Marc Howard - US grants

DESCRIPTION (provided by applicant): Episodic memory refers to the ability to remember specific occurrences from our experience. This ability is fragile, being affected by diverse brain insults, including the early stages of Alzheimer?s disease, normal aging, ischemic incidents, resection for the treatment of epilepsy, closed head injury, and schizophrenia. This application brings together the applicant?s experience in cognitive modeling of episodic association with the sponsor?s experience in neurophysiology to build a detailed model of the region of the brain most closely involved in episodic memory. Computational techniques will be used to develop the components of the model in a way that is consistent with known physiology. Its ability to describe human memory performances will also be evaluated.

DESCRIPTION (provided by applicant): The relationship between episodic and semantic memory has been a source of much controversy in recent years. Previous work has established the Temporal Context Model (TCM) as a viable model of episodic recall. TCM uses retrieved context to model episodic associations. Latent Semantic Analysis (LSA) builds a semantic representation by using information about the contexts in which different words occur in large bodies of text to approximate their meanings. In LSA, the representation of two words becomes similar if they occur in similar textual contexts. This is reminiscent of the mediated associations (A-B, B-C) paradigm, in which A and C become associated by virtue of having been presented in a similar context (B). The proposed research examines the relationship between episodic and semantic memory using experiment and theory. Experiments study the ability of episodic memory to mimic the context-sensitive property of semantic memory. Using double-function lists (long chains of pairs, A-B, B-C, C-D, etc), we will explore the mediated association paradigm in new depth, testing specific predictions of TCM. Theoretically, TCM will be "trained on" large bodies of text. The resulting semantic representation will be compared to the semantic structure predicted by LSA, the semantic structure of English, and free association norms. By applying a quantitative model of episodic recall to semantic learning, the proposed research could provide considerable insight into the bases of episodic and semantic memory.

It can be argued that the primary adaptive function of memory is not to remember the past, but to predict the future. Knowledge of past experiences combined with an understanding of the present state gives organisms the ability to anticipate future rewards or avoid impending danger. The goal of the proposed research is to develop a mathematical theory describing how people use past history and a representation of the present to predict the future. The theory will be built on a mathematical model of how the history leading up to the present moment can be efficiently compressed into a representation that could be maintained by the brain. The investigators pursue the development of the theory using three techniques. First, in order to see if human learners behave in a way consistent with the hypothesis, the investigators will conduct a series of behavioral experiments using undergraduate students as research subjects. The experiments present the subjects with a series of symbols chosen according to a hidden sequence and ask them to predict the symbols that will follow at various stages. Second, the investigators will conduct computer simulations to train the equations on a large body of naturally-occuring language. Language has a rich temporal structure defined by the way words, and combinations of words, follow one another. Third, the investigators will work to extend the mathematics of the hypothesis to enable it to describe a wide range of phenomena in learning and memory. The large scale goal is to reorient several subfields of cognitive psychology---episodic memory, semantic memory, conditioning, and interval timing---around an understanding of how temporal history is represented and utilized by the human brain.

If successful, the proposed research could have far-reaching practical impacts. It could provide insight into how children and adults learn, leading to better instructional tools. If successful, it would also represent a large step forward in completely automated natural language processing. The rise of electronic communication has led to vast quantities of text---more than could ever be read by even a small army of human readers. Algorithms that can extract knowledge from large quantities of text currently find use in applications as far-ranging as essay grading and other educational applications to intelligence uses. The investigators anticipate that the equations will be much better at extracting knowledge from natural text than several widely-used algorithms. Finally, there may be useful technologies that exploit the ability to predict the future from the past and the present with the level of efficiency that humans can.

Computer vision algorithms examine images and make sense of what these images depict. Current computer vision algorithms are able to interpret images at the level of a typical middle school student for many image interpretation tasks. Recent advances in computer vision have led to rapid technological advances which are still unfolding but affect not only the technology industry, but education, national security and health care. However, these new algorithms are as yet poorly understood and do not describe how natural learners such as a typical middle school student learn to understand the visual world. This proposal draws together a team of cognitive psychologists, neuroscientists, and computer scientists to develop a new class of algorithms for computer vision inspired by the way people learn.

The key insight of this proposal is that human learners, unlike many leading computer vision techniques, make extensive use of the temporal structure of visual experience to extract structure. In the real world the image on the human retina is almost never static. Changes in eye position and movements of the head and body create a rich and complex temporal structure over a range of scales from hundreds of milliseconds up to days and weeks. This proposal a) develops databases of realistic and dynamically changing images in the real world and in immersive virtual reality environments, b) develops computational models for learning visual representations from temporally structured experiences and, c) examines the brain structures supporting representations integrating time and space across scales using fMRI. The algorithms pursued in this project are inspired by recent theoretical work in the neuroscience of scale-invariant memory. However, because the databases will be made publicly available, other researchers will be able to develop other algorithms that exploit temporal and spatial correlations. Taken together, these efforts are intended to catalyze a new generation of techniques for human-like machine learning algorithms with applications in computer vision.

Program Director/Principal Investigator (Last, First, Mlddle): Howard, Marc PROJECT SUMMARY (See lnstrucUons): In recent years, the role of the hippocampal place cell system in constructing a .. map of a spatial environment has received a great deal of attention, including a Nobel Prize In 2014. However, a close examination of the firing of neurons in the place cell system reveals a more general function, providing a map of memory space that includes information not only about spatial coordinates, but also about objects in space and reward contingencies. Moreover, empirical work suggests that the hippocampal theta rhythm provides a mechanism for animals to imagine possible future paths to construct appropriate decisions. Using a combination of empirical work, advanced data analyses and computational modeling, this proposal develops a hypothesis for how the hippocampus and frontal cortex cooperate to navigate this general memory space to inform future behavior. Studies of amnesia patients have made it clear that the hippocampus is important in both memory for the past and imagination for the future. Although a great deal is known about the spatial responsiveness of place cells in the hippocampus, it is not clear how those findings bear on the question of memory-the primary cognitive function of the hippocampus. This proposal builds bridges between the spatial function of the hippocampus and memory by building a computational model for both memory for the past and imagination of the future in a general memory space. The project bridges levels of decription from behavior, to an abstract mathematical framework, to systems neuroscience. Advanced data analytic techniques will be used to test specific predictions of the computational model regarding changes in firing rate and firing relative to ongoing theta oscillations. Finally a novel series of experiments will build a bridge in our understanding of the cooperative role of the hippocampus and orbitofrontal cortex in remembering the past and anticipating the future. RELEVANCE (See instructions): The proposed research may shed new light on disorders associated with changes in temporal discounting, including addiction, ADHD, schizophrenia and bipolar disorder. Moreover, because the computational model relates low-level neurophysiological phenomena with a cognitive-level function, this approach may enable us to better anticipate the effect of pharmacological agents on cognition.

PROJECT SUMMARY/ABSTRACT The Weber-Fechner law is perhaps the oldest quantitative relationship in psychology. Detailed neurophysiology of the visual system demonstrates that neural representations of extrafoveal retinal position obey Weber-Fechner scaling such that the width of the receptive ?eld centered at x goes up like x, creating a neural basis for the behavioral Weber-Fechner law. Behavioral work and cognitive modeling suggests that neural representations of other variables such as time and space in the hippocampus should also obey Weber-Fechner scaling, suggesting that Weber-Fechner scaling is a fundamental principle of brain function. Weber-Fechner scaled representations of time and space constitute a signi?cant computational challenge. A recent proposal for constructing a scale- invariant representations of time and one-dimensional space utilizes the Laplace transform of the past as an intermediate representation. This framework has been shown to be roughly consistent with neurophysiological recordings from hippocampal time cells and place cells. Cognitive models built from this representation can account for a broad range of behavioral ?ndings from a variety of different types of memory tasks. Access to the Laplace transform enables rapid and ?exible computation; preliminary theoretical work suggests a mapping hypothesis between translation of a function de?ned over a Weber-Fechner scaled domain and hippocampal theta oscillations. The proposed research builds towards a general theoretical framework for neural representations of time, space and number. One subproject works to test quantitative predictions of the existing mathematical framework by developing new data analysis tools. These tools will be evaluated on existing ensemble recordings from ro- dent and primate hippocampus and medial prefrontal cortex made available through cooperative agreement with leading neurophysiologists. A second subproject works to establish the potential neurophysiological basis of the equations by building detailed circuit models to implement the equations. These circuit models will draw on known properties from both in vitro and in vivo neurophysiology. A third subproject extends the mathematical basis of the framework from 1-dimesional representations of time, space, and number to more general N -dimensional problems. This approach will derive translation operators and control circuits for translation along more general paths using an approach inspired by differential geometry. Taken together, the three subprojects could result in a substantial step towards a general quantitative neural theory for many aspects of memory and ?exible neural computation.

Program Director/Principal Investigator (Last, First, Mlddle): Howard, Marc PROJECT SUMMARY (See lnstrucUons): In recent years, the role of the hippocampal place cell system in constructing a .. map of a spatial environment has received a great deal of attention, including a Nobel Prize In 2014. However, a close examination of the firing of neurons in the place cell system reveals a more general function, providing a map of memory space that includes information not only about spatial coordinates, but also about objects in space and reward contingencies. Moreover, empirical work suggests that the hippocampal theta rhythm provides a mechanism for animals to imagine possible future paths to construct appropriate decisions. Using a combination of empirical work, advanced data analyses and computational modeling, this proposal develops a hypothesis for how the hippocampus and frontal cortex cooperate to navigate this general memory space to inform future behavior. Studies of amnesia patients have made it clear that the hippocampus is important in both memory for the past and imagination for the future. Although a great deal is known about the spatial responsiveness of place cells in the hippocampus, it is not clear how those findings bear on the question of memory-the primary cognitive function of the hippocampus. This proposal builds bridges between the spatial function of the hippocampus and memory by building a computational model for both memory for the past and imagination of the future in a general memory space. The project bridges levels of decription from behavior, to an abstract mathematical framework, to systems neuroscience. Advanced data analytic techniques will be used to test specific predictions of the computational model regarding changes in firing rate and firing relative to ongoing theta oscillations. Finally a novel series of experiments will build a bridge in our understanding of the cooperative role of the hippocampus and orbitofrontal cortex in remembering the past and anticipating the future. RELEVANCE (See instructions): The proposed research may shed new light on disorders associated with changes in temporal discounting, including addiction, ADHD, schizophrenia and bipolar disorder. Moreover, because the computational model relates low-level neurophysiological phenomena with a cognitive-level function, this approach may enable us to better anticipate the effect of pharmacological agents on cognition.

PROJECT SUMMARY/ABSTRACT Episodic memory is characterized by our ability to remember the spatial and temporal context in which events occur. There is substantial evidence that the hippocampal neuronal activity reflects a representation of space but, until recently, little was known about whether or how the hippocampal neurons encode time. However, recent studies by us have shown that hippocampal neuronal activity provides a temporal context signal that contributes to memory. In addition there is now evidence, substantially accumulated in this project, that hippocampal neurons - called time cells - fire during particular moments in a temporally extended experience, similar to hippocampal place cells that fire associated with particular locations in a spatially extended environment. The proposed studies will continue to explore the nature of temporal representation by the hippocampal system and associated brain areas. Experiments so far have focused on time cell activity during a gap between remembered events in order to identify an unambiguous timing signal in the absence of dynamic external events and while holding place and behavior constant. In the next phase of the project we will explore how time cells organize a sequence of events that compose specific episodes. Also, all recordings of hippocampal time cells have so far been examined only in area CA1. We will examine whether temporal coding is limited to CA1 or widespread in the hippocampus and other medial temporal and prefrontal areas. We will also explore whether temporal sequencing is created within intrinsic hippocampal circuits or whether temporal coding within the hippocampus depends on inputs from cortical areas. These studies will challenge the prevalent view that the hippocampal system is dedicated to spatial navigation and advance our understanding of how this system represents events in their spatiotemporal context.