Forwarded message: From: <PLEH(a)IZABELL.elte.hu> To: everyone(a)izabell.elte.hu, dradam(a)ludens.elte.hu, BANYAI(a)IZABELL.elte.hu, h1872csa(a)ella.hu, KURGYIS(a)IZABELL.elte.hu Date: Tue, 8 Nov 1994 11:26:56 GMT+100 Subject: Kovacs Ilona januari vision oraja Priority: normal KEDVES KOLLEGAK, Kovacs Ilona, az ELTE Altalanos Pszicholgiai Tanszek es a Rutgers Egyetem munkatarsa 1995 januar 9 es 24 kozott intenziv kurzust tart angol nyelven a vizuaklis kutatasokrol. A Pszicholopgiaqi Doktori iskola kognitiv alreszenek regisztralt kurzusa ez 1995 tavaszi felevben. Szivesen latnank azonban mind mas doktopri iskolasokat (Adam es Csanyi professzoroktol peldalul), mind egyeb erdeklodoket. A pontos orabeosztast kesobb kozoljuk, valszinuleg minden munkanapon lesz ora. Szeretnenk azonban, ha az erdeklodok elojelentkeznenek pleh(a)izabell.elte.hu es kurgyis(a)izabell.elte.hu cimekene. Ennek oka a kovetkezo: az alabbi tematika es irodalmojegzeknek megfelello szovegyujtemenyt keszitunk meg decemberben. Ehhez kellene becsulnunk a letszamot. Kreditert jaro diakok szamara az olvaskonyv a SOROS Alapitvany HESP alapja tamogatasaval ingyenes. Udvozlettel s erdeklodeseteket elore is koszonve: Pleh Csaba ------------------------------------- The origins of "modularity" in visual science January, 1995 (seminar on visual perception held by Ilona KovVcs at Eptvps Lorand University, Budapest) The idea that the brain can be partitioned into more or less independent modules is one of the basic assumptions of modern neuroscience. Modularity can be viewed as a useful technical approach, and when it is not taken to the extremes, it helps scientist to define problems clearly, and ask questions that can be tested experimentally. Modularity is emphasized usually when the system under investigation seems to be very complex, and without the isolation of its parts, understanding is difficult. Once the existence of the parts is clarified, the interaction between them is the next question, and the system as a whole is coming back to the picture. From this point of view, modularity is not a philosophical issue, it is a natural way of the development of science. VISUAL PSYCHOPHYSICS AND THE ORIGINS OF "MODULARITY" My goal in selecting the papers for this textbook was to present an example of the birth of modularity within a specific field. In vision, a single demonstration drew attention to the fact that there are hidden capabilities in the brain to process raw perceptual information quite independently from high level cognitive operations. Julesz's random-dot stereograms proved that binocular disparity information in itself can be enough for us to extract the three dimensional structure of the environment (see Julesz, 1964, 1965). The demonstration defined the first clearly separable module of vision. As Julesz believed that there must be other modular aspects of vision, and these should be treated separately from the cognitive levels, he introduced the notion of "early vision". Early vision in the strict sense was believed to be mediated by autonomous, fast, parallel processes of the brain that are not penetrable for the higher cognitive levels. The approach was so fruitful that it opened a literally new field of research. As the questions about binocular vision could be now tested experimentally, not only psychologists, but physiologists too were able to join the new trend. The modular approach also opens the field for modelers both in the sense of constructing "vision machines" and in the sense of building a theory of vision. Julesz was one the firsts to offer a computational model of stereoscopic vision, and later of texture perception. As the properties of the modules started to become clear, of course the questions about their cooperation had to come up. That happened some 20 years after the discovery of the random-dot stereograms (it took many years, because it turned out that those "simple" input systems are, in fact, very complicated). Julesz was among the firsts to ask questions about the role of attention in visual processing Bergen and Julesz, 1983; Sagi and Julesz, 1985). The function of visual attention in this context is to serve as a glue that binds together the results of separated modules. And as coordination by attention and neural interactions are gaining more and more emphasize, different puzzles open up for the researchers and new concepts of modularity seem to be required (KovVcs and Julesz, 1994). COMPUTATIONAL VISION How did random-dot stereograms give rise to the field of computational vision? This question is answered in Marr's introductory chapter (Marr, 1982). What are the tacit assumptions underlying computational vision and what are the challenges that have been raised against them (see Ullman, 1991)? What can computational vision accomplish so far? An example, related to the modeling of biological vision is shown in Bergen and Landy's (1991) paper. NEUROPHYSIOLOGY As vision seems to be one if those faculties of mind that we share in a large extent with primates, it gives an exceptional opportunity for experimenters: human subjects can be asked verbally about their experiences, and monkeys can take part in physiological experiments, where the black box is being open. The results of these two different paradigms can be combined because of the similarity between the visual systems of humans and primates. We will see how did Poggio find the physiological correlates of random-dot stereograms (Poggio, 1991). An interesting extreme example of modularity is represented in Newsome et al's (1991) paper, where they report that stimulation of a very small number of neurons by the means of microelectrodes results in altered perceptual decisions, implying that a small bunch of neurons is responsible for the given behavior. I also included some recent findings about the architecture and interactions within the primary visual cortex (Gilbert, 1993) that may serve to explain the psychopysically found interactions (KovVcs and Julesz, 1994). The last two papers (Niebur and Koch, 1994, Fregnac, 1991) are dealing with the problems of integration and visual attention on a larger scale. The reader might be familiar with some of the context represented in this textbook, it might still be a hard task to thoroughly follow every articles since they come from diverse fields within visual science. To help in understanding the basic terminology and conceptual framework, I included a few chapters from a recent introductory Perception book (Sekuler and Blake, 1994). Contents INTRODUCTION TO VISUAL PERCEPTION R. Sekuler, R. Blake: Perception (chap 3-5) MacGraw-Hill, Inc., 1994 VISUAL PSYCHOPHYSICS AND THE ORIGINS OF "MODULARITY" B. Julesz: Binocular depth perception without familiarity cues. Science, Vol 145, 356-362, 1964 B. Julesz: Texture and visual perception. Scientific American, Vol 212, 38-48, 1965 J.R. Bergen and B. Julesz: Parallel versus serial processing in rapid pattern discrimination. Nature, Vol.303, 696-698, 1983. D. Sagi and B. Julesz: "Where" and "what" in vision. Science, Vol.228, 1217-1219, 1985. I. KovVcs and B. Julesz: Perceptual sensitivity maps within globally defined visual shapes. Nature, Vol 370, 644-666, 1994. COMPUTATIONAL VISION D. Marr: The philosophy and the approach. in: D. Marr: Vision. W.H.Freeman and Company, New York, 1982 S. Ullman: Tacit assumptions in the computational study of vision. in: A. Gorea (ed.): Representations of Vision, Cambridge University Press, Cambridge, 1991 J. Bergen and M.S. Landy: Computational modeling of visual texture segregation. in: M.S.Landy and J.A. Movshon (eds) : Computational models of visual processing,a Bradford Book, MIT Press, 1991. NEUROPHYSIOLOGY G. F. Poggio: Physiological basis of stereoscopic vision. in: D. Regan (ed.): Binocular vision, Macmillian Press, 1991 C.D. Gilbert: Circuitry, architecture, and functional dynamics of visual cortex. Cerebral Cortex, 3, 373-386, 1993. W. Newsome et al.: Manipulating perceptual decisions by microstimulation of extrastriate visual cortex. in: A. Gorea (ed.): Representations of Vision, Cambridge University Press, Cambridge, 1991 E. Niebur, C. Koch: A model for the neuronal implementation of selective visual attention based on temporal correlation among neurons. J.Comp.Neurosci, 1, 141-158, 1994 Y. Fregnac: How many cycles make an oscillation? in: A. Gorea (ed.): Representations of Vision, Cambridge University Press, Cambridge, 1991 --------------------------------