E¡ect of connectivity and bistability on the visual potentials evoked by illusory ¢gures Mathieu Brodeur a,b , Franco Lepore b , Caroline Veilleux b , Yasmine Alyanak b and J. Bruno Debruille a a Douglas Hospital Research Centre, McGill University and b Centre de Recherche en Neuropsychologie et Cognition (CERNEC), Universite¤ de Montre¤ al, Montre¤ al, Quebec, Canada Correspondence and requests for reprints to Dr J. Bruno Debruille, PhD, Douglas Hospital Research Centre, McGill University, 6875 Boul. LaSalle, Verdun, Quebec, Canada H4H 1R3 Tel: + 1 514 761 6131 ext. 3405; fax: + 1 514 888 4099; e-mail: bruno.debruille@douglas.mcgill.ca Sponsorship:The study was supported by Grant 96997-10 from the Fonds de la Recherche en Sante¤ du Que¤ bec and the Re¤ seau de la Sante¤ Mentale du Que¤ bec allocated to the last author. Received 14 November 2005; accepted 16 November 2005 The present study aimed at testing functional hypotheses regard- ing two brain potentials elicited by illusory ¢gures. Accordingly, the N1 potential indexes mechanisms connecting the separate parts of the illusory form, whereas a subsequent negative poten- tial indexes compensatory processes triggered by perceptual di/- culty. Here, perceptual di/culty was induced by bistability; that is, by equating the probability of perceiving the illusory form to that of perceiving the independent separate parts. We compared the brain potentials evoked by a strongly connected illusory square, with almost no bistability, with those evoked by a weakly con- nected illusory square presenting strong bistability. Consistent with our hypotheses, the latter ¢gure evoked the smallest N1 and a larger negative component peaking at 360 ms (N360). These results strengthen the link between N1 and connection and bet- ween negativity to perceptual di/culty and perceptual di/culty. NeuroReport 17:157^161  c 2006 Lippincott Williams & Wilkins. Keywords: bistability, contour assignment, electroencephalography, event-related potential, illusory contours, Kanizsa ¢gure, perceptual saliency, visual evoked potential Introduction For more than 30 years, Kanizsa-like figures (e.g. Fig. 1a) have been used for investigating how illusory forms are perceived. These figures are generally composed of three or more black pacmen. The ‘mouths’, or notches, of these pacmen can be perceived as the corners of an illusory form. In this case, the pacmen become the inducers of illusory contours that appear between the corners and circumscribe the form. This figure is sometimes termed the modal figure. Among the techniques used to investigate the processes underlying the perception of illusory forms are visual evoked potentials (VEPs). Computed by averaging the electroencephalography (EEG) epochs following the pre- sentation of visual stimuli, VEPs are made of a succession of negative and positive deflections. The N1 component of the VEPs, which peaks around 170 ms, is the first large negative deflection. Its amplitude has consistently been found to be larger for the Kanizsa figure than for control figures [1–5]. In a recent experiment [6], we attempted to specify the processes underlying this difference by including an amodal version of the Kanizsa figure. Like the modal version, this amodal figure also includes pacmen, the notches of which can be perceived as the corners of a form. In contrast to the modal Kanizsa figure, however, no illusory contours appear because the notches were enclosed in their inducers by circling each pacman with a thin black line. The posterior N1 to the Kanizsa figure was found to be larger than the N1 evoked by the amodal figure. This effect was proposed to be associated to a weaker connection of the corners composing the illusory form in the amodal figure than in the Kanizsa figure. Connection in the amodal figure is much less likely because of the enclosure of the notches in the inducers. Our data also showed that the VEPs to the amodal figure were more negative than the VEPs to the Kanizsa figure between 240 and 290 ms following onset. This effect was part of a broader effect named the Npd effect (Npd standing for ‘negativity to perceptual difficulty’) that was proposed to reflect compensatory processes. These processes were viewed as mechanisms that attempt to strengthen the connection of the corners through further processing or by lowering the impact of information that impedes connec- tion, such as the thin black lines that enclose the notches in the inducers. To test the above interpretations of N1 and Npd effects obtained with illusory figures, other conditions that weaken notches’ connection and increase perceptual difficulty were needed. One such condition may be bistability. A figure is said to be bistable when it can be perceived in two different ways that are mutually exclusive. As a matter of fact, the Kanizsa square itself is bistable as it can be perceived either as four independent pacmen or as a square placed over four disks. Each of these perceptions prevents the occurrence of the other because they assign the notches differently [7,8]. When the Kanizsa figure is perceived as independent pacmen, notches are assigned to the contours of the inducers. When the Kanizsa figure is perceived as a square VISION, CENTRAL NEUROREPORT 0959-4965  c Lippincott Williams & Wilkins Vol 17 No 2 6 February 2006 157 Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.