Nava Rubin - US grants

IBN-9720305 PI: SHAPLEY This project is being funded through the Learning & Intelligent Systems Initiative. The questions involve vision and learning in the central nervous system. The project investigates how the brain represents and processes perceptual information, and the adaptive changes that occur when the system learns and improves its performance. Here the visual system is used as a gateway into the workings of the brain, employing methods from psychophysics, neurophysiology, mathematics and engineering. The work focuses on visual grouping, which is the ability of the visual system to link together local elements of the visual image into coherent wholes. Grouping is one of the most fundamental aspects of human vision, and the goal of this work is to obtain a theory of visual grouping that will explain the physiological and psychophysical data, and lead to new technological ideas to be applied in intelligent artificial systems. Results will be important because understanding the computations in the brain that produce grouping would be a leap forward in our understanding of brain function, and of any systems that can adapt to experience. This work will therefore have an impact in designing learning materials and in the optimal methods of presenting information, and in the design of the next generation of computer vision systems and intelligent control systems. This project is supported in part by the NSF Office of Multidisciplinary Activities in the Directorate for Mathematical & Physical Sciences.

We propose a novel approach to study the mechanisms underlying integration and segmentation in motion perception. This problem is central to motion processing, and it can also be used as a model for how the brain computes a global percept from many isolated local cues in other domains of vision. There has been previous research on this topic with stimuli such as plaids and random-dot displays, using brief-duration presentations. We will approach the problem in a novel way by focusing on the dynamics of perceptual alternations during long presentations of ambiguous motion displays. Such stimuli can evoke the perception of either motion integration long ("coherency") or segmentation ("transparency"). At long presentations, perception alternates between coherency and transparency; it is bi-stable. Our Preliminary Results show that the dynamics approach provides more sensitive measures of the relative strength of segmentation versus integration than brief-presentation methods. Importantly, it allows to measure the strength of the two processes independently. We have already revealed a wealth of new information about the mechanisms of coherency and transparency and the interplay between them. Our approach offers more than just methodological advantages. It will provide an invaluable set of data for testing and further developing models for how the brain resolves the competition between motion coherency and segmentation. Furthermore, expanding the dynamics approach - at present restricted mostly to binocular rivalry - to another domain in vision will be an important step towards understanding general principles underlying competition and cooperation between different brain states. In the work proposed here, we have five specific aims: (I) to apply a new dynamic method we have developed, RTtransp, which provides a better measure of the relative strength of integration and segmentation, to test the effect of motion parameters over much wider ranges than had been possible so far. (II) to validate the dynamic method of Durations of Alternating Percepts (DAP) as a way of measuring the strength of two competing percepts independently; DAP has been extensively used in binocular rivalry but its validity in the motion domains needs further testing. (III) to use DAP to check how motion parameters that are known to affect the relative strength of motion segmentation and integration affect each of those processes independently. (IV) to use DAP to study the effect of static form cues for surface segmentation on the interplay between the segmentation and integration mechanisms in motion. (V) to use fMRI to relate our results to the physiological processes underlying motion integration/segmentation

This is a proposal to renew funding for research using a novel approach - measuring the dynamics of perceptual bi-stability in ambiguous motion displays - to study the mechanisms underlying motion integration and segmentation. This problem is central to motion processing, and it can also be used as a model for how the brain computes a global percept from many isolated local cues in other domains of vision. There has been previous research on this topic using brief-presentation methods. Our approach focuses on the dynamics of perceptual alternations during long presentations of ambiguous motion displays. In the first phase of the project we showed that dynamics-based measures are sensitive to stimulus manipulations in wide parametric ranges, including where brief-presentation methods suffer from "ceiling" and "floor" effects. With this methodological advantage we have uncovered a wealth of new and unsuspected findings about plaid perception. We have also adapted the dynamics methods to the fMRI environment, and identified two localized cortical sub-regions, labeled KOint and LOseg, whose activation is strongly correlated with the perceptual dominance durations of the alternating percepts. Activity in these two regions is in anti-phase, crossing-over with perceptual switches. In the proposed continuation phase of the project we will (i) provide further support for the hypothesis that the perception of the ambiguous plaid stimulus is determined by the outcome of competition between the two occipital regions we have identified, KOint (associated with motion integration) and LOseg (motion segmentation), (ii) test that the behavior exhibited by LOseg and KOint is not specific just for the plaid, but rather generalizes also to other ambiguous stimuli that involve competition between motion integration and segmentation, (iii) test a complementary prediction that is also implied by our hypothesis that LOseg and KOint mediate competition between motion segmentation and integration, namely that these two regions should not exhibit anti-phase modulations when observers undergo perceptual alternations caused by ambiguous motion displays that do not involve integration/segmentation competition, (iv) further elucidate the role of LOseg and KOint in motion-based segmentation and integration compared with form-based segmentation and integration, (v) generalize our results to more complex stimuli, specifically stimuli that require different global motion processing in different portions of the visual field.