Functional Neuroimaging Findings on the Human Perception of Illusory

Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 www.elsevier.com/locate/neubiorev Review Functional neuroimaging ﬁndings on the human perception of illusory contours

M.L. Seghier a,b,*, P. Vuilleumier a,c

a Laboratory for Neurology and Imaging of Cognition, Clinic of Neurology and Department of Neurosciences, University Medical Center of Geneva, Michel-Servet 1, Geneva 1211, Switzerland b Department of Radiology, University Hospital of Geneva (HUG), Micheli-du-Crest 24, 1211 Geneva, Switzerland c Department of Psychology, University of Geneva, Geneva, Switzerland

Received 31 January 2005; received in revised form 14 September 2005; accepted 21 November 2005

Abstract

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. q 2005 Elsevier Ltd. All rights reserved.

Keywords: Illusory contours; Kanizsa ﬁgure; Functional neuroimaging; Visual areas; V1; V2; LOC; Segmentation and grouping mechanisms; Feedback connections; Low and high level vision

Contents

1. Introduction ...... 596 2. Methodological issues ...... 598 2.1. Imaging techniques ...... 598 2.2. Data analysis methods ...... 598 2.3. The subject factor ...... 599 2.4. Stimuli ...... 599 3. IC processing: The ‘where’ debate ...... 600 3.1. V1/V2 areas ...... 600 3.2. The lateral occipital complex (LOC) ...... 602 3.3. Other cortical regions ...... 603 3.4. Retinotopy of IC ...... 603 3.5. IC perception and the right hemisphere ...... 604 4. IC processing: The ‘when’ debate ...... 604

* Corresponding author. Address: Department of Radiology, University Hospital of Geneva (HUG), Micheli-du-Crest 24, 1211 Geneva, Switzerland. Tel.: C41 22 379 5361; fax: C41 22 372 7072. E-mail address: [email protected] (M.L. Seghier).

5. Possible mechanisms inferred from neuroimaging ﬁndings ...... 605 5.1. The fast-local low-level mechanism ...... 605 5.2. The late-global high-level mechanism ...... 606 5.3. Beyond the occipital lobe ...... 606 6. Remaining questions ...... 607 Acknowledgements ...... 608 References ...... 608

1. Introduction Watanabe and Oyama, 1988) have attempted to elucidate the mystery of ‘human perception of illusory contours with real The segregation of sensory visual inputs into coherent neurons’ (Westheimer, 1994). According to Halpern (1981),at objects is one of the essential operations carried out by the least eight different hypotheses have been proposed to explain visual system. This process requires the identification of the perception of illusory contours. Among phenomenological borders between different components and surfaces, as well as approaches, the ‘cues-to-depth’ hypothesis suggests that IC the selection and grouping of individual visual elements in a might be perceived as an elevated surface plane above the complex scene. Differences in luminance, texture, and/or background plane generated by visual inducers, as can be chrominance are the major cues used to define object boundary. apparent when observers use the monocular depth cue of These real boundaries are physically present in the scene and perceived interposition (Coren, 1972; Coren and Porac, 1983). very efficiently identified by the visual system. Behaviorally, On the other hand, the ‘object-cue’ hypothesis or the the time needed to identify such contours may be as short as ‘cognitive’ model argues that IC might be perceived on the 50 ms (Werner, 1935). On the other hand, contours can also be basis of inferences that the observer can make about objects perceived in the absence of ‘real’ physical discontinuity and their spatial relationships in the world (Gregory, 1972; (Ginsburg, 1975; Petry and Meyer, 1987). These contours are Piggins, 1975; Rock and Anson, 1979). Another phenomen- usually called illusory contours (IC), but have also been ological account is based on the ‘Gestalt completion’ referred to using many other names, including subjective hypothesis that suggests an important role of illusory surface contours, phenomenal contours, cognitive contours, anomalous perception, with a secondary completion of the perceived contours, quasiperceptive contours, unfinished contours, surface by forming transitions and contours (Kanizsa, 1976). In incomplete contours, virtual contours, contours without addition, two main categories of explanations have been gradients, and apparent contours (for additional details, see proposed on the basis of physiological hypotheses. One is the Kanizsa, 1979; Petry and Meyer, 1987; Purghe´ and Coren, ‘brightness–contrast’ hypothesis which claims that IC for- 1992). mation is a consequence of the perception of brightness One class of illusory contours, called Kanizsa figures, can be differences between the illusory figure and its background generated by a particular configuration of distant high-contrast (Brigner and Gallagher, 1974; Jory and Day, 1979). The other borders, such as incomplete and co-aligned white (or black) is the ‘feature-edge detection’ hypothesis that gives a circles that induce the illusory perception of a dark (or light) preponderant role to orientation-specific units in the visual shape, placed over the white (or black) circles (Kanizsa, 1979). cortex, which might be triggered by real edges along the The perception of such figures arises very naturally and inducers and lead to illusory perception of contours beyond effortlessly although we are well aware that they are not real these real edges by propagation and interpolation processes (e.g. Ware and Kennedy, 1978). This type of IC has attracted (Smith and Over, 1975, 1977). Note that sophisticated considerable interest in recent years and has been used to mathematical models have also been proposed to simulate derive influential models of how sensory information is the neural networks underlying the formation of IC according processed within the visual system, mostly in the perspective to these different perspectives (e.g. Brigner, 1982; Sarti et al., of constructive aspects of human vision (for review, see Purghe´ 2000). and Coren, 1992; Spillmann and Dresp, 1995; Zeki, 2004). In These accounts of IC have been extensively tested with particular, much interest has focused on visual mechanisms psychophysical experiments in healthy human subjects. involved in perceptual grouping and segmentation (Marr, Results have generally confirmed at least some validity of 1982). However, the cortical networks involved in such visual several of the different hypotheses, which seem able to processes are only beginning to be identified, as we review successfully explain a range of illusions and effects, although here. not necessarily all possible illusory figures. But the possibility Different functional hypotheses based on psychophysics to see IC in some cases without any appearance of visual have been put forward in the past to account for IC perception, occlusion (Purghe, 1995), brightness enhancement (Kennedy, including phenomenological and physiological aspects to 1988; Parks, 1979), solid edges in the inducers (Gregory, various degrees (Purghe´ and Coren, 1992). Several of these 1972), or cues for object recognition (Enns and Rensink, 1993), theoretical models (Bradley and Dumais, 1975; Coren, 1972; has engendered an intense debate between researchers (e.g. Ginsburg, 1975; Gregory, 1972; Halpern, 1981; Kanizsa, 1976; Dresp and Spillmann, 2001; Parks, 2001; Rowbury, 1982; Parks, 1989; Rock and Anson, 1979; Smith and Over, 1975; Wenderoth, 1997). Such variability underlines the possibility M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 597 that IC formation is perhaps not explained by a single exclusive a promising avenue. In this context, explorations using theory. functional brain imaging techniques have begun to provide Besides functional hypothesis on the mechanisms of IC important new data and represent a welcomed opportunity to perception, another essential question has concerned the compare human and non-human visual functions (Denys et al., cortical level where these contours might be encoded (Brigner 2004; Tsao et al., 2003). and Gallagher, 1974; Gregory, 1972; Rock and Anson, 1979; Using functional imaging to track human brain activity Smith and Over, 1975). The importance of this question was during IC perception has become possible only in relatively emphasized when, 20 years ago, the possibility to study recent years, but a wide range of techniques and experimental perception of IC in animals with electrophysiological approaches have been employed. The first functional ‘neuroi- techniques became available (von der Heydt et al., 1984). maging’ approach published on this topic was conducted 15 Since, the work of Zanforlin (1981) on the perception of IC in years ago by Sugawara and Morotomi, in 1991, and performed chicks, further studies have documented that various organisms with Electro-Encephalography (EEG) (Sugawara and Mor- such as mammals, birds, and insects all are able to perceive IC otomi, 1991). The first imaging study using haemodynamic (for review, see Nieder, 2002). The major findings of these techniques, such as functional Magnetic Resonance Imaging electrophysiological studies in animals relate to the type of (fMRI), was published later by Hirsch et al. in 1995. Since neurons implicated in IC perception and the site where they are these two pioneering works, more than 20 other functional located, as well as the nature of visual information to which neuroimaging studies using paradigms with illusory contours they respond. In the studies carried by von der Heydt and have been published. These different studies have been Peterhans (1989) in monkeys, 44% of V2 neurons were found accomplished with either fMRI, positron emission tomography to signal the orientation of IC defined by abutting gratings and (PET), EEG, magneto-encephalography (MEG), or transcra- 32% of V2 neurons responded to illusory bars of Kanizsa-type nial magnetic stimulation (TMS). In the present review, we will figures (Peterhans and von der Heydt, 1989). In cats, both V1 consider 23 studies that used either haemodynamic or electrical and V2 cells were shown to respond to IC (Redies et al., 1986; neuroimaging techniques, with a special focus on the Sheth et al., 1996). For example, 42% of V1 neurons and 60% anatomical and/or temporal stages of visual processing of V2 neurons conveyed information about the orientation of involved in IC perception (Brighina et al., 2003; Ffytche and IC (Sheth et al., 1996). These data have, therefore, led to the Zeki, 1996; Goebel et al., 1998; Grice et al., 2003; Halgren idea that IC might be generated in part by similar population of et al., 2003; Herrmann et al., 1999; Herrmann and Bosch, 2001; neurons and similar computational mechanisms processing real Hirsch et al., 1995; Korshunova, 1999; Kruggel et al., 2001; contours in early stages of cortical vision. Larsson et al., 1999; Mendola et al., 1999; Murray et al., 2002a, More recently, two further studies on monkeys have 2004a; Ohtani et al., 2002; Palva et al., 1997; Pegna et al., provided additional insights into the role of V1 and V2 2002; Proverbio and Zani, 2002; Ritzl et al., 2003; Seghier neurons. Ramsden et al. (2001) used optical imaging and et al., 2000a; Senkowski et al., 2005; Stanley and Rubin, 2003; single-unit recording to show that both V1 and V2 neurons Sugawara and Morotomi, 1991). A few other studies have also responded to IC defined by abutting gratings, but their orientation was negatively signaled in V1 as compared to used illusory figures to map the 40-Hz gamma activity in EEG real contours. These authors proposed that this de-emphasis of during perception of coherent visual objects (Bottger et al., IC orientation in V1, together with the normal positive IC 2002; Herrmann and Mecklinger, 2000; Kaiser et al., 2004; responses in V2, could provide a unique signature for the Keil et al., 1999; Tallon-Baudry et al., 1997), while still others neural representation of IC as compared with real contours. investigated the functional neuroimaging correlates of other Moreover, using stimuli made by Kanizsa figures, Lee and kinds of visual illusions (Hayashi et al., 2004; Lee and van Nguyen (2001) have found that although both V1 and V2 Donkelaar, 2002; Muckli et al., 2005; Sasaki and Watanabe, neurons can respond to IC, the V2 responses were stronger and 2004), or tested the effects of different neurological disorders faster than V1 responses. Taken together, these results suggest on the perception of illusory contours (Fierro et al., 2003; Foxe that IC processing is an interactive process involving complex et al., 2005; Grice et al., 2003; Vuilleumier and Landis, 1998; inter-cortical interactions between early visual areas. Vuilleumier et al., 2001). However, the latter studies will not Bringing together the findings from human psychophysics be directly considered here since they do not concern visual and animal physiology is clearly a central issue to bridge the processes related to IC specifically, but more general questions gap between the different levels of explanation of IC on visual perception and visual awareness. The purpose of our perception (Peterhans and von der Heydt, 1991; Spillmann review is to present and compare the results from the extant and Dresp, 1995). However, several difficulties may arise when neuroimaging studies on IC processing, to discuss the possible confronting the rich visual perception of humans with the sources of discrepancies, and to tentatively bring together their single neuronal responses to such contours measured in findings in a general functional framework. To our knowledge, animals, as illustrated by the classically debated question of no systematic overview of this new neuroimaging literature is ‘do animals see what we see?’ (Ohzawa, 1999). To overcome available. By delineating the commonalities and discrepancies some of these difficulties, new approaches allowing a more between the results from different imaging studies, we point to direct comparison between psychophysical measures and the possible factors implicated during visual responses to IC cortical activity in humans during IC perception should offer and suggest further directions of investigation. 598 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612

2. Methodological issues discrepancies between studies might result from different acquisition protocols. For example, with fMRI, the limited Before examining the major findings of previous functional brain coverage in some studies and the signal loss due to neuroimaging studies, it is important to emphasize the geometric distortion in ventral regions of the temporal and methodological differences that may exist between different frontal lobes (Devlin et al., 2000) should be taken into account studies. These methodological aspects, including acquisition when comparing fMRI studies with PET or MEG studies. protocols, analysis approaches, stimulus types, and experimen- For EEG and MEG, spatial resolution is basically limited by tal paradigms, constitute major sources for the observed the inter-sensor distances and by the fundamental laws of variability across previous studies. Table 1 lists the relevant electromagnetism. In addition, the sensitivity of MEG is characteristics of each of the studies included in this review. weaker in more medial occipital regions like V1/V2 areas than More specific aspects are detailed below. in more lateral regions of the brain (Hamalainen and Sarvas, 1989). This is an important issue to consider when interpreting the findings of activated regions in MEG studies. Also, solving 2.1. Imaging techniques the ‘inverse problem’ in order to identify the intracranial generators of electrical scalp measurements (Koles, 1998; The choice of the technique of investigation can obviously Michel et al., 2004) is a major challenge that will seriously have direct consequences on the ‘nature’ of the observed limit the anatomical accuracy of cortical localization in EEG results (Ojemann et al., 1998; Sadato et al., 1998). Particularly, studies. differences between haemodynamic approaches (PET and fMRI) and electro-magnetic approaches (EEG, MEG, TMS) in terms of sensitivity and resolution are well documented (e.g. 2.2. Data analysis methods Momjian et al., 2003; Toga and Mazziotta, 1996). As listed in Table 1, two studies of IC perception used PET, seven used Conflicting results between previous imaging studies may fMRI, nine used EEG, two used fMRI/EEG combination, two also partly be due to methodological differences in the used MEG, and one used repetitive TMS. In addition, even approaches used for data analysis. For example, the eight when employing the same imaging technique, some studies listed in Table 1 that used fMRI have employed

Table 1 Major characteristics of the methods and stimuli used in different functional neuroimaging studies

Study Techniq. Subjects Particularity Stimulus Real cont. Sugawara and Morotomi (1991) EEG Eight subjects No Static IC (Kanizsa figures) No Hirsch et al. (1995) FMRI Four subjects (one female) Subjects were familiar with Flickered IC (Kanizsa Yes the stimuli figures) Ffytche and Zeki (1996) PET Eight subjects (0 female) Trained prior to the Static IC (Kanizsa figures) Yes experiment Goebel et al. (1998) FMRI 10 subjects (four females) No Apparent motion of IC No (Kanizsa figures) Herrmann et al. (1999) EEG 10 subjects (six females) No Static IC (Kanizsa figures) No Korshunova (1999) EEG 10 subjects No Static IC (Kanizsa figures) No Larsson et al. (1999) PET Six subjects (0 female) No Static IC (abutting-lines) Yes Mendola et al. (1999) FMRI 16 subjects Subjects were sophisticated Static IC (Kanizsa figures, Yes psychophysical observers abutting-lines) Seghier et al. (2000a) FMRI 13 subjects (one female) Nine were fMRI trained Moving IC (Kanizsa No subjects figures) Herrmann and Bosch (2001) EEG 13 subjects (seven females) No Static IC (Kanizsa figures) No Kruggel et al. (2001) FMRI EEG 12 subjects (five females) No Static IC (Kanizsa figures) No Murray et al. (2002a) FMRI EEG Five subjectsa (three females) No Static IC (Kanizsa figures) No Ohtani et al. (2002) MEG Four subjects (0 female) No Static IC (abutting-lines) No Pegna et al. (2002) EEG 20 subjects (10 females) No Static IC (Kanizsa figures) Yes Proverbio and Zani (2002) EEG Eight subjects (four females) No Static IC (Kanizsa figures) No Brighina et al. (2003) RTMS Eight subjects (four females) Subjects were familiar with Static IC (Kanizsa figures) Yes the stimuli Grice et al. (2003) EEG 15 subjects (seven females) Subjects participated in Static IC (Kanizsa figures) Yes practice sessions Halgren et al. (2003) MEG Five subjects (0 female) Subjects with high level of Multiple static IC No compliance (Kanizsa figures) Ritzl et al. (2003) ER-fMRI 11 subjects (three females) No Static IC (Kanizsa figures) Yes Stanley and Rubin (2003) FMRI Eight subjects (three females) Four were novice, and two Static IC contrast reversed No were the authors at 0.5 Hz (Kanizsa figures) Murray et al. (2004a) EEG Nine subjects (four females) No Static IC (Kanizsa figures) No Senkowski et al. (2005) EEG 12 subjects (three females) All were student subjects Static IC (Kanizsa figures) No

a In Murray et al. (2002a), with a total of 28 subjects, 5 subjects participated in both EEG and fMRI experiments. M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 599 different methods to analyze signal changes, such as mean neuroimaging studies. In fact, as summarized in Table 1, the baseline-stimulation comparison (Hirsch et al., 1995), corre- degree of familiarity of subjects with stimuli varied substan- lation analysis (Goebel et al., 1998; Seghier et al., 2000a), tially across previous imaging studies. This factor may play an Fourier domain analysis (Mendola et al., 1999; Stanley and underestimated role, but would be consistent with the fact that Rubin, 2003), general linear models (Murray et al., 2002a; perceptual learning can produce long-lasting functional Ritzl et al., 2003), or regression analysis with nonlinear changes in early cortical visual areas, even for simple visual modeling (Kruggel et al., 2001). Also, even when using the tasks (Schwartz et al., 2002; Maertens and Pollmann, 2005). same approach, some differences in statistical threshold can Moreover, few behavioral reports have also suggested sex have a direct repercussion on the reported results (see differences during illusory perception (Grabowska et al., 2001; illustrations in Friston et al., 1996; Loring et al., 2002). This Heider, 1996), but sex differences have not systematically been issue seems to be particularly important when comparing the taken into account in previous functional neuroimaging previous findings about the right hemisphere preference for IC studies. processing (see below). Furthermore, a higher statistical significance of V1/V2 activation can be apparent when 2.4. Stimuli analyzing individual data within retinotopically defined regions-of-interest as compared with group analysis (Mendola The illusory contours used in brain imaging studies were et al., 1999), and the inter-individual variability in the mainly derived from the Kanizsa type of figures (see Fig. 1), anatomical location of visual cortical areas can lead to except for two studies where co-aligned line segments were important differences between group-based and individual employed (Larsson et al., 1999; Ohtani et al., 2002). Kanizsa region-based approaches (e.g. Hasnain et al., 1998). figures constitute a special class of IC since their perception Similarly, important differences may exist between different may not only imply the extraction of contours proper, but also approaches used to record and analyze EEG data (Picton et al., involve surface segmentation (e.g. Grossberg and Mingolla, 2000). Whereas some studies may focus on amplitude or 1985). Generally, imaging studies of IC perception have, latency measures for well-defined waveforms in the evoked therefore, mostly focused on modal completion processes that potentials recorded at specific electrodes (Picton et al., 1995), lead to IC formation with concurrent brightness enhancement other studies use different measures based on the topographical (Singh, 2004), except for one ERP study that directly compared distribution of electrical activity over all electrodes and cluster both modal and amodal completions (Murray et al., 2004a). In analysis of temporal segments in EEG (Lehmann, 1987; other cases, therefore, the distinct role of edge/contour Michel et al., 1999). processing and surface segmentation processes were not clearly teased apart. 2.3. The subject factor Furthermore, a large variety of shapes formed by Kanizsa IC has been used, including square, triangle, diamond, star, circle, In early behavioral work by Dresp and Bonnet (1991), it was and pentagon shapes (see Fig. 1). These shapes and their shown that the perception of an illusory contour may drop associated IC were generated by a multitude of different visual when the visual fixation point is positioned closer to the inducers, including static and moving features. Table 1 lists for illusory contour itself. However, when this experiment was each study the nature of these illusory figures (static or moving) replicated in a larger group of subjects (McCourt and Paulson, and indicates if real contours were also used for comparison. 1994), the results revealed that some subjects showed the same The size of the illusory figure and the support ratio (i.e. the ratio pattern as initially reported by Dresp and Bonnet, whereas of the length of the real inducing contours relative to the total other subjects showed the inverse pattern, suggesting a length of the illusory figure (Shipley and Kellman, 1992)) are considerable variability across individual observers for the also illustrated for each study in Fig. 2. Generally, a relatively perception of illusory contours, at least in some experimental conditions. However, such inter-individual differences in visual performance are often neglected. Another study of Stanley and Rubin (2003) using fMRI clearly shows that such differences can affect the pattern of functional maps. In this study, four participants were very experienced with IC and four were novice. Activations in the fMRI results of the four novice subjects were found to have a lower statistical significance as compared with those from the experienced subjects, suggesting a role of individual differences related to previous practice with the visual stimuli (Stanley and Rubin, 2003). These findings indicate that the strength of the perception of IC can differ from one observer to another, and that the choice of subjects might be an important factor potentially affecting Fig. 1. Illustration of the variety of shapes based on Kanizsa’s IC that have been the results observed in both behavioral and functional used in functional neuroimaging studies. 600 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612

other arguments such as the absence of any differential cortical responses to various configurations of the same rotated inducers (Lee and Nguyen, 2001) even with collinear inducers (Herrmann and Bosch, 2001), as long as no IC is generated. Such demonstration provides strong support to the claim that the crucial properties of the illusory figures explored by functional neuroimaging specifically concerned the presence and/or perception of IC, with the associated closed surface and brightness enhancement, rather than another non-specific configural property of the figure.

3. IC processing: The ‘where’ debate

Different functional models have been proposed to characterize the processing level of IC (see Section 1). A specific role of early (low) level, midlevel, or high level of Fig. 2. Dimensions of illusory figures used by functional imaging studies. The visual processing has variably been suggested (Spillmann and size of the IC is plotted against the support ratio. Dresp, 1995; Watt and Phillips, 2000; Zeki, 2004). We will high support ratio between 0.4 and 0.5 was used in most now examine in greater details what processing stages have studies, suggesting an important contribution of the real been implicated by neuroimaging studies, in terms of activated borders to the formation of the illusory figures (Grossberg brain regions, and make a brief comparison with previous et al., 1997; Kojo et al., 1993). Finally, the size (i.e. length) of psychological and electrophysiological findings. Generally, illusory figures varied from 2 to 108, suggesting that the these functional studies have not found neural networks involved ‘exclusively’ in processing of IC, supporting an strength of IC perception may have substantially varied across overlap between the cortical representation of real contours and studies (Heider et al., 2002). IC—except for two studies that suggested the existence of IC One critical issue is the choice of the control condition used in ‘specific’ regions (Hirsch et al., 1995) or IC ‘selective’ regions functional neuroimaging paradigms, since the critical acti- (Larsson et al., 1999). The major regions implicated across vations in fMRI and PET studies are typically obtained by different neuroimaging studies are listed in Table 2. ‘contrast’ or ‘subtraction’ of a control condition from the experimental stimulus condition. For instance, to eliminate the 3.1. V1/V2 areas contribution of real objects (i.e. inducers) in Kanizsa figures, different control figures have been developed. The most widely A major debate concerns the implication of the primary used control figures consisted of the same objects as the inducers visual area (V1) in striate cortex during the perception of (i.e. shape, size, contrast, position), but rotated to disrupt the illusory contours. The early study by Hirsch et al. (1995) found percept of IC. In all studies, the statistical differences between activation in extrastriate visual areas only (in particular activations to the Kanizas figures with IC and the control figures Brodmann’s area 18), predominantly in the right hemisphere. without IC were considered as related to the presence of IC. Although, the authors concluded that IC perception may reside Although this approach is generally valid, it is possible that in outside area V1 (Hirsch et al., 1996), the activation observed in some cases the control figures might not be ‘ideal’ because the their study was near the cortical region coding for the vertical spatial distribution of their real edges is ‘retinotopically’ meridian projection between V1 and V2, and was therefore still different from those in Kanizsa figures, and these differences consistent with a possible implication of V1 (Hirsch et al., are critical when low-level visual areas are considered. Such 1995). Similar findings have subsequently been reported by differences might be especially problematic with increasing Palva et al. (1997). numbers of inducers (e.g. as visible in Fig. 1 of (Halgren et al., By contrast, Larsson et al. (1999) used a different kind of 2003) and Fig. 2 of (Senkowski et al., 2005)). Other authors have illusory figures and showed strong activation within V1 itself, argued that these control figures might have different physical bilaterally. This V1 activity was not specific to illusory energy from the Kanizsa figures (e.g. Skottun, 1994). contours since it was also observed during the perception of Furthermore, even the perception of the global shape generated real contours. In addition, a significant implication of V1 has by the spatial arrangement of inducers might be disrupted in also been demonstrated during the perception of moving control figures relative to Kanizsa figures, some authors have illusory contours (Seghier et al., 2000a, 2000b). In the latter emphasized that shapes could sometimes be ‘inferred’ from the study, V1 activity was more prominent during animated IC arrangement of elements, and used the terms of inferred figures than static IC perception (Seghier et al., 2000a,b). Also, in a instead of control figures (Ritzl et al., 2003). recent study using fMRI to investigate the neon color spreading Despite these potential problems and confounds, such illusion (i.e. the perception of an illusory colored transparent control figures are still clearly useful, and their validity in surface), a significant activation was observed in V1 for careful experimental paradigms appears to be supported by those stimuli induced by IC (Sasaki and Watanabe, 2004). M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 601

Table 2 Lists the major ﬁndings about the implicated regions during IC perception

Study V1/V2 activation Other regions Hirsch et al. (1995) Possible activation in V1 and robust V2 implication Extrastriate areas, in the right area 18 of Brodmann in the right hemisphere Ffytche and Zeki (1996) No activation in V1. Robust bilateral activation in Possible implication of V3 V2 Goebel et al. (1998) No activation in V1. Implication of V2 during Implication of V5 area without V3A implication apparent motion of IC Mendola et al. (1999) Very weak activation in V1 and V2 (after individual Strong activation in the Lateral Occipital Complex LOC, including V3A, analysis) V4v, V7 and V8 areas Larsson et al. (1999) Activation of V1 and V2 as with real contours Activation of the fusiform and lingual gyri, and within the cuneus and the parietal lobe. Selective region for the IC was located in the right fusiform gyrus Seghier et al. (2000a) Implication of V1 and V2, in particular during IC Implication of V5 and the LOS/KO region animation Kruggel et al. (2001) No activation in V1 and V2 Different parts of the lateral occipital gyrus, including V5 area Murray et al. (2002a) No activation in V1 and V2 Strong activation in the Lateral Occipital Complex region LOC. Implication of the right parietal cortex Ohtani et al. (2002) Activation of V1 and V2 in half of subjects – Pegna et al. (2002) No activation in V1 and V2 Bilateral activation of the lateral occipital complex LOC Brighina et al. (2003) No activation of V1 and V2 Strong implication of the right extrastriate cortex Halgren et al. (2003) Implication of V1 and V2 Implication of Lateral Occipital Region LOR, with lingual/fusiform and orbitofrontal areas Ritzl et al. (2003) No activation in V1 and V2 Bilateral activation of the LOC, stronger in the left hemisphere Stanley and Rubin (2003) Clear activation in V1 and V2 not observed Implication of the LOC region, interpreted as not speciﬁc to IC. Parietal region are also reported Murray et al. (2004a) No activation in V1 and V2 Strong activation in the LOC region and the posterior parietal regions

Most relevant results reported in imaging studies, with particular emphasis on the implication of low-level visual areas V1/V2.

In addition, a role of early visual areas in the processing of IC is In any case, as can be seen in Table 2, V2 can be considered as the supported by studies using MEG (Ohtani et al., 2002) and some earliest visual area that is most commonly implicated in IC ERP results (Herrmann and Bosch, 2001; Proverbio and Zani, processing using a variety of different stimuli and paradigms. 2002). These neuroimaging data, suggesting a role of the lowest On the other hand, a number of other findings have suggested cortical visual areas in humans, converge with the results from that V1 may not be implicated during the perception of static several electrophysiological studies on animal brains. In cats, it illusory contours (Ffytche and Zeki, 1996; Kruggel et al., 2001; has been shown that neurons in V1 (area 17) can respond to Mendola et al., 1999; Murray et al., 2002a; Stanley and Rubin, illusory contours (Redies et al., 1986; Sheth et al., 1996; Zhou 2003) or during apparent motion of illusory contours (Goebel et al., 2001) just as well as they usually respond to real physical et al., 1998). Although Mendola et al. (1999) concluded that no contours. Also, a single-cell recording study using Kanizsa reliable activation was present in V1, a possible implication of figures found that 14–26% of neurons tested in V1 exhibited V1 could not be entirely ruled out when the authors employed a statistically significant responses to the IC (Lee and Nguyen, more detailed analysis in individual subjects and assessed fMRI 2001; Lee, 2002a). More recently, with electrophysiological signal in a restricted region-of-interest that was retinotopically recording and optical imaging, it was also found that V1 could defined in V1 (Mendola et al., 1999). exhibit significant responses to illusory contours defined by The role of the secondary visual area V2 has been more abutting line gratings (Ramsden et al., 2001). In addition, a consistently established. Some activation in area V2 during significant role of V2 activity was already suggested 20 years IC perception was found in most neuroimaging studies ago by the electrophysiological findings of von der Heydt et al., (Ffytche and Zeki, 1996; Goebel et al., 1998; Hirsch et al., demonstrating that about one-third of cells in this area could 1995; Larsson et al., 1999; Ohtani et al., 2002; Sasaki and respond to the presence of an illusory bar (Peterhans and von Watanabe, 2004; Seghier et al., 2000a), sometimes with der Heydt, 1989; von der Heydt et al., 1984). More recent stronger activity during illusory than real contour perception electrophysiological data have confirmed such a role of V2 in (Ffytche and Zeki, 1996; Hirsch et al., 1995). However, a processing shape edges defined by illusory contours (Baumann few studies did not observe any reliable involvement of V2 et al., 1997; Heider et al., 2000, 2002; Lee and Nguyen, 2001; (Kruggel et al., 2001; Mendola et al., 1999; Murray et al., Leventhal et al., 1998; Ramsden et al., 2001; Sheth et al., 2002a; Ritzl et al., 2003; Stanley and Rubin, 2003), although 1996). Furthermore, it has been shown that the responses of V2 the latter results did not entirely rule out its possible neurons to IC in the alert monkey can signal the figure-ground implication as assumed when using more detailed analysis direction that is perceived at such contours (Baumann et al., approaches in individual subjects (e.g. Mendola et al., 1999). 1997; Peterhans and Heitger, 2001). 602 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612

Besides neurophysiology measures in humans or animals, a the detection of strong activation in higher-level areas instead number of purely psychophysical studies have also strongly (see below), a finding that might potentially be related to recent suggested the implication of low-level visual areas V1/V2 reports suggesting that the implication of early visual areas is during the perception of illusory contours (Banton and Levi, reduced when grouping processes during shape perception is 1992; Brigner and Gallagher, 1974; Davis and Driver, 1994; preformed in higher visual areas (Murray et al., 2002b, 2004b). Dresp and Bonnet, 1991; 1993; Pillow and Rubin, 2002; Furthermore, the recent findings of Ramsden et al. (2001) Ramachandran et al., 1994). The implication of low-level areas suggesting an inverse correlation between the response of V1 in these psychophysical studies was deduced from the neurons to IC and real contours point to more complex interaction of the IC perception with the manipulation of functional scheme underlying the neural representation of IC in local and low-level physical parameters known to influence early visual cortex, such that the common neuroimaging directly the processing in early stages of the human visual approach testing for a subtraction of activation to control system. For example, using a visual search experiment of stimuli may remain insensitive to V1 responses in some cases. Kanizsa figures among figures without IC, it was shown that Finally, the size of the perceived IC may also explain some Kanizsa figures may be detected prior to focal attention, variability across studies. Neurophysiology data have shown suggesting a parallel and automatic process compatible with that responses of V1/V2 cells to IC are reduced when the size the involvement of early visual areas (Davis and Driver, 1994; of IC increases, with an optimal response at 2–48 (Heider et al., Senkowski et al., 2005) (but see, Gurnsey et al., 1996). In 2002). All fMRI studies that observed responses in V1/V2 another behavioral study, when an illusory figure was super- areas have used IC with sizes between 2 and 48 of visual angle, imposed on a checkerboard pattern, a vivid perceptual with a support-ratio R0.4 (Goebel et al., 1998; Hirsch et al., enhancement of IC was observed when the edges of inducers 1995; Seghier et al., 2000a), whereas studies using IC sizes were collinear with the edges of the checks, whereas a between 5 and 108 have reported no significant responses in misalignment between these edges was associated with a V1/V2 areas (Mendola et al., 1999; Murray et al., 2002a; Ritzl subjective disappearance of IC (Ramachandran et al., 1994). et al., 2003; Stanley and Rubin, 2003). Clearly, further work is These findings suggest an interaction between visual processes needed to explore more systematically the role of these subserving the perception of IC with those extracting local parameters, and still others, that all may potentially influence physical edges in the visual field. Similarly, when observers are the involvement of low-level visual cortical areas. asked to discriminate the shape of slightly deformed Kanizsa Overall, these different functional arguments support the figures, performance is much poorer when the IC crosses the hypothesis of an important role of early visual cortex during IC vertical meridian than when it resides entirely within one (left perception, in particular V2. Results from neuroimaging or right) visual hemifield, reflecting some limitations in the studies have also provided some arguments in favor of the cross-hemispheric integration and, therefore, suggesting a implication of area V1, although some uncertainty remains crucial role of early retinotopic areas, perhaps in V1/V2, about the exact conditions and mechanisms that might be since these low-level areas are more sensitive to the interhemi- involved. It is likely that the implication of V1/V2 in IC spheric divide (Pillow and Rubin, 2002). perception could play a crucial role in ‘feature-edge detection’ In sum, as recapitulated in Table 2, there is still no clear mechanisms encoding orientation and luminance contrast, consensus across studies about a specific implication of V1 in which are thought to contribute to the generation of IC by IC perception, while there is now a more compelling evidence extension of the real edges of inducers, in agreement with for a role of V2. Contradictory results between previous earlier physiological and computational theories (Grossberg imaging studies may at least partly be due to methodological and Mingolla, 1985; Smith and Over, 1975). These findings issues, as described above. But a number of other factors may also converge with the suggestion of an implication of V1 in also contribute to these apparent discrepancies. segmentation and grouping processes that might potentially First, the perceptual attributes of illusory figures used as contribute to IC perception, as shown by various neurophy- stimuli in the different studies is likely to play an important role siological studies (Lamme et al., 1993; Lamme, 1995; Marcus in the degree of implication of V1/V2 areas. For example, in and Van Essen, 2002; Sugita, 1999), and more generally fMRI studies reporting significant V1/V2 implication, some support the suggestion of a significant role of V1/V2 in higher temporal/dynamic animations have often been added to the level vision (Lee et al., 1998, 2002). illusory figure, which may produce a stronger perceptual saliency for such figures and consequently more robust 3.2. The lateral occipital complex (LOC) functional responses as compared with static illusory figures (Ni et al., 2003; Seghier et al., 2000a, 2000b). Different types of Different regions in higher-level visual cortex have also animations have been used, including flickering of IC at 4 Hz been proposed to constitute specific processing centers for (Hirsch et al., 1995), inducing apparent motion at 2 Hz (Goebel illusory contours. In the study of Mendola et al. (1999),a et al., 1998), and creating linear motion displacements at 3.28/s region selectively activated by IC was localized in the anterior of velocity (Seghier et al., 2000a). In addition, the absence of occipital lobe, including the putative visual areas V7 and V8, significant activation in V1/V2 areas in some studies (Kruggel bilaterally, but often extending more posteriorly into V3A and et al., 2001; Mendola et al., 1999; Murray et al., 2002a; Ritzl V4v. This region overlapped with the lateral occipital region et al., 2003; Stanley and Rubin, 2003) was associated with LOR that can be strongly driven by visual stimuli in the M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 603 ipsilateral visual field (Tootell et al., 1998b). A similar region In addition, a region containing the Lateral Occipital Sulcus was identified in other fMRI studies (Murray et al., 2002a; (LOS) and the Kinetic Occipital (KO) areas was found to be Stanley and Rubin, 2003) and with ERP source mapping activated during the perception of moving illusory contours in techniques (Murray et al., 2004a; Pegna et al., 2002), probably an fMRI study using moving stimuli (Seghier et al., 2000a). corresponding to a part of the lateral occipital complex (LOC) Area LOS was previously proposed to be part of the motion (Malach et al., 1995). processing network (Sunaert et al., 1999), whereas area KO One explanation for the implication of LOC may simply be was initially thought to be a specific processing center for that this region is generally implicated in the recognition of kinetic contours (Orban et al., 1995; van Oostende et al., 1997; coherent objects and bounded surfaces (Grill-Spector et al., but see Zeki et al., 2003). The KO region has been found to 2001; Vuilleumier et al., 2002). Illusory contours generally respond to second or higher order motion (Dupont et al., 2003; delimitate an illusory figure (i.e. square, triangle, etc.) that has Smith et al., 1998) and was also called V3B (Smith et al., a distinguishable shape from the background, and that 1998). However, this region may not respond only to moving consequently may be processed as an elementary object within illusory contours, but also to static illusory contours (Kruggel the LOC region (Kourtzi and Kanwisher, 2001; Yin et al., et al., 2001), and it is possible that it was partly included in 2002). A possible separation between the perception of activated clusters within the lateral occipital cortex as found by illusory contours and coherent shapes within the LOC region several previous imaging studies (Mendola et al., 1999; Murray (e.g. Mendola et al., 1999) needs additional investigations with et al., 2002a; Stanley and Rubin, 2003). carefully designed stimuli. For example, in a MEG study using Finally, the posterior parietal regions (Halgren et al., 2003; abutting-line grating to induce IC without an associated figure Murray et al., 2004a), lingual gyrus (Halgren et al., 2003), and shape, no activity was detected in the LOC region (Ohtani orbitofrontal cortex (Halgren et al., 2003) have also occasion- et al., 2002). Similarly, Stanley and Rubin (2003) created ally been proposed to be implicated in the perception of IC, but modified Kanizsa figures by rounding and misaligning the these regions appear less consistently seen across studies. corners of inducers, and their fMRI results clearly demonstrated that the LOC region was still responsive to the global 3.4. Retinotopy of IC figure resulting from such inducers even though it was not bounded by illusory contours. These new data suggest that A convincing demonstration of the retinotopic coding of activity in the LOC region may not be specific to the illusory contours within earlier visual areas would provide a perception of illusory contours per se. However, it may direct and strong support for the low-level extraction of such contribute to the extraction of the global shape defined by IC contours, given that retinotopy is a major property of early (Kourtzi and Kanwisher, 2001), and perhaps also to the visual areas (Courtney and Ungerleider, 1997). As previously generation of surface properties induced by IC and other visual shown for more cognitive functions such as mental imagery cues in stimuli such as Kanizsa figures. and visual attention, a modulation of V1 activity in retinotopically manner can now be successfully determined by functional imaging techniques (e.g. Engel et al., 1997; 3.3. Other cortical regions Tootell et al., 1998a). For example, it has been found that the neural activity produced by mentally generating visual images Another candidate region involved in IC perception, or focusing spatial attention at specific location in the visual initially proposed by Larsson et al. (1999), has been localized field may have a corresponding retinotopic organization that in the right fusiform gyrus. This region was significantly more can be documented and quantified by fMRI (e.g. Brefczynski activated by the perception of illusory contours than by real and DeYoe, 1999; Kosslyn et al., 1995; Sasaki et al., 2001; contours in a PET experiment (Larsson et al., 1999), and was Slotnick et al., 2005). more recently also observed in activation source maps obtained Using illusory contours with different length, Mendola et al. with MEG (Halgren et al., 2003). The activation pattern shown (1999) found that the activation patterns in area LOR were in the early fMRI study of Hirsch et al. (1995), with a invariant across a wide range of illusory figure size. These predominant involvement of the right hemisphere, was also findings suggest that there might be no retinotopical projection likely to include the same right fusiform region as subsequently of IC in this area, consistent with the fact that the activated proposed by Larsson et al. (1999). A role of this more ventral LOR region has large and bilateral receptive fields, as visual area might be in agreement with the impaired ability of previously demonstrated by another study of the same group monkeys to see IC after lesions in the inferotemporal cortex (Tootell et al., 1998b). On the other hand, Murray et al. (2002a) (Huxlin et al., 2000) and its involvement during perception of have shown that the retinotopic position of IC (central versus partially occluded shapes (Kovacs et al., 1995). Such lateral presentation) may alter their neural processing. Also, it involvement of the fusiform cortex, known to mediate complex was recently shown by fMRI that IC can activate different shape and object recognition abilities (Grill-Spector et al., visual areas in retinotopically specific manner when these 2001; Vuilleumier et al., 2002), has, therefore, been interpreted induce the neon color spreading illusion (Sasaki and Watanabe, as consistent with the ‘object-cue’ and ‘cognitive’ hypotheses 2004). in the perception of IC (Gregory, 1972; Piggins, 1975; Rock Another recent attempt to identify retinotopic encoding was and Anson, 1979). done by Delon-Martin et al. (2000). They used a new kind 604 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 of illusory contours (modified Ehrenstein figures), affording the processing. Similarly, the right hemisphere might be particu- perception of illusory annular shapes with different eccentri- larly involved in extracting visual information conveyed by the cities. Their results suggested that such illusory contours may low spatial frequencies in images (important to define surfaces generate a retinotopic representation in V1, with fMRI activity and luminance properties), relying on inputs from magnocel- increasing at progressive eccentricities in correspondence with lular visual channels; whereas the left hemisphere might the increasing size of annular IC (Delon-Martin et al., 2000). conversely be more sensitive to high spatial frequencies These data, therefore, provide a relatively strong support to the (important to encode edges and textures), relying on inputs view of a retinotopically organized representation of IC in from parvocellular channels (Iidaka et al., 2004; Peyrin et al., early visual cortex. 2004). Overall, hemispheric asymmetries in IC processing remain 3.5. IC perception and the right hemisphere controversial and probably variable across different tasks and stimulus conditions. More systematic investigations concern- Another matter of debate concerns a possible dominance of ing these questions are crucial for a better understanding of the the human right hemisphere in processing illusory contours. object recognition capabilities of the human visual system and The right hemisphere was found to be more activated than the their underlying neural mechanisms (e.g. Bar, 2003; Gross- left hemisphere during the perception of IC in a few imaging berg, 1980). studies (Hirsch et al., 1995; Larsson et al., 1999), consistent with data from earlier neuropsychological studies in brain- 4. IC processing: The ‘when’ debate lesioned patients (Wasserstein et al., 1987). Neural responses evoked by IC were also more prominent over the right than the Unlike haemodynamic imaging methods, EEG and MEG left hemisphere in a recent MEG study (Halgren et al., 2003), can provide a fine temporal analysis of the time-course of with sources localized in the anterior lateral occipital region. visual perception, although conversely their spatial resolution Likewise, using repetitive TMS, the right extrastriate cortex is much less reliable than PET or fMRI. Studies using ERPs to has been shown to be particularly critical for illusory contour investigate brain responses IC have reported a large variety of perception (Brighina et al., 2003). effects, arising at various latencies post-stimulus onset, and However, these results have not been confirmed by other corresponding to both early and late stages of processing. We functional imaging data (e.g. Murray et al., 2002a; Proverbio will provide only a brief overview of the latencies of and Zani, 2002). Mendola et al. (1999) proposed that the right differential EEG or MEG responses associated with the hemisphere dominance might be observed only when high and formation of IC. restrictive statistical thresholds are chosen, suggesting a Overall, several ERP studies have shown that the perception relative rather than absolute specialization. Furthermore, of IC can modulate the exogenous visual N1 component, using EEG, Proverbio and Zani (2002) showed that the two peaking at approximately 145–160 ms post-stimulus onset hemispheres might be equally activated by IC but at different (Herrmann and Bosch, 2001; Pegna et al., 2002; Proverbio and stages of processing, with a left occipital dominance at the Zani, 2002). This N1 component is generated in extrastriate latency of the N2 component and a right parietal dominance at cortex and modulated by stimulus visibility or attention. The the level of the P300 component. In addition, stronger left than earliest VEP modulation reported for the presence versus right hemisphere activation was observed in a recent event- absence of IC shape was found at 88 ms in one study (Murray related fMRI study (Ritzl et al., 2003). et al., 2002a), but the largest effects arose at much later Similarly, some psychophysics studies have provided latencies after 200 ms. More sustained ERP effects have also conflicting data on hemispheric lateralization. Several studies been observed over a variable period of 130–180 ms (Bottger reported that perceptual mechanisms responsible for illusory et al., 2002; Kruggel et al., 2001), 108–228 ms (Grice et al., contours might be preferentially lateralized to the right 2003), or 140–240 ms (Murray et al., 2004a) post-onset, hemisphere (Atchley and Atchley, 1998), in particular for associated with sources in the LOC and posterior parietal male subjects (Grabowska et al., 2001; Rasmjou et al., 1999), regions (Murray et al., 2004a). With MEG, it was shown that whereas other studies have, in contrast, suggested that both the most prominent responses to IC perception occurred in the hemisphere are equally capable of perceiving IC with modal anterior lateral occipital region at 155 ms (Halgren et al., completion (Corballis et al., 1999). 2003). The question of the lateralization of IC perception might Amplitude differences in the range of the N200 components partly be related to another debate concerning the lateralization have also been often reported. Thus, distinct components such of global versus local visual processing. Mechanisms of as the N170 (Herrmann et al., 1999; Herrmann and Bosch, grouping responsible for the perception of a global illusory 2001), the N180 (Korshunova, 1999; Sugawara and Morotomi, shape have been proposed to implicate the right hemisphere 1991), and the N2 peaking at 250 ms (Proverbio and Zani, preferentially, including regions in the inferior occipital lobe, 2002) have been found to respond differentially in the presence as shown by previous imaging studies of global form versus absence of IC. Such responses are usually thought to perception (Evans et al., 2000; Fink et al., 1996; Han et al., reflect activation of the structural representation and categor- 2002); but see also (Weissman et al., 2002). In contrast, the left ization of visual object. The P230 component (Korshunova, hemisphere might be preferentially implicated in local featural 1999) as well as the P300 component peaking at 390 ms M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 605

(Bottger et al., 2002; Proverbio and Zani, 2002) can also be modulated by IC perception, but these later effects on P300 are likely to reﬂect post-perceptual processes including task- related or response-related factors. In another study, the earliest modulation by IC appeared at 130–180 ms, but the strongest effect modulated the subsequent P300 component (300–500 ms) (Herrmann et al., 1999). Interestingly, indirect evidence for an implication of low- level visual areas (such as V1/V2) was often found in these electromagnetic studies but these effects appeared at a relatively late latency, approximately in the interval of 200–300 ms post-onset. Only two studies using MEG showed a possible response occurring over occipital areas at 110 ms (Halgren et al., 2003) or in the range of 80–150 ms (Ohtani et al., 2002), suggesting a rapid and purely bottom-up response. However, the more frequent EEG effects observed at Fig. 3. Schematic diagram representing the major mechanisms of IC perception 200–300 ms may support a role of feedback mechanisms that are proposed on the basis of neuroimaging ﬁndings. Time post-stimulus arising at later latencies in the anatomically so-called early onset is represented along the x-axis, and anatomical hierarchy of visual areas is visual areas, as we will discuss in more details below. Similar represented along the y-axis. In gray color, the fast-local, low-level mechanism effects indicating a modulation of ‘early’ visual areas with of IC perception; in black, the late-global, high-level mechanism. The star indicates the beginning of the perception of real contours (less than 50 ms in relatively ‘late’ latencies have been suggested in relation to V1). (LOC, lateral occipital complex; IT, inferior temporal cortex; PPR, attentional mechanisms (Noesselt et al., 2002). posterior parietal regions.)

5. Possible mechanisms inferred from neuroimaging for the generation of IC (Lesher and Mingolla, 1993). This findings local process may also involve some initial representation of candidate surfaces associated with the illusory figure, based on The issue of IC processing in the human visual system in relative brightness and contrast information, in particular in V1 terms of implicated brain regions and time course still remains (Haynes et al., 2004; Kinoshita and Komatsu, 2001; MacEvoy largely unresolved. Based on psychophysical and physiological et al., 1998; Rossi and Paradiso, 1999; Sasaki and Watanabe, studies, some valuable modelling attempts (e.g. Gove et al., 2004; van der Smagt et al., 2005). Psychophysical studies also 1995; Hess and Field, 1999; Lee, 2002b; Spillmann and Dresp, suggest that illusory figures might be built up by local 1995) have been made to bring together different visual mechanisms occurring in low-level areas (Dresp and Bonnet, processes, including segmentation and grouping, that might 1993; Pillow and Rubin, 2002). underlie the perception of IC. Here, we propose a summary This local vision taking place in early cortical areas can sketch of the neural mechanisms involved in IC perception occur fast, presumably 100 ms after stimulus presentation (e.g. allowing us to integrate the different neuroimaging findings Lee, 2002b). However, according to some reports, this initial described above. We suggest that two basic mechanisms might response around 100 ms in V1 may already integrate the effect critically be involved in IC perception, with distinct anatomical of ‘late’ processing mediated by reentrant or top-down and temporal characteristics (see Fig. 3). influences (e.g. Foxe and Simpson, 2002). Remarkably, according to Lee and Nguyens’ findings, the first neuronal 5.1. The fast-local low-level mechanism response to IC might occur in fact at 65–95 ms in V2 area, but only slightly later in V1 at 100–120 ms (Lee and Nguyen, The first mechanism of IC perception might constitute a 2001; Lee, 2002a). Such temporal sequence suggests that a fast-local, low-level system. For this mechanism, we assume dynamic interaction between these two visual areas might be that the early visual areas V1/V2 are the first cortical regions to crucial for illusory contours perception, presumably involving be critically involved, as suggested above by the converging some rapid feedback projections from V2 to V1 that can boost evidence from functional neuroimaging and neurophysiologi- and sharpen visual processing at a relatively local scale (Bullier cal findings. Since, the neuronal receptive fields in these low- et al., 2001; Neumann and Sepp, 1999; Roe et al., 2005), and level areas are small and spatially organized in retinotopic with short latencies post-stimulus onset. Then, segmented coordinates, their functional properties may afford an efficient features from the illusory figure might be forwarded to the detection of local details, extracted in parallel throughout the subsequent higher cortical areas in temporal and parietal lobes, visual scene by different segmentation mechanisms (Gulyas for figure recognition and integration into a global shape et al., 1998; Hess and Field, 1999; Kovacs and Julesz, 1993; percept according to the alignment of the detected local Lee et al., 1998) that subserve figure-boundary detection (Li, features (e.g. Saarinen and Levi, 2001). As suggested recently, 2003). This local vision may allow the formation of edges and this integration into a global shape could also involve early contours, even when the contours are illusory or interrupted by retinotopic visual areas (Altmann et al., 2003; Sugita, 1999; gaps (Heider et al., 2002), and therefore provide a critical stage Ursino and La Cara, 2004) by mechanisms of proximity 606 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 grouping (Han et al., 2005), but at later latencies rather than areas at late latencies, around 250–300 ms post-stimulus onset. during an initial sweep of visual inputs. This late-global high-level mechanism would generally accord These successive events at early stages of visual processing with the object-cue or the cognitive model of IC perception are in agreement with the psychophysical model of Ringach (Gregory, 1972; Piggins, 1975; Rock and Anson, 1979), and Shapley (1996) proposing that the first local mechanism suggesting that inferences about the perceived figure in high extracting information about IC may last approximately level areas might drive the formation of IC. The results of 120 ms, followed by an integration of local information into functional neuroimaging studies extend these hypotheses, by a global percept that may in turn last from 140 to 200 ms. indicating that high-level areas suggested in cognitive theories Moreover, this fast-local low-level mechanism incorporates the might specifically correspond to the LOC, usually considered earlier theoretical accounts of IC based on feature-edge as higher-tier area in functional brain imaging studies detection, which attributed a preponderant role to edge- (Mendola et al., 1999; Murray et al., 2002a). orientation processing in low-level visual areas (Smith and More generally, this second late-global mechanism seems to Over, 1975, 1977). Other computational models of vision have converge with the integrated model of vision proposed by also emphasized an essential role for the processing of local Bullier (2001), based on the functional properties of the cues before object recognition (Marr, 1976; Ullman, 1976; parvocelluar and magnocellular channels (Nowak et al., 1995) Zucker and Cavanagh, 1985). (but see Maunsell et al., 1999). According to this integrated model, but assuming a central role of LOC for IC processing 5.2. The late-global high-level mechanism instead of area V5 and the dorsal visual stream (Bullier, 2001) (although V5 area is certainly also involved in IC processing The second mechanism of IC perception constitutes a late- (Goebel et al., 1998; Kruggel et al., 2001; Seghier et al., global, high-level system. The neural substrates for this aspect 2000a)), a very fast activity might be conveyed by the of IC processing might involve higher visual areas, beyond magnocellular channel to the LOC region to generate a first- V1/V2. Functional responses to IC within this system may start pass analysis of the visual scene, then these computations done after 100 ms and be maximal around 200 ms following the at a relatively global level might be retroinjected into areas stimulus presentation (Murray et al., 2002a; Pegna et al., 2002). V1/V2 via rapid feedback connections, in order to modulate the In contrast to earlier visual areas mediating the fast-local local edge and surface processing. An important role of the mechanism, the high-level regions implicated here are magnocellular pathway in the formation of IC was also characterized by large neuronal receptive fields affording a proposed by other researchers (e.g. Li and Guo, 1995; representation of global shape structure and make use of Livingstone and Hubel, 1988; Rogers-Ramachandran and different types of grouping mechanisms. Based on results of Ramachandran, 1998). At subsequent processing stages, these functional neuroimaging studies in humans, the most important computations are combined with the information transmitted region mediating this second mechanism is probably located in by the parvocellular channel in V1/V2, leading to refined the LOC. Cortical areas in LOC have been shown to be critical processing and greater precision of the perceived visual scene for 2D or 3D shape recognition regardless of the visual cues (see also the model of Siegel and Petry, 1991). defining the object’s contour or surface (Grill-Spector et al., 2001; Vuilleumier et al., 2002). Shape recognition processes 5.3. Beyond the occipital lobe contributing to the extraction of illusory figures could be achieved by different strategies of grouping by shape similarity Following initial responses in the occipital visual network, (Han et al., 2001) and shape context (Altmann et al., 2004), and an activation of posterior parietal cortex in both the left be sensitive to the global shape of illusory figures more than to (Murray et al., 2004a) and right hemispheres (Murray et al., their local features (Achtman et al., 2003). After extraction of 2002a; Proverbio and Zani, 2002) is likely to occur in global configural cues in these areas, shape information might association with both types of IC mechanisms (fast-local and be sent back to the low-level areas in V1/V2, for completing or late-global), probably from 240 ms (Murray et al., 2004a)to strengthening the figure-ground segregation processes, e.g. by 390 ms (Proverbio and Zani, 2002) post-stimulus onset. This reconstructing the ‘missing contours’. These segregation parietal activity might relate to general attentional mechanisms processes are probably facilitated by the fine retinotopic involved in object selection and visual exploration. A role of organization of low-level areas, and constitute a crucial step for more anterior regions in the frontal lobe, particularly the the representation of a coherent figure (e.g. Edelman and orbitofrontal cortex, is also possible and was suggested by a Intrator, 2000). recent MEG study (Halgren et al., 2003). This frontal Such ‘retro-injection’ of global shape information would involvement might potentially reflect some interactions again imply an important role of feedback connections between IC perception and memory systems, as proposed by (Angelucci et al., 2002; Bullier et al., 2001; Hupe´ et al., psychophysical (Wallach and Slaughter, 1988) and ERP 1998; Ro et al., 2003) and top-down influences (Li et al., 2004) studies (Korshunova, 1999), and as also advocated for real during IC processing, but during a later time window than the object recognition in situations with degraded or partial images initial local low-level process computing luminance differ- (Bar, 2003). A putative frontal contribution might also arise in ences and edge cues. According to this model, some neuronal association with both IC mechanisms and predominantly take response to IC might therefore arise in the low-level visual place in the range of 340–400 ms post-stimulus onset (Halgren M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 607 et al., 2003), although a more rapid activation of ventral frontal of different imaging modalities (Babiloni et al., 2004) will areas to guide ongoing categorization processes in the visual certainly greatly help integrate the wide diversity of findings cortex was recently suggested on the basis of combined MEG and provide a more complete description of IC processing. and fMRI data (Kassam et al., 2003). Finally, these high level Among important questions to clarify in the future, the issue areas may potentially also influence IC processing in early of individual differences related to gender (Grabowska et al., visual areas via long-distance feedback connections (e.g. 2001; Rasmjou et al., 1999), age (Bottger et al., 2002), and Clavagnier et al., 2004). perceptual expertise should benefit from more systematic In sum, accumulating results from recent functional experimental designs and refined neuroimaging investigations, neuroimaging studies have highlighted the fact that IC including the possibility to track any corresponding differences perception may involve more than just a single brain area or in terms of anatomical networks and temporal courses. a single perceptual process, operating at a unique moment in Hemispheric dominance during IC perception is another matter time. Rather, the processing of IC seems to engage several of controversy (Atchley and Atchley, 1998; Corballis et al., brain areas at ‘early’ as well as ‘intermediate’ and perhaps 1999) requiring further functional neuroimaging studies to ‘late’ stages in the visual hierarchy. Based on these imaging assess the contribution of each hemisphere, as done for results, we propose that at least two distinct kinds of neural example for the language system (e.g. Seghier et al., 2004). mechanisms may support IC processing, involving different Even more importantly, the question of the retinotopic coding cortical networks, from the primary visual cortex through to of IC within early cortical visual areas is a fundamental issue higher extrastriate cortical areas and parieto-frontal regions, that has strong implications for our understanding of IC each with a different time-course. This framework suggests that processing level, but still needs to be more fully explored by both feedforward signals from lower-level visual areas and systematic studies using modern retinotopic mapping pro- feedback signals from higher processing stages are likely to cedures and different types of IC stimuli (Delon-Martin et al., play critical role for the integration of local features into IC and 2000). These investigations could also establish the cortical coherent shapes. Importantly, these distinct mechanisms might basis of the advantage of IC perception in the lower visual have different weight depending on the particular type of hemifield (Rubin et al., 1996). In addition, various character- stimulus and task. This flexibility might underlie the dynamic istics of IC stimuli that can potentially modulate the pattern and and complex nature of the perception of illusory contours, as magnitude of activation in different brain areas should also be well as the variability of their neural signatures in previous clarified. imaging work. In addition, our framework is consistent not Furthermore, the necessity and/or the contribution of only with previous theoretical accounts of IC, as detailed selective attention during IC processing (Gurnsey et al., above, but also with neuropsychological findings about the 1992; Montaser-Kouhsari and Rajimehr, 2004; Vuilleumier disorders of perceptual organization and shape perception and Landis, 1998) is another important debate that need to be following brain lesions (e.g. Behrmann and Kimchi, 2003). more systematically addressed by imaging studies, for instance by dissociating attentional effects from stimulus- 6. Remaining questions driven effects that can modulate neural activity at identical retinotopic locations in early visual areas. Such investigations Many important findings about IC perception have been would be useful to characterize the mechanisms of IC documented in the neuroimaging literature and we can now formation beyond purely bottom-up and low-level processes. begin to integrate them in a dynamic neural architecture, but Moreover, it has been shown with EEG that Kanizsa figures several questions still remain unresolved. The use of novel can automatically capture spatial attention when used as imaging approaches allowing high temporal and spatial visual cues (Senkowski et al., 2005), consistent with previous resolutions with better sensitivity and reliability will offer psychophysics (Davis and Driver, 1994) and neuropsycholo- new avenues to elucidate in greater details the ‘exact’ gical observations (Vuilleumier and Landis, 1998; Vuilleu- processing network of IC. For example, it has been suggested mier et al., 2001). Additional experiments are needed to that fMRI, in addition to its high spatial resolution (Kim et al., elucidate the ‘exact’ role of attentional processes and their 2000; Menon and Goodyear, 1999), might potentially achieve a interaction with IC perception, as previously done with fMRI much better temporal resolution at the level of tens of for the perception of visual objects defined by real contours milliseconds (Formisano and Goebel, 2003; Menon et al., (Corbetta and Shulman, 1998). Here again, a combined 1998; Ogawa et al., 2000; Xiong et al., 2003) and could be functional analysis of fMRI with connectivity analysis and employed to assess functional connectivity between cortical EEG should be useful to differentiate the cortical basis of top- regions (Friston et al., 2003), which may prove very valuable in down and bottom-up interactions (e.g. Mechelli et al., 2004; order to further dissect the neural dynamics of IC processing. Noesselt et al., 2003). Conversely, EEG techniques can improve their spatial Finally, the anatomical and/or temporal differences between localization capability by using new analysis algorithms that the perception of illusory and real contours (Ginsburg, 1975; allow more precise source estimation (Grave de Peralta Imber et al., 2005; Larsson et al., 1999; Ramsden et al., 2001) Menendez et al., 2004; Michel et al., 2004) or by using a could also be addressed by functional neuroimaging tech- combination of EEG with fMRI to constrain the problem of the niques, in particular to identify any additional stages used by inverse solution (Menon et al., 1997). Such combination the visual pathways to process IC. In addition, the influence 608 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 of specific physical parameters such as contrast polarity and Babiloni, F., Mattia, D., Babiloni, C., Astolfi, L., Salinari, S., Basilisco, A., shape of inducers (Lesher and Mingolla, 1993; Matthews and Rossini, P.M., Marciani, M.G., Cincotti, F., 2004. Multimodal integration Welch, 1997; Prazdny, 1985), the presence of animation or of EEG, MEG and fMRI data for the solution of the neuroimage puzzle. Magn. Reson. Imag. 22, 1471–1476. motion (Rajimehr, 2004; Ramachandran et al., 1973), the Banton, T., Levi, D.M., 1992. The perceived strength of illusory contours. modulation by current visual fixation (Dresp and Bonnet, 1991; Percept. Psychophys. 52, 676–684. Liinasuo et al., 1997), and the symmetry/asymmetry of IC Bar, M., 2003. A cortical mechanism for triggering top-down facilitation in (Albert, 1993; Proverbio and Zani, 2002), could also help visual object recognition. J. Cogn. Neurosci. 15, 600–609. characterize the response of low-level visual areas that are Baumann, R., van der Zwan, R., Peterhans, E., 1997. Figure-ground segregation at contours: a neural mechanism in the visual cortex of the sensitive to these physical parameters. Examining the neural alert monkey. Eur. J. Neurosci. 9, 1290–1303. sites modulated by adaptation phenomena (Ramachandran Behrmann, M., Kimchi, R., 2003. What does visual agnosia tell us about et al., 1994) and priming effects (Kourtzi and Kanwisher, 2001; perceptual organization and its relationship to object perception? J. Exp. Vuilleumier et al., 2002) during IC perception might also offer Psychol. Hum. Percept. Perform. 29, 19–42. a useful approach to determine the role of specific cortical Bottger, D., Herrmann, C.S., von Cramon, D.Y., 2002. Amplitude differences of evoked alpha and gamma oscillations in two different age groups. Int. regions and their selectivity for different visual characteristics, J. Psychophysiol. 45, 245–251. as previously done for real figures (Grill-Spector et al., 2001; Bradley, D.R., Dumais, S.F., 1975. Ambiguous cognitive contours. Nature 257, Vuilleumier et al., 2002). For instance, testing for adaptation to 582–584. similar figures defined by different types of contours or Brefczynski, J.A., DeYoe, E.A., 1999. A physiological correlate of the different figures defined by similar contours at identical or ‘spotlight’ of visual attention. Nature Neurosci. 2, 370–374. Brighina, F., Ricci, R., Piazza, A., Scalia, S., Gislia, G., Fierro, B., 2003. different retinal locations might help disentangle the distinct Illusory contours and specific regions of human extrastriate cortex: contributions of early visual areas and higher-level regions in evidence from rTMS. Eur. J. Neurosci. 17, 2469–2474. LOC and fusiform cortex. Other aspects that should be usefully Brigner, W.L., 1982. Mathematical model for the filling-in of illusory contour. addressed with functional neuroimaging techniques include the Percept. Motor Skill. 54, 815–820. interaction of IC perception with object recognition and Brigner, W.L., Gallagher, M.B., 1974. Subjective contour: apparent depth or simultaneous brightness contrast. Percept. Motor Skill. 38, 1047–1053. memory systems (Bar, 2003; Wallach and Slaughter, 1988) Bullier, J., 2001. Integrated model of visual processing. Brain Res. Rev. 36, and other surface segmentation mechanisms (Nakayama and 96–107. Shimojo, 1992; Sasaki and Watanabe, 2004). Bullier, J., Hupe, J.M., James, A.C., Girard, P., 2001. The role of feedback Understanding these questions, and several other intriguing connections in shaping the responses of visual cortical neurons. Prog. Brain aspects of IC, will constitute a challenging work that can Res. 134, 193–204. Clavagnier, S., Falchier, A., Kennedy, H., 2004. Long-distance feedback undoubtedly provide crucial new insights about human projections to area V1: implications for multisensory integration, spatial perception. Beyond the special interest for a fascinating visual awareness, and visual consciousness. Cogn. Affect. Behav. Neurosci. 4, phenomenon revealed by IC per se, elucidating the cerebral 117–126. mechanisms underlying IC perception should give us many Corballis, P.M., Fendrich, R., Shapley, R.M., Gazzaniga, M.S., 1999. Illusory precious cues about the way our brain can so efficiently extract contour perception and amodal boundary completion: evidence of a dissociation following callosotomy. J. Cogn. Neurosci. 11, 459–466. a rich visual world from impoverished retinal images. Corbetta, M., Shulman, G.L., 1998. Human cortical mechanisms of visual attention during orienting and search. Philos. Trans. R. Soc. Lond. B Biol. Acknowledgements Sci. 353, 1353–1362. Coren, S., 1972. Subjective contours and apparent depth. Psychol. Rev. 79, 359–367. This work was supported by grants from the Swiss National Coren, S., Porac, C., 1983. Subjective contours and apparent depth: a direct Science Foundation (No. 3200B0-100717 to MS; and No. test. Percept. Psychophys. 33, 197–200. 632.065935 to PV). Courtney, S.M., Ungerleider, L.G., 1997. What fMRI has taught us about human vision. Curr. Opin. Neurobiol. 7, 554–561. Davis, G., Driver, J., 1994. Parallel detection of Kanizsa subjective figures in References the human visual system. Nature 371, 791–793. Delon-Martin, C., Dojat, M., Seghier, M., Warnking, J., Segebarth, C., Achtman, R.L., Hess, R.F., Wang, Y.Z., 2003. Sensitivity for global shape Bullier, J., 2000. Illusory contours have retinotopic representations in V1. detection. J. Vis. 3, 616–624. Neuroimage 11, S695. Albert, M.K., 1993. Parallelism and the perception of illusory contours. Denys, K., Vanduffel, W., Fize, D., Nelissen, K., Peuskens, H., Van Essen, D., Perception 22, 589–595. Orban, G.A., 2004. The processing of visual shape in the cerebral cortex of Altmann, C.F., Bulthoff, H.H., Kourtzi, Z., 2003. Perceptual organization of human and nonhuman primates: a functional magnetic resonance imaging local elements into global shapes in the human visual cortex. Curr. Biol. 13, study. J. Neurosci. 24, 2551–2565. 342–349. Devlin, J.T., Russell, R.P., Davis, M.H., Price, C.J., Wilson, J., Moss, H.E., Altmann, C.F., Deubelius, A., Kourtzi, Z., 2004. Shape saliency modulates Matthews, P.M., Tyler, L.K., 2000. Susceptibility-induced loss of signal: contextual processing in the human lateral occipital complex. J. Cogn. comparing PET and fMRI on a semantic task. Neuroimage 11, 589–600. Neurosci. 16, 794–804. Dresp, B., Bonnet, C., 1991. Psychophysical evidence for low-level processing Angelucci, A., Levitt, J.B., Walton, E.J., Hupe, J.M., Bullier, J., Lund, J.S., of illusory contours and surfaces in the Kanizsa square. Vis. Res. 31, 1813– 2002. Circuits for local and global signal integration in primary visual 1817. cortex. J. Neurosci. 22, 8633–8646. Dresp, B., Bonnet, C., 1993. Psychophysical measures of illusory Atchley, R.A., Atchley, P., 1998. Hemispheric specialization in the detection of form perception: further evidence for local mechanisms. Vis. Res. 33, subjective objects. Neuropsychology 36, 1373–1386. 759–766. M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 609

Dresp, B., Spillmann, L., 2001. The brain may know more than cognitive Gulyas, B., Cowey, A., Heywood, C.A., Popplewell, D., Roland, P.E., 1998. theory can tell us: a reply to Ted Parks. Perception 30, 633–636. Visual form discrimination from texture cues: a PET study. Hum. Brain Dupont, P., Sary, G., Peuskens, H., Orban, G.A., 2003. Cerebral regions Mapp. 6, 115–127. processing first- and higher-order motion in an opposed-direction Gurnsey, R., Humphrey, G.K., Kapitan, P., 1992. Parallel discrimination of discrimination task. Eur. J. Neurosci. 17, 1509–1517. subjective contours defined by offset gratings. Percept. Psychophys. 52, Edelman, S., Intrator, N., 2000. (Coarse coding of shape fragments) C 263–276. (retinotopy) approximatelyZrepresentation of structure. Spat. Vis. 13, Gurnsey, R., Poirier, J.A.M., Gascon, E., 1996. There is no evidence that 255–264. Kanizsa-type subjective contours can be detected in parallel. Perception 25, Engel, S.A., Glover, G.H., Wandell, B.A., 1997. Retinotopic organisation in 861–874. human visual cortex and the spatial precision of fMRI. Cereb. Cortex 7, Halgren, E., Mendola, J., Chong, C.D.R., Dale, A.M., 2003. Cortical activation 181–192. to illusory shapes as measured with magnetoencephalography. Neuroimage Enns, J.T., Rensink, R.A., 1993. A model for rapid interpretation of line 18, 1001–1009. drawings in early vision. In: Brogan, D., Gale, A., Carr, K. (Eds.), Visual Halpern, D.F., 1981. The determinants of illusory contour perception. Search II: Proceedings of the Second International Conference on Visual Perception 10, 199–213. Search. Taylor and Francis, London, pp. 73–89. Hamalainen, M.S., Sarvas, J., 1989. Realistic conudctivity geometry model of Evans, M.A., Shedden, J.M., Hevenor, S.J., Hahn, M.C., 2000. The effect of the human head for interpretation of neuromagnetic data. IEEE Trans. variability of unattended information on global and local processing: Biomed. Eng. 36, 165–171. evidence for lateralization at early stages of processing. Neuropsychology Han, S., Song, Y., Ding, Y., Yund, E.W., Woods, D.L., 2001. Neural substrates 38, 225–239. for visual perceptual grouping in humans. Psychophysiology 38, 926–935. Ffytche, D.H., Zeki, S., 1996. Brain activity related to the perception of illusory Han, S., Weaver, J.A., Murray, S.O., Kang, X., Yund, E.W., Woods, D.L., contours. Neuroimage 3, 104–108. 2002. Hemispheric asymmetry in global/local processing: effects of Fierro, B., Ricci, R., Piazza, A., Scalia, S., Giglia, G., Vitello, G., Brighina, F., stimulus position and spatial frequency. Neuroimage 17, 1290–1299. 2003. 1 Hz rTMS enhances extrastriate cortex activity in migraine: Han, S., Jiang, Y., Mao, L., Humphreys, G.W., H., Gu, 2005. Attentional evidence of a reduced inhibition? Neurology 61, 1446–1448. modulation of perceptual grouping in human visual cortex: Functional MRI Fink, G.R., Halligan, P.W., Marshall, J.C., Frith, C.D., Frackowiak, R.S., studies. Hum. Brain Mapp. 25, 424–432. Dolan, R.J., 1996. Where in the brain does visual attention select the forest Hasnain, M.K., Fox, P.T., Woldorff, M.G., 1998. Intersubject variability of and the trees? Nature 382, 626–628. functional areas in the human visual cortex. Hum. Brain Mapp. 6, 301–315. Formisano, E., Goebel, R., 2003. Tracking cognitive processes with functional Hayashi, E., Kuroiwa, Y., Omoto, S., Kamitani, T., Li, M., Koyano, S., 2004. MRI mental chronometry. Curr. Opin. Neurobiol. 13, 174–181. Visual evoked potential changes related to illusory perception in normal Foxe, J.J., Simpson, G.V., 2002. Flow of activation from V1 to frontal cortex in human subjects. Neurosci. Lett. 359, 29–32. humans. A framework for defining ‘early’ visual processing. Exp. Brain Haynes, J.D., Lotto, R.B., Rees, G., 2004. Responses of human visual cortex to Res. 142, 139–150. uniform surfaces. Proc. Natl Acad. Sci. USA 101, 4286–4291. Foxe, J.J., Murray, M.M., Javitt, D.C., 2005. Filling-in in Schizophrenia: a Heider, B., 1996. Sex-related engagement of the hemispheres in visuospatial high-density electrical mapping and source-analysis investigation of processing. Int. J. Neurosci. 88, 83–95. illusory contour processing. Cereb. Cortex. 15, 1914–1927. Heider, B., Meskenaite, V., Peterhans, E., 2000. Anatomy and physiology of a Friston, K.J., Holmes, A., Poline, J.B., Price, C.J., Frith, C.D., 1996. Detecting neural mechanism defining depth order and contrast polarity at illusory activations in PET and fMRI: levels of inference and power. Neuroimage 4, contours. Eur. J. Neurosci. 12, 4117–4130. 223–235. Heider, B., Spillmann, L., Peterhans, E., 2002. Stereoscopic illusory contours— Friston, K.J., Harrison, L., Penny, W., 2003. Dynamic causal modelling. cortical neuron responses and human perception. J. Cogn. Neurosci. 14, Neuroimage 19, 1273–1302. 1018–1029. Ginsburg, A.P., 1975. Is the illusory triangle physical or imaginary? Nature Herrmann, C.S., Bosch, V., 2001. Gestalt perception modulates early visual 257, 219–220. Goebel, R., Khorram-Sefat, D., Muckli, L., Hacker, H., Singer, W., 1998. The processing. NeuroReport 12, 901–904. constructive nature of vision: direct evidence from functional magnetic Herrmann, C.S., Mecklinger, A., 2000. Magnetoencephalographic responses to resonance imaging studies of apparent motion and motion imagery. Eur. illusory figures: early evoked gamma is affected by processing of stimulus J. Neurosci. 10, 1563–1573. features. Int. J. Psychophysiol. 38, 265–281. Gove, A., Grossberg, S., Mingolla, E., 1995. Brightness perception, illusory Herrmann, C.S., Mecklinger, A., Pfeifer, E., 1999. Gamma responses and ERPs contours, and corticogeniculate feedback. Vis. Neurosci. 12, 1027–1052. in a visual classification task. Clin. Neurophysiol. 110, 636–642. Grabowska, A., Nowicka, A., Szymansk, O., Szatkowski, I., 2001. Subjective Hess, R., Field, D., 1999. Integration of contours: new insights. Trends Cogn. contour illusion: sex-related effect of unilateral brain damage. NeuroReport Sci. 3, 480–486. 12, 2289–2292. Hirsch, J., DeLaPaz, R.L., Relkins, N.R., Victor, J., Kim, K., Li, T., Borden, P., Grave de Peralta Menendez, R., Murray, M.M., Michel, C.M., Martuzzi, R., Rubin, N., Shapley, R., 1995. Illusory contours activate specific regions in Gonzalez Andino, S.L., 2004. Electrical neuroimaging based on biophysi- human visual cortex: evidence from functional magnetic resonance cal constraints. Neuroimage 21, 527–539. imaging. Proc. Natl Acad. Sci. USA 92, 6469–6473. Gregory, R.L., 1972. Cognitive contours. Nature 238, 51–52. Hirsch, J., DeLaPaz, R., Kim, K., Victor, J., Relkin, N., Lee, K.M., Moreno, D., Grice, S.J., Haan, M.D., Halit, H., Johnson, M.H., Csibra, G., Grant, J., Rubin, N., Shapley, R.M., 1996. FMRI and FSE images indicate that Karmiloff-Smith, A., 2003. ERP abnormalities of illusory contour cortical areas activated exclusively by illusory contours and stereo depth lie perception in Williams syndrome. NeuroReport 14, 1773–1777. outside primary visual cortex. Proc. Int. Soc. Magn. Res. Imag. 3, 1857. Grill-Spector, K., Kourtzi, Z., Kanwisher, N., 2001. The lateral occipital Hupe´, J.M., James, A.C., Payne, B.R., Lomber, S.G., Girard, P., Bullier, J., complex and its role in object recognition. Vis. Res. 41, 1409–1422. 1998. Cortical feedback improves discrimination between figure and Grossberg, S., 1980. How does a brain build a cognitive code? Psychol. Rev. background by V1, V2 and V3 neurons. Nature 394, 784–787. 87, 1–51. Huxlin, K.R., Saunders, R.C., Marchionini, D., Pham, H.A., Merigan, W.H., Grossberg, S., Mingolla, E., 1985. Neural dynamics of form perception: 2000. Perceptual deficits afetr lesions of inferotemporal cortex in boundary completion, illusory figures, and neo color spreading. Psychol. macaques. Cereb. Cortex 10, 671–683. Rev. 92, 173–211. Iidaka, T., Yamashita, K., Kashikura, K., Yonekura, Y., 2004. Spatial Grossberg, S., Mingolla, E., Ross, W.D., 1997. Visual brain and visual frequency of visual image modulates neural responses in the temporo- perception: how does the cortex do perceptual grouping? Trends Neurosci. occipital lobe. An investigation with event-related fMRI. Brain Res. Cogn. 20, 106–111. Brain Res. 18, 196–204. 610 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612

Imber, M.L., Shapley, R.M., Rubin, N., 2005. Differences in real and illusory Lesher, G.W., Mingolla, E., 1993. The role of edges and line-ends in illusory shape perception revealed by backward masking. Vis. Res. 45, 91–102. contour formation. Vision Res. 33, 2253–2270. Jory, M.K., Day, R.H., 1979. The relationship between brightness contrast and Leventhal, A.G., Wang, Y., Schmolesky, M.T., Zhou, Y., 1998. Neural illusory contours. Perception 8, 3–9. correlates of boundary perception. Vis. Neurosci. 15, 1107–1118. Kaiser, J., Buhler, M., Lutzenberger, W., 2004. Magnetoencephalographic Li, Z., 2003. V1 mechanisms and some figure-ground and border effects. gamma-band responses to illusory triangles in humans. Neuroimage 23, J. Physiol. Paris 97, 503–515. 551–560. Li, C.Y., Guo, K., 1995. Measurements of geometric illusions, illusory contours Kanizsa, G., 1976. Subjective contours. Sci. Am. 234, 48–52. and stereo-depth at luminance and colour contrast. Vis. Res. 35, 1713– Kanizsa, G., 1979. Organization in Vision. Essays on Gestalt Perception. 1720. Praeger, New York. Li, W., Piech, V., Gilbert, C.D., 2004. Perceptual learning and top-down Kassam, K.S., Aminoff, E., Bar, M., 2003. Spatial–temporal cortical processing influences in primary visual cortex. Nat. Neurosci. 7, 651–657. of contextual associations. Soc. Neurosci. Abstr. 128, 8. Liinasuo, M., Rovamo, J., Kojo, I., 1997. Effects of spatial configuration and Keil, A., Mu¨ller, M.M., Ray, W.J., Gruber, T., Elbert, T., 1999. Human gamma number of fixations on Kanizsa triangle detection. Invest. Ophthalmol. Vis. band activity and perception of gestalt. J. Neurosci. 19, 7152–7161. Sci. 38, 2554–2565. Kennedy, J.M., 1988. Line endings and subjective contours. Spat. Vis. 3, Livingstone, M., Hubel, D., 1988. Segregation of form, color, movement and 151–158. depth: anatomy, physiology, and perception. Science 240, 740–749. Kim, D.S., Duong, T.Q., Kim, S.G., 2000. High-resolution mapping of iso- Loring, D.W., Meador, K.J., Allison, J.D., Pillai, J.J., Lavin, T., Lee, G.P., orientation columns by fMRI. Nature Neurosci. 3, 164–169. Balan, A., Dave, V., 2002. Now you see it, now you don’t: statistical and Kinoshita, M., Komatsu, H., 2001. Neural representation of the luminance and methodological considerations in fMRI. Epilepsy Behav. 3, 539–547. brightness of a uniform surface in the macaque primary visual cortex. MacEvoy, S.P., Kim, W., Paradiso, M.A., 1998. Integration of surface J. Neurophysiol. 86, 2559–2570. information in primary visual cortex. Nat. Neurosci. 1, 616–620. Kojo, I., Liinasuo, M., Rovamo, J., 1993. Spatial and temporal properties of Maertens, M., Pollmann, S., 2005. FMRI reveals a common neural substrate of illusory figures. Vision Res. 33, 897–901. illusory and real contours in v1 after perceptual learning. J. Cogn. Neurosci. Koles, Z.J., 1998. Trends in EEG source localization. Electroencephalogr. Clin. 17, 1553–1564. Neurophysiol. 106, 127–137. Malach, R., Reppas, J.B., Benson, R.R., Kwong, K.K., Jiang, H., Kennedy, W.A., Korshunova, S.G., 1999. Visual evoked potentials induced by illusory outlines Ledden, P.J., Brady, T.J., Rosen, B.R., Tootell, R.B.H., 1995. Object-related (Kanizsa’s square). Neurosci. Behav. Physiol. 29, 695–701. activity revealed by functional magnetic resonance imaging in human Kosslyn, S.M., Thompson, W.L., Kim, I.J., Alpert, N.M., 1995. Topographical occipital cortex. Proc. Natl Acad. Sci. USA 92, 8135–8139. representations of mental images in primary visual cortex. Nature 378, Marcus, D.S., Van Essen, D.C., 2002. Scene segmentation and attention in 496–498. primate cortical areas V1 and V2. J. Neurophysiol. 88, 2648–2658. Kourtzi, Z., Kanwisher, N., 2001. Representation of perceived object shape by Marr, D., 1976. Early processing of visual information. Philos. Trans. R Soc. the human lateral occipital complex. Science 293, 1506–1509. Lond. B Biol. Sci. 275, 483–519. Kovacs, I., Julesz, B., 1993. A closed curve is much more than an incomplete Marr, D., 1982. A Computational Investigation into the Human Representation one: effect of closure in figure-ground segmentation. Proc. Natl Acad. Sci. and Processing of Visual Information. Freeman, San Fransisco. USA 90, 7495–7497. Matthews, N., Welch, L., 1997. The effect of inducer polarity and contrast on Kovacs, G., Vogels, R., Orban, G.A., 1995. Selectivity of macaque inferior the perception of illusory figures. Perception 26, 1431–1443. temporal neurons for partially occluded shapes. J. Neurosci. 15, 1984– Maunsell, J.H., Ghose, G.M., Assad, J.A., McAdams, C.J., Boudreau, C.E., 1997. Noerager, B.D., 1999. Visual response latencies of magnocellular Kruggel, F., Herrmann, C.S., Wiggins, C.J., von Cramon, D.Y., 2001. and parvocellular LGN neurons in macaque monkeys. Vis. Neurosci. Hemodynamic and electroencephalographic responses to illusory figures: 16, 1–14. recording of the evoked potentials during functional MRI. Neuroimage 14, McCourt, M.E., Paulson, K., 1994. The influence of illusory contours on 1327–1336. the detection of luminance increments and decrements. Vis. Res. 34, 2469– Lamme, V.A.F., 1995. The neurophysiology of figure-ground segregation in 2475. primary visual cortex. J. Neurosci. 15, 1605–1615. Mechelli, A., Price, C.J., Friston, K.J., Ishai, A., 2004. Where bottom-up meets Lamme, V.A.F., van Dijk, B.W., Spekreijse, H., 1993. Contour from motion top-down: neuronal interactions during perception and imagery. Cereb. processing occurs in primary visual cortex. Nature 363, 541–543. Cortex 14, 1256–1265. Larsson, J., Amunts, K., Gulya`s, B., Malikovic, A., Zilles, K., Roland, P.E., Mendola, J.D., Dale, A.M., Fischl, B., Liu, A.K., Tootell, R.B.H., 1999. The 1999. Neuronal correlates of real and illusory contour perception: representation of illusory and real contours in human cortical visual areas functional anatomy with PET. Eur. J. Neurosci. 11, 4024–4036. revealed by functional magnetic resonance imaging. J. Neurosci. 19, 8560– Lee, T.S., 2002a. Top-down influence in early visual processing: a Bayesian 8572. perspective. Physiol. Behav. 77, 645–650. Menon, R.S., Goodyear, B.G., 1999. Submillimeter functional localization in Lee, T.S., 2002b. The nature of illusory contour computation. Neuron 33, 667– human striate cortex using bold contrast at 4 Tesla: implications for the 668. vascular point-spread function. Magn. Res. Med. 41, 230–235. Lee, T.S., Nguyen, M., 2001. Dynamics of subjective contour formation in the Menon, V., Ford, J.M., Lim, K.O., Glover, G.H., Pfefferbaum, A., 1997. early visual cortex. Proc. Natl Acad. Sci. USA 98, 1907–1911. Cobined event-related fMRI and EEG evidence for temporal–parietal Lee, J.H., van Donkelaar, P., 2002. Dorsal and ventral visual stream cortex activation during target detection. NeuroReport 8, 3029–3037. contributions to perception–action interactions during pointing. Exp. Menon, R.S., Luckowsky, D.L., Gati, J.S., 1998. Mental chronometry using Brain Res. 143, 440–446. latency-resolved functional magnetic resonance imaging. Proc. Natl Acad. Lee, T.S., Mumford, D., Romero, R., Lamme, V.A.F., 1998. The role of the Sci. USA 95, 10902–10907. primary visual cortex in higher level vision. Vis. Res. 38, 2429–2454. Michel, C.M., Seeck, M., Landis, T., 1999. Spatio-temporal dynamics of Lee, T.S., Yang, C.F., Romero, R.D., Mumford, D., 2002. Neural activity in human cognition. News Physiol. Sci. 14, 206–214. early visual cortex reflects behavioral experience and higher-order Michel, C.M., Murray, M.M., Lantz, G., Gonzalez, S., Spinelli, L., Grave de perceptual saliency. Nat. Neurosci. 5, 589–597. Peralta, R., 2004. EEG source imaging. Clin. Neurophysiol. 115, 2195– Lehmann, D., 1987. Principles of spatial analysis. In: Gevins, A.S., Remond, A. 2222. (Eds.), Handbook of Electroencephalography and Clinical Neurophysiol- Momjian, S., Seghier, M., Seek, M., Michel, C.M., 2003. Mapping the neuronal ogy. Methods of Analysis of Brain Electrical and Magnetic Signals. networks of human cortical brain functions. Adv. Tech. Stand. Neurosurg. Elsevier, Amsterdam, pp. 309–354. 28, 91–142. M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612 611

Montaser-Kouhsari, L., Rajimehr, R., 2004. Attentional modulation of Petry, S., Meyer, G.E., 1987. The Perception of Illusory Contours. Springer, adaptation to illusory lines. J. Vis. 4, 434–444. Berlin. Muckli, L., Kohler, A., Kriegeskorte, N., Singer, W., 2005. Primary visual Peyrin, C., Baciu, M., Segebarth, C., Marendaz, C., 2004. Cerebral regions and cortex activity along the apparent-motion trace reflects illusory perception. hemispheric specialization for processing spatial frequencies during natural PLoS Biol. 3, e265. scene recognition. An event-related fMRI study. Neuroimage 23, 698–707. Murray, M.M., Wylie, G.R., Higgins, B.A., Javitt, D.C., Schroeder, C.E., Picton, T.W., Lins, O., Scherg, M., 1995. The recording and analysis of event- Foxe, J.J., 2002a. The spatiotemporal dynamics of illusory contour related potentials. In: Boller, F., Grafman, J., Johnson, R. (Eds.), Handbook processing: combined high-density electrical mapping, source analysis, of Neuropsychology. Event-Related Brain Potentials and Cognition. and functional magnetic resonance imaging. J. Neurosci. 22, 5055–5073. Elsevier, Amsterdam, pp. 3–73. Murray, S.O., Kersten, D., Olshausen, B.A., Schrater, P., Woods, D.L., 2002b. Picton, T.W., Bentin, S., Berg, P., Donchin, E., Hillyard, S.A., Johnson, R.J., Shape perception reduces activity in human primary visual cortex. Proc. Miller, G.A., Ritter, W., Ruchkin, D.S., Rugg, M.D., Taylor, M.J., 2000. Natl Acad. Sci. USA 99, 15164–15169. Guidelines for using human event-related potentials to study cognition: Murray, M.M., Foxe, D.M., Javitt, D.C., Foxe, J.J., 2004a. Setting boundaries: recording standards and publication criteria. Psychophysiology 37, 127– brain dynamics of modal and amodal illusory shape completion in humans. 152. J. Neurosci. 24, 6898–6903. Piggins, D.J., 1975. Cognitive space. Perception 4, 337–340. Murray, S.O., Schrater, P., Kersten, D., 2004b. Perceptual grouping and the Pillow, J., Rubin, N., 2002. Perceptual completion across the vertical meridian interactions between visual cortical areas. Neural Netw. 17, 695–705. and the role of early visual cortex. Neuron 33, 805–813. Nakayama, K., Shimojo, S., 1992. Experiencing and perceiving visual surfaces. Prazdny, K., 1985. On the nature of inducing forms generating perceptions of Science 257, 1357–1363. illusory contours. Percept. Psychophys. 37, 237–242. Neumann, H., Sepp, W., 1999. Recurrent V1–V2 interaction in early visual Proverbio, A.M., Zani, A., 2002. Electrophysiological indexes of illusory boundary processing. Biol. Cybern. 81, 425–444. contours perception in humans. Neuropsychology 40, 479–491. Ni, R., Wang, Z., Wu, X., Wang, Y., Li, D., 2003. Perception of illusory contours enhanced in motion. Sci. Chin 46, 570–583. Purghe, F., 1995. Illusory figures from stereoscopically three-dimensional Nieder, A., 2002. Seeing more than meets the eye: processing of illusory inducers depicting no occlusion event. Perception 24, 905–918. contours in animals. J. Comp. Physiol. A 188, 249–260. Purghe´, F., Coren, S., 1992. Subjective contours 1900–1990: research trends Noesselt, T., Hillyard, S.A., Woldorff, M.G., Schoenfeld, A., Hagner, T., and bibliography. Percept. Psychophys. 51, 291–304. Jancke, L., Tempelmann, C., Hinrichs, H., Heinze, H.J., 2002. Delayed Rajimehr, R., 2004. Static motion aftereffect does not modulate positional striate cortical activation during spatial attention. Neuron 35, 575–587. representations in early visual areas. Cogn. Brain Res. 20, 323–327. Noesselt, T., Shah, N.J., Jancke, L., 2003. Top-down and bottom-up Ramachandran, V.S., Rao, V.M., Vidyasagar, T.R., 1973. Apparent motion modulation of language related areas—an fMRI study. BMC Neurosci. 4, with subjective contours. Vis. Res. 13, 1399–1401. 13. Ramachandran, V.S., Ruskin, D., Cobb, S., Rogers-Ramachandran, D., Nowak, L.G., Munk, M.H., Girard, P., Bullier, J., 1995. Visual latencies in Tyler, C.W., 1994. On the perception of illusory contours. Vision Res. areas V1 and V2 of the macaque monkey. Vis. Neurosci. 12, 371–384. 34, 3145–3152. Ogawa, S., Lee, T.M., Stepnoski, R., Chen, W., Zhu, X.H., Ugurbil, K., 2000. Ramsden, B.M., Hung, C.P., Roe, A.W., 2001. Real and illusory contour An approch to probe some neural systems interaction by functional MRI at processing in area V1 of the primate: a cortical balancing act. Cereb. Cortex neural time scale down to milliseconds. Proc. Natl Acad. Sci. USA 97, 11, 648–665. 11026–11031. Rasmjou, S., Hausmann, M., Gunturkun, O., 1999. Hemispheric dominance Ohtani, Y., Okamura, S., Shibasaki, T., Arakawa, A., Yoshida, Y., Toyama, K., and gender in the perception of an illusion. Neuropsychologia 37, 1041– Ejima, Y., 2002. Magnetic responses of human visual cortex to illusory 1047. contours. Neurosci. Lett. 321, 173–176. Redies, C., Crook, J.M., Creutzfeldt, O.D., 1986. Neuronal responses to borders Ohzawa, I., 1999. Do animals see what we see? Nat. Neurosci. 2, 586–588. with and without luminance gradients in cat visual cortex and dorsal lateral Ojemann, J.G., Buckner, R.L., Akbudak, E., Snyder, A.Z., Ollinger, J.M., geniculate nucleus. Exp. Brain Res. 61, 469–481. McKinstry, R.C., Rosen, B.R., Petersen, S.E., Raichle, M.E., Conturo, T.E., Ringach, D.L., Shapley, R., 1996. Spatial and temporal properties of illusory 1998. Functional MRI studies of word-stem completion: reliability across contours and amodal boundary completion. Vis. Res. 36, 3037–3050. laboratories and comparison to blood flow imaging with PET. Hum. Brain Ritzl, A., Marshall, J.C., Weiss, P.H., Zafiris, O., Shah, N.J., Zilles, K., Mapp. 6, 203–215. Fink, G.R., 2003. Functional anatomy and differential time courses of Orban, G.A., Dupont, P., De Bruyn, B., Vogels, R., Vandenberghe, R., neural processing for explicit, inferred, and illusory contours. An event Mortelmans, L., 1995. A motion area in human visual cortex. Proc. Natl related fMRI study. Neuroimage 19, 1567–1577. Acad. Sci. USA 92, 993–997. Ro, T., Breitmeyer, B., Burton, P., Singhal, N.S., Lane, D., 2003. Feedback Palva, M., Heslenfeld, D., Aronen, H.J., Ilmoniemi, R.J., 1997. Early visual contributions to visual awareness in human occipital cortex. Curr. Biol. 13, processing of illusory and real contours studied with fMRI. Int. 1038–1041. J. Psychophysiol. 25, 52. Rock, I., Anson, R., 1979. Illusory contours as the solution to a problem. Parks, T.E., 1979. Subjective figures: does brithness enhancement depend on Perception 8, 665–681. subjective boundary definition. Percept. Psychophys. 26, 419. Roe, A.W., Lu, H.D., Hung, C.P., 2005. Cortical processing of a brightness Parks, T.E., 1989. Illusory-figure lightness: evidence for a two-component illusion. Proc. Natl Acad. Sci. USA 102, 3869–3874. theory. Perception 18, 783–788. Rogers-Ramachandran, D.C., Ramachandran, V.S., 1998. Psychophysical Parks, T.E., 2001. Rock’s cognitive theory of illusory figures: a commentary. Perception 30, 627–631. evidence for boundary and surface systems in human vision. Vis. Res. Pegna, A.J., Khateb, A., Murray, M.M., Landis, T., Michel, C.M., 2002. Neural 38, 71–77. processing of illusory and real contours revealed by high-density ERP Rossi, A.F., Paradiso, M.A., 1999. Neural correlates of perceived brightness in mapping. NeuroReport 13, 965–968. the retina, lateral geniculate nucleus, and striate cortex. J. Neurosci. 19, Peterhans, E., Heitger, F., 2001. Simulation of neuronal responses defining 6145–6156. depth order and contrast polarity at illusory contours in monkey area V2. Rowbury, C.R., 1982. A comment on Day and Jory’s note on a second stage in J. Comput. Neurosci. 10, 195–211. the formation of illusory contours. Percept. Psychophys. 31, 395–396. Peterhans, E., von der Heydt, R., 1989. Mechanisms of contour perception in Rubin, N., Nakayama, K., Shapley, R., 1996. Enhanced perception of illusory monkey visual cortex. II. Contours bridging gaps. J. Neurosci. 9, 1749–1763. contours in the lower versus upper visual hemifields. Science 271, 651–653. Peterhans, E., von der Heydt, R., 1991. Subjective contours—bridging the gap Saarinen, J., Levi, D.M., 2001. Integration of local features into a global shape. between psychophysics and physiology. Trends Neurosci. 14, 112–119. Vis. Res. 41, 1785–1790. 612 M.L. Seghier, P. Vuilleumier / Neuroscience and Biobehavioral Reviews 30 (2006) 595–612

Sadato, N., Yonekura, Y., Yamada, H., Nakamura, S., Waki, A., Ishii, Y., 1998. Tootell, R.B.H., Mendola, J.D., Hadjikhani, N.K., Liu, A.K., Dale, A.M., Activation patterns of covert word generation detected by fMRI: 1998b. The representation of the ipsilateral visual field in human cerebral comparison with 3D PET. J. Comput. Assist. Tomogr. 22, 945–952. cortex. Proc. Natl Acad. Sci. USA 95, 818–824. Sarti, A., Malladi, R., Sethian, J.A., 2000. Subjective surfaces: a method for Tsao, D.Y., Freiwald, W.A., Knutsen, T.A., Mandeville, J.B., Tootell, R.B., completing missing boundaries. Proc. Natl Acad. Sci. USA 97, 6258–6263. 2003. Faces and objects in macaque cerebral cortex. Nat. Neurosci. 6, Sasaki, Y., Watanabe, T., 2004. The primary visual cortex fills in color. Proc. 689–695. Natl Acad. Sci. USA 101, 18251–18256. Ullman, S., 1976. On visual detection of light sources. Biol. Cybern. 21, Sasaki, Y., Hadjikhani, N., Fischl, B., Liu, A.K., Marrett, S., Dale, A.M., 205–212. Tootell, R.B., 2001. Local and global attention are mapped retinotopically Ursino, M., La Cara, G.E., 2004. A model of contextual interactions and in human occipital cortex. Proc. Natl Acad. Sci. USA 98, 2077–2082. contour detection in primary visual cortex. Neural Netw. 17, 719–735. Schwartz, S., Maquet, P., Frith, C., 2002. Neural correlates of perceptual van der Smagt, M.J., Wehrhahn, C., Albright, T.D., 2005. Contextual masking learning: a functional MRI study of visual texture discrimination. Proc. Natl of oriented lines: interactions between surface segmentation cues. Acad. Sci. USA 99, 17137–17142. J. Neurophysiol. 94, 576–589. Seghier, M., Dojat, M., Delon-Martin, C., Rubin, C., Warnking, J., von der Heydt, R., Peterhans, E., 1989. Mechanisms of contour perception in Segebarth, C., Bullier, J., 2000a. Moving illusory contours activate primary monkey visual cortex. I. Lines of pattern discontinuity. J. Neurosci. 9, visual cortex: an fMRI study. Cereb. Cortex 10, 663–670. 1731–1748. Seghier, M., Dojat, M., Delon-Martin, C., Segebarth, C., Bullier, J., 2000b. von der Heydt, R., Peterhans, E., Baumgartner, G., 1984. Illusory contours and Motion of illusory contours enhances activation in V1. Neuroimage 11, cortical neuron responses. Science 224, 1260–1262. S697. van Oostende, S., Sunaert, S., van Hecke, P., Marchal, G., Orban, G.A., 1997. Seghier, M.L., Lazeyras, F., Pegna, A.J., Annoni, J.M., Zimine, I., Mayer, E., Michel, C.M., Khateb, A., 2004. Variability of fMRI activation during a The kinetic occipital (KO) area in human visual cortex: a fMRI study. phonological and semantic language task in healthy subjects. Hum. Brain Cereb. Cortex 7, 690–701. Mapp. 23, 140–155. Vuilleumier, P., Landis, T., 1998. Illusory contours and spatial neglect. Senkowski, D., Rottger, S., Grimm, S., Foxe, J.J., Herrmann, C.S., 2005. NeuroReport 9, 2481–2484. Kanizsa subjective figures capture visual spatial attention: evidence from Vuilleumier, P., Valenza, N., Landis, T., 2001. Explicit and implicit perception electrophysiological and behavioral data. Neuropsychologia 43, 872–886. of illusory contours in unilateral spatial neglect: behavioural and Sheth, B.R., Sharma, J., Rao, S.C., Sur, M., 1996. Orientation maps of anatomical correlates of preattentive grouping mechanisms. Neuropsycho- subjective contours in visual cortex. Science 274, 2110–2215. logy 39, 597–610. Shipley, T., Kellman, P., 1992. Strength of visual interpolation depends on the Vuilleumier, P., Henson, R.N., Driver, J., Dolan, R.J., 2002. Multiple levels of physically specified to total edge length. Percept. Psychophys. 52, 97–106. visual object constancy revealed by event-related fMRI of repetition Siegel, S., Petry, S., 1991. Evidence for independent processing of subjective priming. Nat. Neurosci. 5, 491–499. contour brightness and sharpness. Perception 20, 233–241. Wallach, H., Slaughter, V., 1988. The role of memory in perceiving subjective Singh, M., 2004. Modal and amodal completion generate different shapes. contours. Percept. Psychophys. 43, 101–106. Psychol. Sci. 15, 454–459. Ware, C., Kennedy, J.M., 1978. Perception of subjective lines, surfaces, and Skottun, B.C., 1994. Illusory contours and linear filters. Exp. Brain Res. 100, volumes in 3-dimensional constructions. Leonardo 11, 111–114. 360–364. Wasserstein, J., Zappulla, R., Rosen, J., Gerstman, L., Rock, D., 1987. In search Slotnick, S.D., Thompson, W.L., Kosslyn, S.M., 2005. Visual mental imagery of closure: subjective contour illusions, Gestalt completion tests, and induces retinotopically organized activation of early visual areas. Cereb. implications. Brain Cogn. 6, 1–14. Cortex 15, 1570–1583. Watanabe, T., Oyama, T., 1988. Are illusory contours a cause or a consequence Smith, A.T., Over, R., 1975. Tilt aftereffects with subjective contours. Nature of apparent differences in brightness and depth in the Kanizsa square? 257, 581–582. Perception 17, 513–521. Smith, A.T., Over, R., 1977. Orientation masking and the tilt illusion with Watt, R.J., Phillips, W.A., 2000. The function of dynamic grouping in vision. subjective contours. Perception 6, 441–447. Trends Cogn. Sci. 4, 447–454. Smith, A.T., Greenlee, M.W., Singh, K.D., Kraemer, F.M., Hennig, J., 1998. Weissman, D.H., Mangun, G.R., Woldorff, M.G., 2002. A role for top-down The processing of first- and second-order motion in human visual cortex attentional orienting during interference between global and local aspects assessed by functional magnetic resonance imaging (fMRI). J. Neurosci. of hierarchical stimuli. Neuroimage 17, 1266–1276. 18, 3816–3830. Wenderoth, P., 1997. Comments on Parks’s ‘Prior experience of form and Spillmann, L., Dresp, B., 1995. Phenomena of illusory form: can we bridge the illusory figures: new demonstrations’. Perception 26, 97–100. gap between levels of explanation? Perception 24, 1333–1364. Werner, H., 1935. Studies on contour. Am. J. Psychol. 47, 40–64. Stanley, D.A., Rubin, N., 2003. fMRI activation in response to illusory contours and Westheimer, G., 1994. Illusory figures and real neurons. Nature 371, 745–746. salient regions in the human lateral occipital complex. Neuron 37, 323–331. Xiong, J., Fox, P.T., Gao, J.H., 2003. Directly mapping magnetic field effects of Sugawara, M., Morotomi, T., 1991. Visual evoked potentials elicited by neuronal activity by magnetic resonance imaging. Hum. Brain Mapp. 20, subjective contour figures. Scand. J. Psychol. 32, 352–357. Sugita, Y., 1999. Grouping of image fragments in primary visual cortex. Nature 41–49. 401, 269–272. Yin, C., Shimojo, S., Moore, C., Engel, S.A., 2002. Dynamic shape integration Sunaert, S., Van Hecke, P., Marchal, G., Orban, G.A., 1999. Motion-reponsive in extrastriate cortex. Curr. Biol. 12, 1379–1385. regions of the human brain. Exp. Brain Res. 127, 355–370. Zanforlin, M., 1981. Visual perception of complex forms in chicks. Ital. Tallon-Baudry, C., Bertrand, O., Delpuech, C., Pernier, J., 1997. Oscillatory J. Psychol. 8, 1–16. g-band (30–70 Hz) activity induced by visual search task in humans. Zeki, S., 2004. The neurology of ambiguity. Conscious. Cogn. 13, 173–196. J. Neurosci. 17, 722–734. Zeki, S., Perry, R.J., Bartels, A., 2003. The processing of kinetic contours in the Toga, A.W., Mazziotta, J.C., 1996. Brain Mapping: the Methods. Academic brain. Cereb. Cortex 13, 189–202. Press, London. Zhou, Y., Jia, F., Tao, H., Shou, T., 2001. The responses to illusory contours of Tootell, R.B.H., Hadjikhani, N.K., Vanduffel, W., Liu, A.K., Mendola, J.D., neurons in cortex areas 17 and 18 of the cats. Science China 44, 136–145. Sereno, M.I., Dale, A.M., 1998a. Functional analysis of primary visual Zucker, S.W., Cavanagh, P., 1985. Subjective figures and texture perception. cortex (V1) in humans. Proc. Natl Acad. Sci. USA 95, 811–817. Spatial Vis. 1, 131–139.