Visual spatial integration. Each location in the retinal image is being analyzed by a large number of local detectors that process different aspects of the image. After such an analysis, in order to arrive at a unified percept of any visually perceived object or event, the activity of local analyzers responding to the same object has to be integrated. I apply visual psychophysics to reveal visual integration mechanisms. I use a contour-integration task to study the integration of orientation information across the visual field (e.g., Kovács & Julesz, 1993, 1994; Kovács, 1996; Kovács et al, 1999, 2000). The task involves the detection of spatially extended patterns with continuous paths of Gabor signals and orientation noise. Linking of local Gabor elements in this task requires both local orientation analysis, and lateral interactions among the local analyzers. These relatively low-level interactions are sensitive to factors of global perceptual organization. We found superiority of closed paths over open paths in terms of maximal separation between adjacent elements in this task (Kovács & Julesz, 1993) and enhanced local contrast sensitivity within closed contours (Kovács & Julesz, Nature, 1994). The superiority of closure – a classic Gestalt concept – under these circumstances is striking because the global constraint of closure affects local association rules: adjacent, nearly collinear segments can be linked across larger spatial distances when they belong to a closed contour.

      Contour                            Contour in noise            Open contour in noise      Closed contour in noise

Interocular interactions in binocular rivalry. Global context has a significant impact not only on spatial but on interocular interactions. We introduced the ‘patchwork’ rivalry paradigm as a critical test for investigating the role of eye competition versus pattern competition in binocular rivalry (Kovács et al, 1996). This work shows that binocular rivalry – which was considered for over a hundred years to be a result of an autonomous, low-level process – might be driven by higher order organizing processes. Beyond the early competitive mechanisms between the eyes in binocular rivalry, the visual system seems to rely on cooperative processes in an attempt to arrive at a consistent percept. This is relevant with respect to the neural locus of perceptual awareness, because it can help to identify the neural units that are correlated with subjective experience.

Development and plasticity of spatial integration. In a series of recent experiments, we tested the development of visual spatial integration in children, and the effect of abnormal visual input in adults. In addition to significant improvement of performance in children between 5 to 14 years in the contour integration task (Kovács et al, 1999, Kovács, 2000), we found a deficit in the performance of adults who had abnormal binocular visual experience during their developmental period (Kovács et al, 2000, Pennefather et al, 1999). We have also started to test very young subjects (3-month-old babies, Gerhardstein et al, 1999, 2000), and patient populations with a possibility of perceptual organization problems (schizophrenics, Silverstein et al, 2000; agnosic patients, Giersch et al, 2000). In all these studies we used a card-test version of the contour integration task that were designed for the purposes of testing children and clinical patient populations.

Normal development. Over four hundred children with normal vision were tested with the contour-integration cards (Kovács et al, 1999; Kozma et al, 1997, 1999).  The children ranged in age from 5 to 14 years. Children in the 13-14 year old group were able to detect most of the contours in the set, while 5-6 year old  children missed the contours in about half of the cards. This is a great difference in contour-integration performance between the two age groups. There is gradually increasing performance, and adult-like levels are not reached until after adolescence in this task. In order to see whether the surprisingly slow development that we found is due to the immaturity of perceptual skills, or to the lack of motivation or attention, we looked at the effect of practice and at the specificity of learning (Kovács et al, 1999). We found significant learning after a few days of practice both in children and in adults, and the learning was specific to visual cues of orientation and color. A high degree of stimulus specificity usually suggests that the plastic neuronal changes of learning took place at early cortical levels where the basic stimulus dimensions are still separable. The lack of transfer of learning across orientation and color in the contour-detection task indicates that the involved mechanisms are perceptual mechanisms with access to relatively low-level perceptual representations. Why do young children perform poorly in the contour integration task? The development of horizontal connections in layer 2/3 of the primary visual cortex of humans has been shown to extend well into childhood, and a delayed postnatal development of feedback connections between V1 and V2 has also been indicated in humans. In order to see whether absolute cortical distances spanned by lateral interactions are limiting performance in the contour-integration task, we conducted an experiment where we varied the spacing among contour elements while we kept the relative noise level constant. We found that adult performance is independent of contour spacing, while we found better performance at smaller contour spacings in children, and the difference between children and adults in contour detection performance was greater at larger contour spacings. We concluded that long-range interactions span a shorter spatial range in children than in adults.

Abnormal development. We have shown that abnormal binocular input during an early developmental period of binocular vision not only leads to reduced visual acuity (usually in one eye only: the condition of amblyopia), but to reduced contour-integration performance as well (Kovács et al, 1996, Kovács et al, 2000; Pennefather et al, 1999), and that performance in the contour-integration task depends on the type of the treatment patients received (Chandna et al, 1998).  Our results also show that even in the absence of amblyopia, a history of abnormal binocular input can lead to reduced contour-integration performance in both eyes (Kovács et al, 2000).  While strabismic amblyopes are at the 6-7 year old performance level in the contour-integration task as measured by the card test, strabismic patients without amblyopia perform at the 10-year-old level (Kovács et al, 2000). The conclusion of these studies was that there is a complex picture in terms of visual deficits following abnormal binocular input, and the type and severity of the deficit depends not only on the type and timing of input abnormality, but on the treatment history of the patients as well. We are in the process of preparing a longitudinal study to look at the perceptual organization abilities of children with early and late onset eye problems (e.g., strabismus or anisometropia).

Functional and developmental dissociation of higher visual function. (Kovács, 2000) The conclusion that spatial integration carried out in the occipital lobe has a very slow human development raises further questions: What is the consequence of the immaturity of this specific integrative function with respect to other perceptual and cognitive functions? What is the developmental pattern of visual functions related to higher cortical areas? A possible functional dissociation of higher visual function is in terms of parallel pathways for “action” and “perception”. The action system is responsible for the control of visually guided action (mediated by the dorsal visual stream), while the perceptual system is responsible for perceptual and cognitive representations of objects and events (mediated by the ventral visual stream). I propose that the two systems are developmentally dissociated as well, and the perceptual system mediated by the ventral visual stream is delayed with respect to the action system mediated by the dorsal stream. The hypothesis is based on the following considerations: (a) there is anatomical indication for the slower functional development of the occipitotemporal pathway compared to the occipitoparietal pathway in the macaque; (b) there is human anatomical indication for slower establishment of long-range connectivity in the ventral compared to the dorsal stream based on the analysis of a human anatomical database (Kovács et al, 1999); (c) protracted developmental courses of specific temporal lobe functions indicate immature connectivity in the ventral stream (Káldy and Kovács, 2000; Gandhi and Kovács, in preparation); (d) the dorsal stream is phylogenetically older, which might suggest more genetic preprogramming, and faster maturation; (e) during ontogeny, there might be a greater need for early availability of structures mediating the visual control of action, which is the prerequisite of exploration (dorsal stream) than for those representing object identity (ventral stream); (f) if the ventral stream is mediating perceptual representations that are part of a high-level cognitive network enabling us to understand the environment, a protracted developmental course with plasticity preserved beyond the earliest ages would be desirable. Some of these considerations are hypothetical, and require further investigation.

Mapping the exact pattern of human visual development will not only contribute to the general knowledge in the field, it will also be particularly relevant in some aspects of primary school education, such as diagnosing children with slowly developing visual systems.

Recent Publications: (to be updated!!)
1. A. Giersch, G. Humphreys, M. Boucart, I. Kovács: The computation of occluded contours in visual agnosia: evidence for early computation prior to shape binding and figure-ground coding. Cognitive Neuropsychology, in press.

2. S. Silverstein, I. Kovács, R. Corry & C. Valone: Perceptual organization, the disorganization syndrome, and context processing in chronic schrizophrenia. Schizophrenia Research,  43(1), 11-20, 2000. full text in pdf | PubMed | Science Direct

3. I. Kovács, U. Polat, A. M. Norcia, P. M. Pennefather and A. Chandna: A new test of contour integration deficits in patients with a history of disrupted binocular experience during visual development. Vision Research, 40(13), 1775-1783, 2000. full text in pdf | PubMed | Science Direct

4. I. Kovács: Human development of perceptual organization. Vision Res. Special Issue on Attention, 40(10-12), 1301-1310, 2000. full text in pdf | PubMed | Science Direct

5. I.Kovacs, P. Kozma, A. Feher and G. Benedek: Late maturation of visual spatial integration in humans. Proc.Natl.Acad.Sci. USA, 96(21):12204-12209, 1999. full text in pdf | PubMed |  PNAS online

6. P. M. Pennefather, A. Chandna, I. Kovacs. U. Polat and A. M. Norcia: Contour detection threshold: repeatability and learning with "contour cards." Spatial Vision,, 2(3):257-266, 1999. PubMed

7. I. Kovács, Á. Fehér and B. Julesz: Medial-point description of shape: a representation for action coding and its psychophysical correlates.  Vision Res. Special Issue on Recognition, 38, 2323-2333, 1998. full text in pdf | PubMed | Science Direct

8. I. Kovács and Á. Fehér. Non-Fourier information in bandpass noise patterns. Vision Res. 37(9), 1167-1175, 1997. full text in pdf | PubMed | Science Direct

9. I. Kovács. Gestalten of today: Early processing of visual contours and surfaces. Behav. Brain Res. Invited review, 82(1):1-11, 1996. full text in pdf | PubMed | Science Direct

10. I. Kovács, T. V. Papathomas, and M. Yang, Á. Fehér: When the brain changes its mind: Interocular grouping during binocular rivalry. Proc.Natl.Acad.Sci. USA, 93: 15508-15511, 1996. full text in pdf | PubMed | PNAS online