Cheryl A. Olman, Ph.D. - US grants

Our long-term goal is to understand how humans perform natural tasks given realistic visual input. Object perception is critical for the everyday tasks of recognition, planning, and motor actions. Through vision, we infer intrinsic properties of objects, including their shapes, sizes, materials, as well as their identities. We also infer their depths and movement relationships to each other and ourselves, as well as determine how to use this information. The remarkable fact is that the human visual system provides a high level of functionality despite complex and objectively ambiguous retinal input. Current machine vision systems do not come close to normal human visual competence. In contrast, our daily visual judgments are unambiguous, and our actions are reliable. How is this accomplished? Our conceptual approach to this question is motivated by our previous work on object perception as Bayesian statistical inference, and its implications for how human perception gathers and integrates information about scenes and objects to reduce uncertainty, resolve ambiguity and achieve action goals. Our experimental approach to this question grows out of our team's past accomplishments in using behavioral techniques such as interocular suppression, high-field functional magnetic resonance imaging and analysis, and Bayesian observer analysis of human behavioral performance. We combine our conceptual and experimental approaches to address a new set of questions. In three series of experiments, we aim to better understand: 1) the relationship between cortical activity and the perceptual organization of image features into unambiguous object properties and structures (Within-object interactions);2) how visual information about other objects and surfaces reduces uncertainty about the representation of an object's properties and depth relations (Between-object interactions);and 3) whether and how information and uncertainty may be processed differently depending on the viewer-object interactions demanded by task, as predicted by theory (Viewer-object interactions).

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. We are using a visual stimulus paradigm that produces well-characterized localized neural inhibition in primary visual cortex to test whether, on a fine spatial scale, the positive BOLD response is more closely related to the neural output or the local neural inhibition. In the past year we have published one paper demonstrating that the V1 BOLD response correlates better with the magnitude of the local neural inhibition than with the local neural output;a second manuscript is in preparation.

DESCRIPTION (provided by applicant): Functional magnetic resonance imaging (fMRI) is undeniably the neuroimaging methodology that has become the workhorse for neuroscience and psychology researchers who want access to localized measurements of physiological changes in the brain that correlate with human behavior. The high value of fMRI measurements is based on the fact that they have been shown, time and again, to exhibit a linear correlation with the local neural population response. There are, however, a recent smattering of articles in the literature indicating a mismatch between the fMRI response and the measured or presumed neural activity. These mismatches appear limited to experiments in which only a small neural population is stimulated; they also seem most likely to occur when the balance between local neural excitation and inhibition is tipped in favor of inhibition. These reports of fMRI responses that fail to correlate with neural responses are puzzling at best, and potentially troublesome for scientists who want to draw quantitative conclusions about neural population activity from fMRI data. Our first series of proposed experiments will characterize the effects of sampling resolution on the interpretability of the fMRI response to small stimuli. Not only do flanking negative responses confound accurate interpretation of the fMRI response to small patches of neural activity, because the boundary regions are large compared to the total neural response, but size-dependent intrinsic inhibition shapes the neural response yet has an unknown representation in the hemodynamic response. The result of the first series of experiments will be a computational model characterizing (1) neuro-hemodynamic coupling at the edges of isolated patches of neural activity, and (2) the contribution of inhibitory neural activity to the fMRI response. Our second series of experiments will characterize fMRI response evoked by neural networks with different balances between excitation and inhibition. All local neural codes contain a balance between input and output; local computations use a balance of excitation and inhibition to shape the input and define output spiking rates. In this series of experiments, we will investigate the implications of our recent study showing that localized fMRI of individual image patches cannot be predicted simply from the responses of the neurons that respond best to those stimuli. Working with a computational model that demonstrates how the entire local neural population response can be used to predict fMRI responses, this second series of experiments will seek to identify signature hemodynamic response characteristics that are present when heterogeneous neural responses mask key information encoded in a sub-population of neurons. Together, these experiments will improve our ability to use high-resolution fMRI to characterize patterned neural activity, improving the utility of fMRI for clinicl applications such as neurosurgical planning and seizure locus detection. PUBLIC HEALTH RELEVANCE: Scientists who study the brain need high-resolution imaging tools in order to understand how different patterns of neural activity correlate with different aspects of behavior; functional magnetic resonance imaging (fMRI) is one of the tools that can provide the highest imaging resolution. However, we still need to answer some fundamental questions about the relationship between the fMRI signal and the underlying neural activity in the brain. The work funded by this grant will develop more accurate models for linking neural activity patterns to fMRI responses when (1) the neural response occupies only a small portion of cortex, and (2) sub- populations of neurons right next to each other have different responses. This ability to detect activity or dysregulation of activity in a subpopulation of neurons is key fr high-resolution localization of neural function, for neurosurgical planning or seizure locus detection, as well as for quantifying biomarkers of diseases such as schizophrenia, which differentially affects inhibitory neurons in visual cortex.

ABSTRACT Functional magnetic resonance imaging (fMRI) is undeniably the workhorse for neuroscientists who want localized measurements of neural correlates of human behavior. However, recent studies have presented fMRI results that conflict with traditional invasive (direct) electrophysiological measurements of neural population responses: attention strongly modulates the fMRI response in primary visual cortex (V1) but only weakly modulates neural firing rates; fMRI responses in V1 predict perceptual states while only subtle aspects of direct recordings of V1 neural responses correlate with perception. An important step toward reconciling the differences between fMRI data and invasive electrophysiology is to obtain fMRI data that can map out the details of the local neural population code on a scale that is closer to that sampled by electrodes. The proposed experiments will use 7 Tesla fMRI with sub-millimeter resolution ? a spatial scale comparable to that of the local field potential in electrophysiology ? to determine how the spatial details of the neural population response in V1 depend on visual stimulus properties as well as perceptual state. Because input, output and local connections are segregated according to depth, the primary focus of this research is to dissociate signals at different cortical depths. Several recent literature reports show that the fMRI response amplitude is different at different cortical depths. However, it is not yet established whether depth-dependent fMRI actually reflects local neural network changes at different cortical depths. The following aims seek to verify that fMRI has differential laminar sensitivity that corresponds meaningfully to neural activity. The first aim is to validate layer-specific fMRI against known properties of the intrinsic neural network. While fMRI and electrophysiological measurements of V1 response modulation due to behavioral or perceptual processes (putative feedback processes) disagree, there has been ample demonstration of the agreement between fMRI and electrophysiological measurements of intrinsic neural responses such as contrast sensitivity or orientation selectivity. Two sets of experiments will therefore quantify the depth-dependent fMRI response to simple visual stimuli, validating the fMRI laminar profile against laminar profiles predicted by electrophysiology. The second aim is to study modulation of neural responses in V1 by visual information that is extracted over a larger spatial scale. Enhancement of V1 neural responses by global scene structure (figure/ground segmentation) and suppression of V1 neural responses by uniform texture (orientation-dependent surround suppression) differ importantly in the role of awareness in regulating the modulation in V1. We will quantify fMRI laminar profiles under these two different kinds of contextual modulation. Because both visual feature grouping and iso-orientation suppression are affected by neurological disorders such as autism and schizophrenia, validating a technique for monitoring these mechanisms of visual contextual modulation has utility in the clinical setting.

PROJECT SUMMARY Our knowledge of signal processing in various parts of the human brain has been heavily influenced by non- invasive functional magnetic resonance imaging (fMRI) experiments. FMRI infers the location and selectivity of neural activity from vascular signals. However, brain circuits are much more complex than regional differences in neuronal selectivity. Specifically, the largest part of the brain (neocortex) accounts for up to 80% of the brain volume and is divided into six distinct layers. Specific computations, e.g., local processing vs. feedforward inputs vs. vs. feedback inputs, are done in specific cortical laminae. Thus, if high-resolution layer-specific fMRI is shown to reflect the repertoire of neural computations performed across these cortical layers, it would be an invaluable refinement to non-invasive imaging. However, despite the widespread usage of low-resolution fMRI, a detailed understanding of how neural activity generates vascular responses remains unknown. The goal of this project is to elucidate the link between neural and vascular signals across laminae by combining two-photon imaging of neural and vascular responses with ultra-high-field (UHF) fMRI. Experiments will use sensory visual stimuli that induce layer-specific responses. In cat primary visual cortex (V1), which has a functional architecture (e.g., maps for stimulus orientation) similar to human V1, we will measure neural activity (synaptic and spiking) with single-cell resolution together with vascular signals (blood flow, blood volume, and oxygenation) in individual vessels across the entire cortical thickness. We will also perform UHF lamina-specific fMRI in cat (9.4 and16.4 T) and human (7 and 10.5 T) V1 to relate fMRI signals to the single- vessel responses. Lastly, we will develop a model to relate lamina-specific vascular signals to neural activity. In Aim 1, we test the hypothesis that vascular signals selective for stimulus orientation are present in cortical layers 2/3 (and 5/6) while untuned responses occur in layer 4 and pial vessels. Grating visual stimuli will be used, while varying orientation and eye preference (ocular dominance) systematically. Since binocular integration is stronger outside layer 4, eye preference vascular signals should be most prominent in layer 4. In Aim 2, we will test the hypothesis that in any given cortical lamina, glutamate release in regions around an individual blood vessel best accounts for the selectivity of vascular responses compared to spiking activity?in terms of the preferred stimulus orientation and tuning width. Aim 3 is to build a computational model to determine effective minimum voxel size for BOLD fMRI. The model will be tested against simultaneously measured vascular and neural activity to natural scene stimuli using two-photon imaging. If the source signals at the finest spatial scales have laminar specificity, we can correlate laminar-specific fMRI signals to differences in neural processing. To our knowledge, this is the first study that brings together such a wide repertoire of approaches into a single project to understand the neural and laminar basis of fMRI.

NEUROIMAGING CORE ? SUMMARY The Neuroimaging module provides a valuable service to Core Grant investigators who study system-level visual function in humans and monkeys using magnetic resonance (MR) imaging and psychophysics. The module provides the tools and the expertise of our skilled neuroimaging specialist, Dr. Andrea Grant, to support investigators conducting visual neuroimaging experiments. The Neuroimaging module provides software protocols and hardware for visual stimulus presentation to subjects in MR scanners and protocols for experiments ranging from basic retinotopic mapping to studies of the interactions of multiple visual areas during complex tasks with sub-millimeter resolution. The module also provides equipment for monitoring subjects' direction of gaze and patterns of eye movements during experiments. The Neuroimaging specialist advises investigators on experimental design, equipment and software protocols for conducting experiments and trains laboratory personnel on the software used to identify visual areas of the brain. She maintains the visual display equipment at the Center for Magnetic Resonance Research (CMRR).

ABSTRACT The remarkable efficiency of human perception derives from the fact that we do not process each stimulus as a novel event. Instead, past experiences and scene context inform internal, working models of the world that allow us to generate predictions for our physical environment. A leading theory suggests that perceptual predictions are accomplished via flexible normalization: local inhibitory neuronal populations are regulated by long-range connections so that responses are suppressed when they do not provide helpful information about object boundaries. However, the precise neural mechanisms by which the healthy human brain accomplishes this flexible normalization are not known. In order to understand exactly how neural population responses are suppressed or enhanced in response to different scene contexts, we will perform 2-photon imaging in ferret primary visual cortex (V1) to quantify the responses of excitatory and inhibitory neural populations in superficial layers of cortex during several different visual stimulus paradigms. The ferret model is chosen because the imaging techniques necessary to quantify inhibitory neuronal responses are not yet well established in primate models, and while our current knowledge about neural morphology and connections has been derived from mouse models, mouse visual cortex lacks the ?columnar organization? (spatial grouping of neurons with similar response properties) that is a hallmark of primate visual cortex and is present in ferrets. Thus, the ferret model is well-positioned to bridge the gap between mouse models and primate models. First, in order to understand neuronal behaviors in the absence of contextual modulation, we will characterize interactions within a single hypercolumn to small, simple stimuli (sinusoidally modulated luminance gratings) at a range of orientations and contrasts. We hypothesize that parvalbumin-containing (PV+) inhibitory interneurons will demonstrate the sharpest orientation tuning, followed by somatostatin-containing (SOM+) and serotonin-positive (5HTR+) populations. Next. using a Cross Orientation Suppression paradigm, we will test the hypothesis that that SOM+ responses track the overall contrast energy in the stimulus, while PV+ populations reflect suppression of individual grating component representations. Additional experiments with naturalistic textures will test whether these behaviors generalize to stimuli with a broad range of contrasts, orientations, and spatial frequencies. Finally, we will use classical Orientation-Dependent Surround Suppression and Collinear Facilitation paradigms to study how the local inhibitory pool responds to scene context. We hypothesize that the responses of local 5HTR+ neurons will reflect the surrounding stimuli rather than the center stimuli. Together, these experiments will constrain an open-source computational model articulated at the level of the single neuron that will constrain hypotheses about how human perceptual behaviors are linked to specific neuronal populations; this model will be valuable for understanding how perceptual aberrations associated with psychosis might be mapped to the function of specific neuronal subpopulations.