Thesis: S.M. in Neuroscience, Massachusetts Institute of Technology, Department of Brain and Cognitive Sciences, 2015. / Cataloged from PDF version of thesis. / Includes bibliographical references (page 17). / This work establishes a benchmark by which to measure models of the ventral stream using crowd-sourced human behavioral measurements. We collected human error patterns on an object recognition task across a variety of images. By comparing the error pattern of these models to the error pattern of humans, we can measure how similar to the human behavior the model's behavior is. Each model we tested was composed of two parts: an encoding phase which translates images to features, and a decoding phase which translates features to a classifier decision. We measured the behavioral consistency of three encoder models: a convolutional neural network, and a particular view of neural activity of either are V4 or IT. We measured three decoder models: logistic regression and 2 different types of support vector machines. We found the most consistent error pattern to come from a combination of IT neurons and a logistic regression but found that this model performed far worse than humans. After accounting for performance, the only model that was not invalidated was a combination of IT neurons and an SVM. / by Diego Ardila. / S.M. in Neuroscience

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1997. / Vita. / Includes bibliographical references. / by Stephen A.B. Gilbert. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2008. / Includes bibliographical references (p. 101-106). / Two key fields of computational neuroscience involve, respectively, the analysis of experimental recordings to understand the functional properties of neurons, and modeling how neurons and networks process sensory information in order to represent the environment. In both of these endeavors, it is crucial to understand and quantify uncertainty - when describing how the brain itself draws conclusions about the physical world, and when the experimenter interprets neuronal data. Bayesian modeling and inference methods provide many advantages for doing so. Three projects are presented that illustrate the advantages of the Bayesian approach. In the first, Markov chain Monte Carlo (MCMC) sampling methods were used to answer a range of scientific questions that arise in the analysis of physiological data from tuning curve experiments; in addition, a software toolbox is described that makes these methods widely accessible. In the second project, the model developed in the first project was extended to describe the detailed dynamics of orientation tuning in neurons in cat primary visual cortex. Using more sophisticated sampling-based inference methods, this model was applied to answer specific scientific questions about the tuning properties of a recorded population. The final project uses a Bayesian model to provide a normative explanation of sensory adaptation phenomena. The model was able to explain a range of detailed physiological adaptation phenomena. / by Beau D. Cronin. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2010. / Cataloged from PDF version of thesis. / Includes bibliographical references. / Most neurons are born during the embryonic period to become the building blocks for a variety of brain circuits. However, two brain regions only start to assemble during the postnatal period. Both brain areas, olfactory bulb and dentate gyrus, mainly accommodate the integration of new neurons during the postnatal period, and continuously receive new neurons throughout animals' life. In this thesis, I used the rat olfactory bulb (OB) as a model system to address two important issues regarding the integration and plasticity of new neurons generated during the postnatal period. The first feature of postnatal neurogenesis is that when new neurons arrive and integrate into an adult OB, only half of neurons can ultimately survive. However, what form of activity pattern determines the survival of new neurons remains unclear. Using NaChBac sodium channels to selectively alter the intrinsic excitability of new neurons in vivo, this manipulation reveals that neuronal survival critically depends on the level of membrane depolarization. Once neurons integrate and survive in the brain circuits, neurons have the capability of monitoring their activity level and adaptively maintain their membrane excitability within the operational range. How they achieve the long-term stability of membrane excitability remains unclear. By altering the resting membrane potential of individual neurons in vivo, OB granule neurons are found to use a subthreshold parameter, resting membrane potential, to guide the compensatory changes of intrinsic ion channels and synaptic receptors. In summary, studies from this thesis have revealed the cellular mechanisms underlying neuronal survival in an in vivo brain circuit. I also uncover a novel form of homeostatic computation by which granule neurons preferentially use the subthreshold membrane potential response rather than spiking rates as a set point. / by Chia-Wei Lin. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2007. / Includes bibliographical references. / The effects of observed object motion on object perception are examined in two sets of studies. The first section of the thesis provides a thorough examination of various untested aspects of the basic "temporal association" hypothesis, which suggests that object motion provides a principled basis for linking distinct images together if they appear within small time intervals. Using familiar and unfamiliar objects undergoing various forms of non-rigid motion, I ask how well this simple hypothesis predicts behavior in change detection and categorization tasks. The results favor a modified version of the hypothesis which operates over a population of units, such that increases in generalization also produce increases in image sensitivity. The observed effects of long-term knowledge concerning object appearance and expected patterns of motion also force additional modifications of the initial hypothesis to incorporate interactions between learned predictions and recent experience. Specifically, the tendency to alter patterns of generalization following dynamic exposure appears to be contingent on the stability of the direction of movement through appearance space. / (cont.) Consistent with this expanded model, performance in our categorization task appears to depend heavily on whether or not a coherent direction of movement through appearance space can be determined across both categories to be learned. In the second section of the thesis, I report the results of two parametric analyses of image encoding following dynamic exposure. In each case, I ask how the movement of an object up to the presentation of particular image affects an observers' ability to accurately recall that image. Novel, rigidly rotating objects are used in both cases to characterize the influence of appearance dynamics on short and long-term image encoding. In both cases, I find that local appearance change over time exerts a powerful influence on encoding, suggesting that both immediate percepts and visual memory are modulated by the recent past. The result is a complex picture of dynamic object perception that goes far beyond the basic principle of object motion as a tool for learning invariant recognition. / by Benjamin J. Balas. / Ph.D.

Thesis (Ph. D. in Neuroscience)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, February 2013. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. "February 2013." Page 192 blank. / Includes bibliographical references. / The neuronal serine/threonine kinase cyclin-dependent kinase 5 (Cdk5) is activated by its regulatory subunit, p35, to post-translationally modify substrates through phosphorylation. In this thesis, I provide several lines of evidence that Cdk5 plays a critical role in synaptic function and plasticity. First, we characterized the function of Cdk5 in learning and memory by region-specific Cdk5 ablation. From multiple Cdk5 conditional knockout mouse models, we determined that Cdk5 is essential for memory formation and synaptic plasticity. Loss of Cdk5 in the hippocampus disrupts the cAMP pathway due to increased phosphodiesterase proteins. This dysregulation of cAMP signaling can be attenuated by a phosphodiesterase inhibitor to restore levels of protein phosphorylation, synaptic plasticity, and memory. Moreover, forebrain-specific deletion of Cdk5 affected multiple aspects of behavior that can partially be rescued by lithium treatment. We next identified the N-type calcium channels as a presynaptic substrate of Cdk5. We described how Cdk5-mediated phosphorylation of the N-type calcium channel increased calcium influx and channel open probability. This in turn enhanced the association of the N-type calcium channel with the active zone protein RIM1, which impacted vesicle docking and neurotransmission. Finally, we identified the postsynaptic density protein Shank3 as a Cdk5 substrate and observed that Cdk5-mediated phosphorylation of Shank3 plays a critical role in maintaining dendritic spine morphology and synaptic plasticity. Our collective results demonstrate a central role for Cdk5 in regulating both presynaptic and postsynaptic functions and provide better insight into how specific targets of Cdk5 can impact a general mechanism underlying synaptic transmission, synaptic plasticity, and cognitive function. / by Susan C. Su. / Ph.D.in Neuroscience

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Brain and Cognitive Sciences, 2015. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 221-236). / This thesis develops formal computational models of intuitive theories, in particular intuitive physics and intuitive psychology, which form the basis of commonsense reasoning. The overarching formal framework is that of hierarchical Bayesian models, which see the mind as having domain-specific hypotheses about how the world works. The work first extends models of intuitive psychology to include higher-level social utilities, arguing against a pure 'classifier' view. Second, the work extends models of intuitive physics by introducing a ontological hierarchy of physics concepts, and examining how well people can reason about novel dynamic displays. I then examine the question of learning intuitive theories in general, arguing that an algorithmic approach based on stochastic search can address several puzzles of learning, including the 'chicken and egg' problem of concept learning. Finally, I argue the need for a joint theory-space for reasoning about intuitive physics and intuitive psychology, and provide such a simplified space in the form of a generative model for a novel domain called Lineland. Taken together, these results forge links between formal modeling, intuitive theories, and cognitive development. / by Tomer David Ullman. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1995. / Includes bibliographical references (p. 193-212). / by Zoubin Ghahramani. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1995. / Includes bibliographical references (p. 409-433). / by Diana Kathryn Smetters. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2011. / Cataloged from PDF version of thesis. / Includes bibliographical references. / The field of neuroscience has the potential to address profound questions including explaining how neural activity enables complex behaviors and conscious experience. However, currently the field is a long way from understanding these issues, and progress has been slow. One of the main problems holding back the pace of discovery is that it is still unclear how to interpret neural activity once it has been recorded. This lack of understanding has led to many different data analysis methods, which makes it difficult to evaluate the validity and importance of many reported results. If a clearer understanding of how to interpret neural data existed, it should be much easier to answer other questions about how the brain functions. In this thesis I describe how to use a data analysis method called 'neural population decoding' to analyze data in a way that is potentially more relevant for understanding neural information processing. By applying this method in novel ways to data from several vision experiments, I am able to make several new discoveries, including the fact that abstract category information is coded in the inferior temporal cortex (ITC) and prefrontal cortex (PFC) by dynamic patterns of neural activity, and that when a monkey attends to an object in a cluttered display, the pattern of ITC activity returns to a state that is similar to when the attended object is presented alone. These findings are not only interesting for insights that they give into the content and coding of information in high level visual areas, but they also demonstrate the benefits of using neural population decoding to analyze data. Thus, the methods developed in this thesis should enable more rapid progress toward an algorithmic level understanding of vision and information processing in other neural systems. / by Ethan M. Meyers. / Ph.D.