Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1998. / Includes bibliographical references. / by Matthew C. Tresch. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Brain and Cognitive Sciences, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 189-204). / Why do languages have the words they do instead of some other set of words? In the first part of this thesis, I argue that cognitive and communicative demands strongly influence the structure of the lexicons of natural languages. It is known that words in natural language are distributed such that shorter words are more frequent and occur after more predictive contexts. I provide evidence that, at least in part, this pattern is driven by word shortenings (i.e., chimp -+ chimpanzee) and that word shortenings can be predicted by principles of efficient communication. I also show that, using nonce words with no pre-existing semantic meaning, a Zipfian correlation between length and frequency emerges in freely produced text and that this correlation is driven by participants' tendency to reuse short words more readily than longer words. In addition to word length, I investigate phonetic probability in a corpus of 97 languages. Across a wide variety of languages and language families, phonetic forms are optimized for efficient communication. And, using baseline phonetic models, I show that the words in the lexicons of four languages (English, Dutch, German, and French) are more tightly clustered in phonetic space than would be suggested by chance alone. This thesis depends on standard methods in language research. How reliable is the data that we work with as a field? In the second part of this thesis, I tackle that question by examining two dominant methods in modern language research: behavioral experiments (specifically syntactic priming) and linguistic acceptability judgments. I present data, based on large-scale surveys, showing that many of the standard syntactic and semantic judgments in a mainstream linguistic journal are flawed. Using this data, I construct a Bayesian prior over judgments and give recommendations for performing small sample-size experiments in linguistics that will not overly burden researchers. Finally, I present a large-scale meta-analysis of syntactic priming (the largest meta-analysis of a psycholinguistic phenomenon) and find that, while many priming studies are severely underpowered, there is no evidence of intense p-hacking. / by Kyle Mahowald. / Ph. D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2006. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Includes bibliographical references (p. [175]-211). / In this thesis, I describe a quantitative model that accounts for the circuits and computations of the feedforward path of the ventral stream of visual cortex. This model is consistent with a general theory of visual processing that extends the hierarchical model of [Hubel and Wiesel, 1959] from primary to extrastriate visual areas. It attempts to explain the first few hundred milliseconds of visual processing and "immediate recognition". One of the key elements in the approach is the learning of a generic dictionary of shape components from V2 to IT, which provides an invariant representation to task-specific categorization circuits in higher brain areas. This vocabulary of shape-tuned units is learned in an unsupervised manner from natural images, and constitutes a large and redundant set of image features with different complexities and invariances. This theory significantly extends an earlier approach by [Riesenhuber and Poggio, 1999a] and builds upon several existing neurobiological models and conceptual proposals. First, I present evidence to show that the model can duplicate the tuning properties of neurons in various brain areas (e.g., V1, V4 and IT). / (cont.) In particular, the model agrees with data from V4 about the response of neurons to combinations of simple two-bar stimuli [Reynolds et al., 1999] (within the receptive field of the S2 units) and some of the C2 units in the model show a tuning for boundary conformations which is consistent with recordings from V4 [Pasupathy and Connor, 2001]. Second, I show that not only can the model duplicate the tuning properties of neurons in various brain areas when probed with artificial stimuli, but it can also handle the recognition of objects in the real-world, to the extent of competing with the best computer vision systems. Third, I describe a comparison between the performance of the model and the performance of human observers in a rapid animal vs. non-animal recognition task for which recognition is fast and cortical back-projections are likely to be inactive. Results indicate that the model predicts human performance extremely well when the delay between the stimulus and the mask is about 50 ms. This suggests that cortical back-projections may not play a significant role when the time interval is in this range, and the model may therefore provide a satisfactory description of the feedforward path. / (cont.) Taken together, the evidences suggest that we may have the skeleton of a successful theory of visual cortex. In addition, this may be the first time that a neurobiological model, faithful to the physiology and the anatomy of visual cortex, not only competes with some of the best computer vision systems thus providing a realistic alternative to engineered artificial vision systems, but also achieves performance close to that of humans in a categorization task involving complex natural images. / by Thomas Serre. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1991. / Includes bibliographical references (p. 149-155). / by Joshua J. Stern. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1992. / Includes bibliographical references. / by Daphne Bavelier. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2011. / 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. / Includes bibliographical references (p. 254-302). / The ease with which we recognize visual objects belies the computational difficulty of this feat. Despite the concerted efforts of both biological and computer vision research communities over the last forty years, human-level visual recognition remains an unsolved problem. The impact of a robust yet inexpensive solution would dramatically change computer science and neuroscience, unleashing a host of innovative applications in our modern society. In this thesis, we identify two operational barriers that have obstructed progress towards finding a solution { namely the lack of clear indicators and operational definitions of success, and the currently limited exploration of the staggeringly large hypothesis space of biologically- inspired solutions. To break down these barriers, we first establish new neuroscience-motivated baselines and new suites of fully-controlled benchmarks for object and face recognition. We also compare and contrast a variety of high-level visual systems, both artificial (state-of-the- art computer vision) and biological (humans). Then, we propose a simple high-throughput approach to undertake a systematic exploration of the biologically-inspired model class. By leveraging recent advances in massively parallel computing, we show that it is possible to generate a multitude of candidate models, screen them for desirable properties and discover robust solutions. Finally, we validate the scalability of our approach by showing its potential on large-scale real-world" applications. Taken together, this thesis represents a humble attempt towards an integrated approach to the problem of brain-inspired object recognition { spanning the engineering, specification, evaluation, and application of an interesting set of biologically-inspired ideas, driven and enabled by massively parallel technology. Even relatively early instantiations of this approach yield algorithms that achieve state-of-the-art performance in object recognition tasks and can generalize to other image domains. In addition, it offers insight into which computational ideas may be important for achieving this performance. Such insights can then be "fed back" into the design of new candidate models, constraining the search space and suggesting improvements, further guiding "evolutionary" progress. We hope that our results will point a new way forward, both in the creation of powerful yet simple computer vision systems and in providing insights into the computational underpinnings of biological vision. / by Nicolas Pinto. / 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 (p. 399-407). / This work serves as the first comprehensive investigation into typically developing children's acquisition of subject-to-subject (StS) raising. In particular, it asks how English-speaking children come to comprehend a StS raised sentence like (1) versus its semantically equivalent unraised counterpart (2), and how the presence of an experiencer-phrase affects interpretation (3): (1) John[subscript i] seems [t[subscript i] to be dancing]. (2) It seems that John is dancing. (3) John[subscript i] seems to Mary [t[subscript i] to be dancing]. The acquisition of StS raising is of particular bearing given renewed interest in the acquisition of verbal passives, which share syntactic traits with StS raising and which have been argued to develop late under genetic guidance. Using sentence-picture matching and truth-value judgment experiments, the following comprehension results obtain: e Unraised sentences are acquired early, with most three-year-old children demonstrating mastery. - Sentences involving StS raising over an experiencer (ROE) are delayed until around age seven, with many children incorrectly interpreting them as involving raising-to-object (RtO) syntax. e Sentences involving StS raising with no experiencer (RNE) are likewise delayed until around age seven, with many children incorrectly interpreting them as involving either subject control or copular syntax. - Subject control sentences are acquired early, by at least age three. - The delayed acquisition of StS raising (ROE and RNE) appears to be developmentally linked with the delayed acquisition of verbal passives. The noted delay in comprehension of StS raising occurs despite the fact that StS raising sentences are found to be relatively common in child-directed speech. These data serve to rule out several grammatical acquisition accounts in the literature that have attempted to capture children's delayed comprehension of verbal passives (e.g. External Argument Requirement Hypothesis, Canonical Alignment Hypothesis, and Universal Freezing Hypothesis). The data, however, are both compatible with and predicted by the Universal Phase Requirement (UPR; Wexler, 2004), a grammatical account on which children are claimed to take all vPs to define strong phases for maturational reasons. Only UPR correctly predicts delay for StS raising (with and without an experiencer-phrase) in early child grammar, and a positive correlation with the acquisition of verbal passives. / by Christopher K. Hirsch. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1995. / Includes bibliographical references (leaves 157-165). / by Marc A. Sommer. / Ph.D.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 1994. / Includes bibliographical references (p. 164-167). / by Richard C. Carter. / Ph.D.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Brain and Cognitive Sciences, February 2017. / Cataloged from PDF version of thesis. "September 2016." Vita. / Includes bibliographical references (pages 189-220). / in 68 children born in the United States meets the diagnostic criteria for Autism Spectrum Disorder (ASD), a psychiatric illness that shares a high comorbidity with intellectual disability (ID). Despite the high prevalence of ASD, there are currently no mechanism-based treatments available due to a lack of understanding of the pathophysiological processes in the brain that disrupt behavior in affected individuals. Identifying convergent molecular pathways involved in known genetic causes of ASD and ID may broaden our understanding of these disorders and help advance potential targeted treatments for ASD. Synaptic protein synthesis is essential for modification of the brain through experience and is altered in several genetically-defined disorders, notably fragile X (FX), a heritable cause of ASD and ID. Neural activity directs local protein synthesis via activation of metabotropic glutamate receptor 5 (mGlu₅), yet the mechanism by which mGlu₅ couples to the intracellular signaling pathways that regulate synaptic mRNA translation is poorly understood. In this dissertation, we show that manipulation of two novel targets, [beta]-arrestin2 and glycogen synthase kinase 3[alpha] (GSK3[alpha]) are able to independently modulate translation downstream of mGlu₅ Avoiding dose-limiting consequences and unwanted side effects of globally targeting mGlu₅ signaling, pharmacological inhibition of these targets has the potential to provide significant advantages over first-generation mGlu₅ inhibitors for the treatment of FX. Finally, we show that a mouse model of 16p1 1.2 microdeletion disorder, a polygenic disorder known to confer risk for ASD and ID in humans, shares common features of synaptic dysfunction downstream of mGlu₅ with the Fmr KO mouse. Chronic administration of pharmaceutical agents previously shown to restore synaptic function in the Fmr KO mouse successfully corrected many biochemical, cognitive and behavioral impairments in 16p1 1 .2 df/+ mice supporting the hypothesis that troubled translation downstream of mGlu₅ may be a convergent point of dysfunction between these two genetically-defined disorders. / by Laura J. Stoppel. / Ph. D.