Russell Epstein - US grants

There are few circumstances more fundamental to our existence in the fact that we are creatures that move through space. Almost all of the behaviors are essential to our survival - locating food, finding a mate, and avoiding predators - require navigation through the external world. A half-century of research suggests that humans and animals use cognitive "maps" of the large-scale spatial structure of the world to facilitate this ability. Much of the information used to form these maps is initially acquired from vision. But this presents a puzzle: how do we transform visuospatial information initially acquired from specific views and specific instances into a viewpoint-invariant representation of environmental space that can be used for navigational planning? The research proposed here will use functional magnetic resonance imaging (fMRI) to address this question. On the basis of previous work, we hypothesize that the human brain supports at least three distinct hierarchical levels of spatial representation: (1) it represents the locations of objects and surfaces relative to various body parts such as the eye and hand (body-space);(2) it represents the relationship between the trunk of the body and a coordinate frame anchored on the set of immovable surfaces that defines the local scene (scene-space);(3) it represents the the location and orientation of the observer within the larger environment, including relative to locations that are currently "over the horizon" (world-space). The first part of the proposed research will examine scene-space representations in parahippocampal and retrosplenial cortices with particular emphasis on uncovering the mechanisms by which these representations are learned from experience and change over time. The second part of the proposed research will examine the neural mechanisms involved in representing body-space and will functionally distinguish these mechanisms from those involved in representing scene-space. The third part of the proposed research will examine the neural mechanisms involved in representing world-space. Although this research does not examine spatial processing in nonsighted individuals, the results obtained will help us to distinguislvbetween processes that are inextricably tied to vision and those that are relatively amodal. As such, this knowledge could be critical for development of rehabilitation strategies in the blind.

DESCRIPTION (provided by applicant): A core concern of cognitive neuroscience is understanding the representational distinctions made by different brain regions. Two techniques are commonly used in functional magnetic resonance imaging (fMRI) studies to investigate this issue: multivoxel pattern analysis (MVPA) and fMRI adaptation (fMRIa). Whereas MVPA examines the voxelwise response patterns elicited by different stimuli to determine which items elicit patterns that are distinguishable, fMRIa examines the effect of repeating items over time under the hypothesis that repetition of representationally-similar items will elicit a reduced response. Although MVPA and fMRIa have become part of the central toolkit of cognitive neuroscience, little is known for certain about the neuronal mechanisms that underlie these two techniques. Consequently, it has been difficult to relate results obtained through MVPA to results obtained through fMRIa. Indeed, previous work from our lab has shown that apparently inconsistent results can be obtained from fMRIa and MVPA, even when these techniques are applied to the same dataset. The main goal of the current project is to test several hypotheses about the different neuronal mechanisms that might underlie MVPA and fMRIa. Specifically, we will test whether fMRIa indexes neuronal or subneuronal tuning (Exp. 1), stimulus similarity or perceptual expectations (Exp. 2), and we will test the spatial scale of functional organization indexed by MVPA (Exp. 3). The focus of this preliminary project will be on fMRI signals in the parahippocampal place area (PPA) during viewing of real-world visual scenes and in the lateral occipital complex (LOC) during the viewing of objects. Although the current experiments do not directly examine recovery or disease, they will, if successful, provide important analytical tools for such investigations. For example, attempts to use MVPA and fMRIa to understand plasticity and cortical reorganization in amblyopia, or after recovery of sight in the blind, can only bear fruit if the interpretation of the findings in terms of underlying neural mechanisms is fully understood.

DESCRIPTION (provided by applicant): A fundamental aspect of our existence is the fact that we move through space. We do not do so randomly~ rather, we use a variety of different strategies to efficiently reach our navigational goals. One such strategy is landmark-based piloting, which is the use of stable topological features to determine one's location and orientation relative to the enduring spatial structure of the world. The current proposal describes a research program in which functional magnetic resonance imaging (fMRI) and cognitive behavioral testing are used to understand the neural mechanisms that underlie landmark-based piloting. Previous neuroimaging and neuropsychological studies, including several from our laboratory, have identified a network of brain regions that might be critical. These include the parahippocampal place area (PPA), the retrosplenial complex (RSC), and the hippocampus. However, the precise way in which this network implements this function remains unclear. We will use recent technical innovations for the study of neural representation-multi-voxel pattern analyses (MVPA) and fMRI adaptation (fMRIa) - to test the idea that this network can be fractionated into functional subsystems tied to three cognitive mechanisms: a landmark-recognition mechanism, a spatial orientation mechanism, and a cognitive map. More specifically, Aim 1 will test the hypothesis that the PPA encodes fixed landmarks and local spatial coordinate frames that are anchored to these landmarks. Aim 2 will test the hypothesis that RSC supports recovery of location and facing direction relative to global (beyond-the-horizon) coordinate frames. Aim 3 will test the idea that the hippocampus encodes a cognitive map, by characterizing the spatial code indexed by a recently-discovered hippocampal distance signal. If successful, this research will result in a detailed theory of the neural basis of landmak-based piloting. This knowledge will have important health implications in two domains. First, understanding the mechanisms that underlie landmark-based piloting is critical for the development of rehabilitation strategies and navigational aids for the blind. Second, because the brain regions investigated are often impacted early in neurodegenerative diseases such as Alzheimer's dementia, the knowledge gained about these systems will be useful for diagnosing and managing these diseases.

Project Summary A fundamental aspect of our existence is the fact that we live in a spatially extended world. To survive and flourish in this world, we must have some method for navigating efficiently from place to place. The current project focuses on the neural mechanisms that underlie landmark- based navigation (LBN)?navigation that is guided by spatially stable elements of the environment. To implement LBN, a person must be able to 1) perceive the local environment, 2) use features of the local environment to determine their location and orientation in the world, and 3) plan a route that takes them from their current location to their navigational goal. In previous funding periods, we have used functional magnetic resonance imaging (fMRI) and other methods to identify regions of the human brain that mediate these operations and to assign functional roles to these regions. Now we seek to integrate these findings to understand how these neural/cognitive components work together to implement LBN during realistic navigational episodes. In aim 1 we will identify the spatial representations that are simultaneously active in the human brain during dynamic navigation, including representations of location and heading, and representations of the navigational goal. In aim 2 we seek to understand the remapping mechanisms that allow a navigator to negotiate a complex world that contains multiple local environments. In aim 3 we will delineate the reorientation mechanisms that allow a navigator to establish (or re-establish) their sense of place and direction after losing their bearings. Together, these operations?knowing where we are in the world and the location of our goal, distinguishing between different spatial environments, and recovering our bearings when we are disoriented?constitute core elements of spatial navigation. Understanding the neural mechanisms that underlie these elements would be a major and sustained intellectual advance in an area that has long been a central topic of investigation in psychology and neuroscience.

Project Summary: Spatial navigation is a challenge that must be met by all mobile organisms. Although much is known about the neural mechanisms that underlie spatial navigation in animals, the neurocognitive basis of spatial navigation in humans is much less clear. This is partly because of technical limitations: the best tool currently available for noninvasive measurement of brain activity is functional magnetic resonance imaging (fMRI), but traditional fMRI analysis methods are not well-suited for identifying representations elicited in dynamic, naturalistic situations such as unconstrained navigational episodes. The current project will attempt to overcome this limitation by using a recent technical innovation in fMRI analysis?voxel-wise encoding modeling?to identify the neural representations that mediate active navigation. Specifically, we will model the fMRI response during navigation within a virtual-reality city in terms of the spatial variables that are known to have known cellular coordinates correlates in rodents and non-human primates, and we will then evaluate the model by testing whether it predicts fMRI response in a held-out dataset not used to train the model. Additionally, we will explore whether the representations thus revealed suffice to solve the problem of simultaneous localization and mapping (SLAM); that is, whether they allow one to keep track of one's position during exploration of a novel environment while simultaneously learning its layout. Aim 1 is to build the voxel-wise encoding model and to use it to identify the neural representations within specific brain regions that mediate dynamic navigation. Aim 2 is to examine generalization across environments by testing whether a model trained in one virtual environment suffices to solve the problem of SLAM in another. If successful, we anticipate that this project will have a major and sustained impact on the field by achieving a quantitative description of how spatial information is encoded in multiple regions of the human brain during dynamic real-time navigation. This will allow us to test specific hypotheses about spatial representations developed from the animal literature, and potentially allow this literature to be leveraged to better understand multiple cognitive functions that rely on the same underlying neural architecture, including spatial cognition, episodic memory, and imagination. Moreover, this project will provide the essential technical foundation for future tests of novel hypotheses about the physiological basis of navigation.

Project Summary The ability to navigate from one place to another is essential for a flourishing and autonomous human life. Cognitive scientists have long believed that navigation in humans and animals is guided by mental representations of the spatial structure of the world, which are referred to as ?cognitive maps? because they play a functional role that is similar to physical maps. Consistent with this idea, electrophysiologists have identified neurons in rodent brains that fire as a function of spatial variables that are essential elements of a cognitive map, such as location, distance, and heading direction, while cognitive neuroscientists have investigated possible neural correlates of cognitive maps in several regions of the human brain, including the hippocampal formation (HF) and the retrosplenial complex (RSC). Notably, these brain regions are also known to be essential for several important cognitive functions besides spatial navigation, including memory, imagination, and thinking about the future. However, despite this previous work, there remain two crucial gaps in our knowledge. First, we have an incomplete understanding of how cognitive maps are represented in the human brain. Behavioral studies indicate that our spatial knowledge is often fragmented, hierarchically organized, and distorted in multiple ways compared to metric truth, and we do not yet understand how these ?real? cognitive maps are represented in brain structures such as HF and RSC. Most notably, we do not understand how the brain divides environments into spatial parts (such as rooms within a building, or neighborhoods in a city), and how it then combines these parts into a larger whole. Second, we do not yet have a good theory of how spatial cognitive maps can be applied to non-spatial domains, thus allowing brain structures such as HF and RSC to mediate both spatial and nonspatial functions. The current project will address these issues by using advanced neuroimaging techniques, such as multivoxel pattern analysis and individual difference analyses to: (i) identify the neural mechanisms that allow the brain to encode subspaces within a larger space; (ii) delineate the neural processes by which subspaces representations are combined into a larger cognitive map, and (iii) understand how the principles underlying spatial cognitive maps can be applied to nonspatial domains. This project has the potential to make a major and sustained advance in the field by resolving longstanding questions about the cognitive and neural systems underlying spatial navigation, and by providing fundamental knowledge about how the brain mediates a wide range of basic cognitive functions, including not just navigation, but also semantic and episodic memory, prospective thinking, and reasoning.