Jonathan Winawer - US grants

[unreadable] DESCRIPTION (provided by applicant): The broad goals of this proposal are to increase understanding of how surface properties are represented by the visual system. In particular, the project seeks to deepen our understanding of perceptual constancies, the means by which objects and surfaces maintain their appearance in changing contexts. This kind of research falls under the NEI's stated mission of understanding visual function. The first aim of the proposal is to study the visual cues that are important in understanding how scenes affect the perception of surface lightness. In particular, the aim will use psychophysical methods to study the image features that the visual system uses to discount the effects of shadows, illumination changes, and transparency in recovering surface lightness. The second aim will build on the findings of the first; it will seek to identify regions in human visual cortex that respond as the interpretation of shadows and transparencies vary, influencing perceived surface lightness. The third aim will examine the effects of scene interpretation on the representation of the size of surfaces and objects. In this aim, we will study how perceived depth influences the perceived size of objects, and how size constancy mechanisms are represented in visual cortex. PUBLIC HEALTH RELEVANCE Basic mechanisms of visual processing are impaired in a wide range of disabilities including psychiatric conditions such as schizophrenia and autism. One impairment found in a number of conditions is an abnormal effect of context on visual processing. Understanding how the normal visual system processes information and takes into account contextual information may guide development of tests for early diagnosis of such conditions. [unreadable] [unreadable] [unreadable]

7. Project Summary Seeing requires pooling information over space and time. The nervous system must appropriately integrate and segregate inputs in order to correctly interpret visual scenes. The proposal seeks to build experimental and computational infrastructure to characterize how the human visual system pools information from simple space-time stimuli using multiple imaging modalities, and to relate these measurements to visual perception. A key component of the proposal is to combine measurements from functional MRI with scalp and intracranial electrodes. The different instruments for measuring human brain function have very different sensitivities. Spatial summation measurements and temporal summation measurements will be conducted to derive models of cortical space-time receptive fields. We will test hypotheses about how these receptive fields are organized throughout the visual pathways, characterizing differences between central and peripheral cortical representations and between early and late visual areas. The organization of spatial summation across the visual pathways is better understood than the organization of temporal summation. Spatial summation will therefore be emphasized in model development and validation; temporal summation will be emphasized in later stages. In the mentored phase of the award, one set of studies will be conducted to establish the basic computational infrastructure. Methods will be developed to use measurements in one modality to predict data in another modality. This work will involve modeling spatial summation across the many visual field maps using fMRI, and then using these models to predict the pattern of activity across electrode arrays in response to simple visual stimuli. For a second set of studies, these models will be used as constraints to resolve temporal responses with EEG at the spatial resolution of fMRI. Findings from both sets of studies will be validated with measurements of intracranial electrodes in a clinical subject population. Computational models will be developed to integrate visualization and analysis across the modalities. In the independent phase of the award, temporal summation will be studied with fMRI and the results compared against those from EEG. Measurements of temporal summation will characterize the temporal integration period of different brain areas. Concurrent EEG and psychophysical experiments will be conducted to identify the neural activity patterns associated with perception of various stimulus classes. Together the pattern of results will provide an account of how temporal processing across the visual hierarchy shapes perception. The mentored phase will take place at Stanford University, mostly in the Center for Neurobiological Imaging. The studies will build on the candidate's expertise in functional MRI experiments, computational modeling, and psychophysics. Training in conducting and analyzing EEG experiments and relating the results to MRI data will be a significant part of the training in the mentored phase. The candidate will seek an independent faculty position during the second year of the mentored phase.

Project Summary / Abstract A major obstacle in the study of human brain function is that we currently have limited understanding of how the measurements made by different instruments, such as fMRI and EEG, relate to one another and to the underlying neuronal circuitry. Significant efforts have led to development of models within various specialist fields, but fragmentation has held us back from advancing our interpretation of the spatiotemporal characteristics of non-invasive imaging signals. Bringing together the various models that pertain to signal interpretation would constitute a significant advance in what we can learn from non-invasive neuroimaging. In this project we take such an integrative approach to the study of cortical sensory systems. We intend to develop a set of connecting, empirically driven models that will predict how sensory stimuli are encoded in neuronal population activity underlying electrophysiological measures (AIM 1), and hemodynamic measures (AIM 2), leading to a comprehensive integrative model (AIM 3). The pivotal integrative model (AIM 3) will improve our understanding of, and revolutionize the information we can obtain from fMRI, the modality with the highest potential for mapping detailed functions non-invasively in humans. To achieve this we will combine hemodynamic and electrophysiological measurements at multiple spatial scales in humans, and in rodents at very high resolutions. This will include non-invasive (fMRI at 3T and 7T, MEG and EEG) and invasive (optical imaging, ECoG) modalities obtained from healthy humans. By obtaining multiple modality recordings from the same individuals, using the same stimuli and tasks, we will be able to unequivocally link clear and specific electrophysiological information to widely used fMRI technology, while significantly improving our understanding of the electrical and hemodynamic phenomena underlying brain activity. The research constitutes a multicenter endeavor to A) develop a comprehensive model to link external inputs to neuronal population physiology to non-invasive imaging measures, B) obtain state of the art multimodal recordings from the same individuals in order to bridge modalities and inform the models, C) validate the models with data from multiple modalities (ECoG, fMRI, MEG/EEG, optical recordings) and brain systems (visual, somatosensory and motor), and D) make algorithms and data available to the neuroscience community to foster further development beyond the project's lifetime. Moreover, the research will foster reconciliation of different theories about the relation between electrophysiology and fMRI and will lead to `breakthroughs in understanding the dynamic activity of the human brain'. Such breakthroughs will be essential in improving disease models of the nervous system, which rely on inferences about neuronal population activity from non-invasive imaging of human brain activity.

Project Summary The long-term objective is to understand how visual performance varies across the visual field during childhood, adolescence and adulthood, and the neural correlates and possible oculomotor correlates underlying such changes. Vision at the center of gaze (fovea) has high sensitivity and resolution, facilitating good performance in many tasks, but performance decreases with increasing distance from fovea. Simple radial distance from fovea ?eccentricity? is not the only relevant retinotopic parameter for perception. At any given eccentricity stimuli can fall anywhere along circle, and the polar angle of a stimulus at ?isoeccentric? locations also has pronounced effects on perception in human adults, captured by a Performance Fields (PF) map. And yet the role of the polar angle has not been studied in children and adolescents, and the underlying neural representations and possible oculomotor correlates are practically unknown. PF present an opportunity for establishing tight quantitative links between behavior and neural representations of visual space. The proposed studies will measure PF in children, adolescents and adults, under different attentional states (Aim 1), along with fMRI visual field maps (Aim 2) and fixational oculomotor behavior (Aim 3). By highlighting and characterizing the functional, oculomotor and neural correlates of PF and their modulations across multiple timescales, the proposed studies will elucidate critical constraints on perception and performance as a function of polar angle. We will implement a computational observer model that is the glue that binds the three aims together. In addition to advancing our knowledge of visual perception, the proposed research will enable us to make predictions about human performance. The characterization of PF has significant implications for ergonomic and human factors applications, which should take into account functional plasticity during development, compensation via selective attention, and oculomotor behavior. For example, it is of critical importance for user-interfaces that present information at different locations of the visual field. With an understanding of how attention affects PF we can extend our knowledge to real-world displays, such as navigation and cockpit alerting systems, control panel layouts in cars, computer-aided detection systems, and software for presenting radiological images. Furthermore, the gained knowledge can aid the design of artificial image recognition systems. This research will inform the development of compensatory strategies and interventions to deal more effectively with the limitations of the visual system. In addition, understanding the functional causes of performance differences across the visual field and how attention and/or oculomotor behavior interact with them will improve our models of visual dysfunction (e.g., macular degeneration, retinitis pigmentosa), as well as the diagnosis of these disorders.

Data; gray matter; Human; Visual Cortex; white matter;