preattentive object files are loose collections of basic features. Thus, we can know preattentively that an object has the attributes "red" and "vertical" and yet have no idea if any part of the object is red and vertical. Experiment 4 shows that some information about the structure of an object is available preattentively, but Experiments 5-12 search for and fail to find any preattentive representation of overall shape. Appreciation of the overall shape of an object appears to require the binding together of local form features--a process that requires attention.

There is no evidence that Kanizsa-type subjective contours can be detected in parallel

Davis and Driver presented evidence suggesting that Kanizsa-type subjective contours could be detected in a visual search task in a time that is independent of the number of nonsubjective contour distractors. A linking connection was made between these psychophysical data and the physiological data of Peterhans and von der Heydt which showed that cells in primate area V2 respond to subjective contours in the same way that they respond to luminance-defined contours. Here in three experiments it is shown that there was sufficient information in the displays used by Davis and Driver to support parallel search independently of whether subjective contours were present or not. When confounding properties of the stimuli were eliminated search became slow whether or not subjective contours were present in the display. One of the slowest search conditions involved stimuli that were virtually identical to those used in the physiological studies of Peterhans and von der Heydt to which Davis and Driver wish to link their data. It is concluded that while subjective contours may be represented in the responses of very early visual mechanisms (eg in V2) access to these representations is impaired by high-contrast contours used to induce the subjective contours and nonsubjective figure distractors. This persistent control problem continues to confound attempts to show that Kanizsa-type subjective contours can be detected in parallel.

The abutting grating illusion

Two line gratings abutting each other with a phase shift of half a cycle elicit the perception of an illusory line running orthogonally between the two sets of grating lines. We found that rating strength increases with increasing number of lines, line length, and phase angle. In contrast, rating strength decreases with increasing spacing of lines, lateral misalignment, rotation of one grating relative to the other, and line width. There is a pronounced oblique effect at 45 deg when the orientation of the abutting gratings is changed from horizontal through diagonal to vertical. Findings are interpreted in terms of a neurophysiological model. We conclude that the end-stopped receptive fields activated by the grating lines are about 6 deg long and 2 deg wide. On the other hand, the "response fields" of the cells, integrating orthogonally across line ends, are assumed to be 5 deg long and less than 1 deg wide. The psychophysical data compare favorably with available neurophysiological data in Area V2 of the macaque suggesting that the perception of illusory contours in human observers may be based on cortical cell properties similar to those found in the monkey.

Phenomena of illusory form: can we bridge the gap between levels of explanation?

The study of illusory brightness and contour phenomena has become an important tool in modern brain research. Gestalt, cognitive, neural, and computational approaches are reviewed and their explanatory powers are discussed in the light of empirical data. Two well-known phenomena of illusory form are dealt with, the Ehrenstein illusion and the Kanizsa triangle. It is argued that the gap between the different levels of explanation, bottom-up versus top-down, creates scientific barriers which have all too often engendered unnecessary debate about who is right and who is wrong. In this review of the literature we favour an integrative approach to the question of how illusory form is derived from stimulus configuration which provide the visual system with seemingly incomplete information. The processes that can explain the emergence of these phenomena range from local feature detection to global strategies of perceptual organisation. These processes may be similar to those that help us restore partially occluded objects in everyday vision. To understand better the Ehrenstein and Kanizsa illusions, it is proposed that different levels of analysis and explanation are not mutually exclusive, but complementary. Theories of illusory contour and form perception must, therefore, take into account the underlying neurophysiological mechanisms and their possible interactions with cognitive and attentional processes.

Object recognition contributions to figure-ground organization: operations on outlines and subjective contours

In previous research, replicated here, we found that some object recognition processes influence figure-ground organization. We have proposed that these object recognition processes operate on edges (or contours) detected early in visual processing, rather than on regions. Consistent with this proposal, influences from object recognition on figure-ground organization were previously observed in both pictures and stereograms depicting regions of different luminance, but not in random-dot stereograms, where edges arise late in processing (Peterson & Gibson, 1993). In the present experiments, we examined whether or not two other types of contours--outlines and subjective contours--enable object recognition influences on figure-ground organization. For both types of contours we observed a pattern of effects similar to that originally obtained with luminance edges. The results of these experiments are valuable for distinguishing between alternative views of the mechanisms mediating object recognition influences on figure-ground organization. In addition, in both Experiments 1 and 2, fixated regions were seen as figure longer than nonfixated regions, suggesting that fixation location must be included among the variables relevant to figure-ground organization.

Parallel detection of Kanizsa subjective figures in the human visual system

Subjective figures, seen in the absence of luminance gradients (Fig. 1), provide a phenomenal illusion that can be related to the properties of single cells in the visual cortex, offering a rare bridge between brain function and visual awareness. It remains controversial whether subjective figures arise from intelligent cognitive mechanisms, or from lower-level processes in early vision. The cognitive account implies that the perception of subjective figures may require serial attentive processing, whereas on the low-level account they should arise in parallel at earlier visual stages. Physiological evidence apparently fits a low-level account and indicates that some types of subjective contour may be detected earlier than the conventional Kanizsa type. Here we report that even Kanizsa subjective figures can be detected without focal attention at parallel stages of the human visual system.