FAS Theses and Dissertations

Schizophrenic hallucinations were studied with regard to their phenomenology. A comparison was made with the "model psychoses"-sleep, hallucinogen psychosis, amphetamine psychosis, and the organic psychoses (especially temporal lobe epilepsy). The dopamine theory of schizophrenia was presented, and then the author's own hypothesis for a neurophysiological mechanism for schizophrenic hallucinations was introduced. This hypothesis is an attempt to explain both the etiology and pathogenesis of schizophrenia. It is based on the dopamine theory, and in particular the mesolimbic (A10) dopamine system. Electrical brain stimulation studies, which evoke experiential hallucinations, along with experimental results showing that these hallucinations are dependent upon the subject's psychological state have caused the author to label this phenomenon contextual hallucinations. Recently, dopamine nerve terminals have been discovered in the limbic cortex and it is known that the neuroleptic drugs antipsychotic effects are in the mesolimbic dopamine system. These findings have led the author to propose that schizophrenic hallucinations are very similar to contextual hallucinations and they are elicited by an improper balance of dopamine in the mesolimbic system. It is thought that stress might psychosocially activate this mechanism. The author also suggests that the additional symptoms of schizophrenia are secondary and peripheral to hallucinations.

This past decade has seen unprecedented success in Artificial Intelligence (AI), pushing the frontiers in ways most experts could have never predicted. However, most of this success has come in the form of performing well inside the data distribution the models have been trained with. Out-of-distribution (OOD) generalization still remains the Achilles’ heel of modern AI. In contrast, biological systems exhibit a remarkable ability to adapt to novel situations. This thesis addresses this critical generalization gap, by studying biological and artificial intelligence in tandem. The work presented includes new mathematical frameworks designed to better formalize generalization, behavioral benchmarks to identify the limits of both human and AI generalization capabilities, experiments to identify the underlying mechanisms driving generalization in both brains and neural networks, and engineering solutions to incorporate these findings to improve AI. To this end, this thesis presents scientific contributions made to the fields of Machine Learning, Computer Vision, Computer Graphics, Computational Neuroscience, and Psychophysics. Throughout the thesis, the goal of this work has been to advance our understanding and improve OOD generalization by working at the intersection of biological and artificial intelligence.

“Is credit given to where credit is due?” This paper examines how economic exposure to COVID affected household credit behavior in Thailand using loan-level panel data from the National Credit Bureau, combined with remotely-sensed night-time light intensity as a proxy for local economic disruption. I find that more COVID-exposed postal codes experienced significantly worse repayment outcomes, with delinquency rising especially strongly for loans originated during the pandemic. At the same time, total outstanding debt often declined in more exposed areas. However, this decline does not appear to reflect healthier household balance sheets. Rather, it is more consistent with tighter effective access to credit, as more exposed borrowers became more likely to make loan inquiries, while also facing a higher probability that those inquiries did not result in new loans. Further product-level results show that these dynamics were not uniform across loan categories — unsecured products such as credit cards and personal loans exhibit clearer signs of repayment distress, while agricultural credit appears more shaped by policy intervention, with compositional shifts toward smaller-amount loans. In addition, borrowers who entered the pandemic with only unsecured debt appear to be systematically more vulnerable to COVID exposure. These findings suggest that COVID-19 affected household credit markets in Thailand through a combination of various channels, including worsening repayment capacity, tighter effective credit supply, and changes in the composition of lending, with effects that react to the end of government debt relief policy and persisting well beyond the initial COVID period. This paper advances the use of remotely-sensed variables in development economics to proxy economic shock at granular geographic scales and holds implications for financial health monitoring and targeted government policy.

Gram-positive bacteria are surrounded by a cell envelope consisting of an inner membrane and a thick peptidoglycan (PG) cell wall that protects cells from lysis due to their high internal turgor pressure. To maintain the integrity of this essential layer, bacteria have evolved signaling pathways that sense and respond to defects that arise during its biogenesis. My thesis focused on defining the molecular basis of these signaling pathways and the logic that underlies them. These studies led to the discovery of two bacterial autoproteolytic domains involved in mechanical force sensing; a signal transduction pathway that increases the levels of the second messenger cyclic-di-AMP in response to defects in the cell wall, thereby reducing the cytoplasmic turgor pressure to prevent lysis; and the identification and characterization of a broadly conserved regulator of the cell wall elongation machinery in Gram-positive bacteria.

Mechanotransduction allows cells to sense and respond to mechanical forces. In eukaryotes, well-studied examples include adhesion GPCRs (aGPCRs) and Notch receptors. aGPCRs undergo autoproteolysis in their extracellular GPCR Autoproteolysis INducing (GAIN) domain but the cleaved fragments remain associated. In response to force they are pulled apart, exposing a tethered agonist. In the case of Notch, the mechanical force exerted on its extracellular domain results in a conformation change, which reveals the cleavage site for a protease that triggers signaling. I discovered two bacterial autoproteolytic domains (SEAL and MAIN) that mediate force sensing in bacteria and have striking parallels with these eukaryotic systems.

Gram-positive bacteria use the sigma/anti-sigma factor pair SigI/RsgI to sense and respond to defects in their cell wall. RsgI is a membrane-embedded sensor-transducer that holds SigI inactive at the cytoplasmic membrane. RsgI contains an extracytoplasmic intrinsically disordered region (IDR) that functions as a cell wall integrity probe. Previous work suggested that RsgI is cleaved in its juxtamembrane domain, but the cleavage products remain stably associated. When the IDR encounters defects in the PG meshwork it was hypothesized that a pulling force separates the cleavage products triggering intramembrane proteolysis of RsgI and activation of SigI. I discovered that RsgI's juxtamembrane domain undergoes enzyme-independent autoproteolysis and established that the cleavage products remain non-covalently associated. In addition, I discovered that this bacterial autoproteolytic domain is structurally and functionally similar to eukaryotic SEA domains that undergo autoproteolysis and have been implicated in mechanotransduction. My findings indicate that RsgI-family members, with their SEA-like (SEAL) domains, share features with SEA and Notch signaling pathways.

This discovery prompted my interest in GAIN domains, which are central to aGPCR signaling but have never been identified in bacteria. Using structural homology searches, I discovered that GAIN domains are broadly conserved among bacteria and archaea. Like their eukaryotic homologs, I demonstrated that these Microbial Autoproteolysis INducing (MAIN) domains are autoproteolytic and undergo cleavage at a conserved motif. Interestingly, MAIN domains are not components of transmembrane signaling complexes. Instead, these domains tether diverse adhesion and enzymatic domains to microbial cell surfaces. The MAIN domain enables the release of these scavenging enzymes in response to force-based stimuli. Strikingly, aGPCRs and MAIN-containing proteins share many of the same adhesion domains, suggesting these protein families share a common origin.

Returning to my interest in stress-response signaling, I identified a mechanistic link between cell envelope stress and the essential second messenger cyclic-di-AMP (c-di-AMP). The targets of c-di-AMP were identified over the past decade, but the signals and logic behind modulating this second messenger were unknown. Working in Bacillus subtilis, I discovered that a membrane complex composed of a c-di-AMP synthase (CdaA) and its IDR-containing regulator (CdaR) sense defects in the cell wall and increase c-di-AMP levels in response. I then established that changes in c-di-AMP levels modulate cellular turgor pressure through changes in cytoplasmic osmolarity. My data indicate that CdaR's IDR senses defects in the cell wall and activates CdaA in response. The resulting increase in c-di-AMP reduces cytoplasmic turgor and prevents lysis. These findings defined the first signal transduction pathway that modulates the levels of c-di-AMP, established that c-di-AMP controls cytoplasmic turgor pressure, and provided the logic behind modulating this second messenger.

Finally, inspired by the rise of in silico protein interaction screening, I built a pipeline for one-versus-proteome Alphafold-Multimer screens. I paired this in silico screening approach with transposon-sequencing screens to identify biologically relevant protein-protein interactions involved in cell envelope synthesis. Using this approach, I discovered and characterized a broadly conserved lipoprotein that regulates cell wall elongation in Gram-positive bacteria.

Altogether, my thesis work uncovered ancient and conserved strategies for trans-envelope signaling in bacteria and revealed common principles of mechanotransduction in microbial and eukaryotic signaling systems. It further uncovered distinct mechanisms by which bacteria sense and respond to the synthesis of their cell wall to prevent explosive lysis.

Across homeostasis, development, and disease, hematopoietic stem cells (HSCs) orchestrate blood production by deploying cis‑regulatory programs that balance quiescence, self‑renewal, and differentiation. With aging, the hematopoietic system undergoes significant changes, characterized by expansion of phenotypic HSCs, reduced lymphoid potential and a dominant myeloid bias. While chronic low grade inflammation and somatic mutations are known drivers of this decline, the extent to which aging related immune dysfunction is encoded as a stable, cell intrinsic epigenetic memory remains an important question for immune rejuvenation. This thesis explores the hypothesis that aging installs persistent chromatin states in long lived HSCs that are inherited by downstream myeloid progeny, thereby driving systemic immune dysfunction. First, we used a combination of aging mouse models and multiomics to establish the stability of hematopoietic aging signatures. By utilizing an ex vivo expansion system to isolate HSCs from the aged bone marrow micro environment, we demonstrate that aging associated chromatin accessibility signatures are not only transient responses to extrinsic inflammatory cues but are cell intrinsic features that persist through expansion and differentiation. Second, we translated these findings to humans by generating a high resolution single cell multiomic atlas of CD34+ HSPCs from young and aged donors. Through ex vivo culture and expansion, we distinguished aging associated cell intrinsic mechanisms from the extrinsic, niche dependent signals, identifying persistent chromatin states that remain stable even after two weeks of culture. And to make direct measurements of these chromatin states from progenitors to effector cells such as CD14 monocytes, we leveraged mitochondrial somatic mutations for clonal tracing (Mito-SHARE-seq). This approach enables us to directly map HSC derived chromatin states to transcriptional phenotypes of CD14 monocytes with the same clonal lineage. Through this analysis, we identified a robust, age specific regulatory program anchored by RUNX and ETS transcription factors. This program facilitates increased accessibility at inflammatory and myeloid primed enhancers, which is clonally propagated to drive a proinflammatory, antigen presenting gene expression profile in CD14 monocytes. Our results reveal that the aged HSC epigenome serves as a stable reservoir of regulatory memory, ensuring the continuous production of primed myeloid cells that could potentially contribute to systemic inflammation. By defining the specific cis-reulatory module and transcription factors that maintain these aging programs, this works provides a mechanistic foundation for targeted rejuvenation strategies to reset the immune system to a more youthful state.

Cellular metabolism powers the mechanical processes that shape biological form and function. During mammalian embryogenesis, mechanical forces organize subcellular organelles and tissue architecture, while mitochondrial metabolism fuels these transformations. Both metabolism and mechanics vary across space and time, and disruptions in their organization are associated with infertility and developmental failure. Despite extensive knowledge of the molecular constituents involved, the physical principles that govern how mechanics and metabolism are spatially organized—and how they interact—remain poorly understood. This dissertation investigates the biophysical coupling between these processes across multiple scales, from subcellular organelles to the early embryo. We first study the meiotic spindle in mouse oocytes, a dynamic microtubule-based structure that organizes chromosomes during cell division. Using polarization microscopy, we show that the spindle microtubule network behaves as a nematic liquid crystal. We further find that chromosomes create tactoid-like voids that are spatially ordered within the metaphase plate. We propose that the deformation of the nematic field around these embedded chromosomes induces long-range repulsion, establishing their regular spacing at the metaphase plate. Next, we investigate how mitochondrial metabolism is spatially patterned relative to the meiotic spindle. We demonstrate that mitochondria are intrinsically heterogeneous in protein composition and that their spatial distribution is governed by a two-step mechanism involving actin-driven cytoplasmic flows and metabolism-dependent microtubule binding. A minimal advection-binding model, in which mitochondria are transported by flows but preferentially bind to microtubules based on their metabolic state, quantitatively explains the emergence and maintenance of these metabolic gradients. Finally, we examine the metabolic changes of human embryos from the zygote to the blastocyst stage using fluorescence-lifetime imaging microscopy (FLIM). Using causal inference, we show that metabolic transitions occur independently of morphological progression and the metabolic state of the one-cell zygote predicts subsequent developmental outcomes. Together, these findings help lay the foundation for understanding how energy metabolism and mechanics interact during early mammalian development.

Conducted in partnership with the Harvard REACT Lab and Project CETI (Cetacean Translation Initiative), a global effort to decode sperm whale communication using machine learning and robotics, this project supports the advancement of autonomous sensing and signal analysis technologies for marine research. To advance CETI’s research in Dominica, our team designed and built a low-cost, battery-efficient, and mechanically robust aerial robotic platform tailored for deployment in harsh marine environments. The primary objective is to localize whales tagged with VHF transmitters, which do not provide GPS data, by measuring signal strength at multiple positions to triangulate their location. The system will autonomously locate and rendezvous with VHF-tagged whales, establish wireless communication with whale-mounted tags, and guide the UAV toward likely surfacing locations. Meeting these goals requires solving key challenges in waterproofing, compact payload integration, endurance, and stability during launch and recovery from research vessels. By combining structural analysis (FEA) and aerodynamic modeling (CFD), we aim to optimize performance and ensure reliability in the field. Ultimately, this project contributes to the development of novel tools for whale bio-acoustics, advancing inter-species communication research and marine conservation.

Scientific human capital plays a central role in knowledge production and innovation within organizations. Yet, the conditions under which scientific expertise translates into organizational performance remain imperfectly understood. This dissertation examines the challenges organizations face in selecting and deploying scientific human capital.

The first essay studies firms’ hiring of scientists from academia. I show that scientists' productivity in academia is only weakly correlated with their productivity in industry. As a result, making hiring decisions based on academic productivity can lead firms to select scientists who perform poorly in industry and are subsequently laid off. The second essay, co-authored with John McKeon, examines how constraints in high-performance computing shape scientists' output. We show that resource scarcity not only reduces the number of publications produced, but also shifts researchers toward less exploratory and less novel projects. The third essay, co-authored with Maria Roche, studies the commercialization of scientific knowledge in entrepreneurial settings. We show that startups whose technologies rely heavily on their inventors’ own scientific research are less likely to achieve successful exit outcomes, especially when the inventor is not involved in founding the venture.

Together, these essays document the challenges organizations face in selecting and deploying scientific human capital, and the conditions under which scientific expertise can be effectively translated into successful outcomes.

Matrices: Female Printmakers in Revolutionary France asserts the identities and artistic oeuvres of 286 women who worked professionally as printmakers in France between the revolutions of 1789 and 1848. This dissertation marks the first dedicated study of French female printmakers of this period, and one of the few comprehensive studies of female printmakers at all, significantly broadening and honing understanding of French female printmakers and of their impact on the artistic, professional, and commercial worlds they operated within. Focused on political imagery, broadly defined, this dissertation foregrounds how female printmakers used print to participate in politics. Informed by hands-on experience printmaking, this dissertation also centers materials and technique of intaglio and lithographic print. In doing so, it proposes a new conceptual methodology – defined as a matrixial approach – to address how individual female printmakers navigated gender and politics. This matrixial methodology knits together deep archival research, social and political context, feminist, queer, and trans theory, and printmaking materiality to propose a way to address and navigate archival absences, particularly for artists or artworks that have been historically marginalized. Merging techniques of expansive archival research and critical fabulation, this dissertation reanimates these historical female printmakers and imagine how they navigated their artistic, gendered, and political identities – critically, through printmaking materials and process. Matrices argues that female artists found in printmaking the ability to work through gender as an identity, societal group, and political class. Female printmakers encountered, pushed upon, experimented within, and challenged print’s fluid conceptualization of gender to unsettle gender definitions and binaries in their social, professional, artistic, and political worlds. Their gender troublings, in this intense political moment, amounted to assertions and redefinitions of citizenship, which was constructed along gendered lines. Crafting a web – a matrix – of the archival, the material, and the theoretical, this dissertation elucidates how female artists encountered the entwined discourses of print and gender through their artistic processes. In the skilled manipulation of their materials, female printmakers found a way to navigate their gender and professional identities, to assert female agency and authorship, and to problematize the gender binaries and biases that circumscribed their participation in artistic and political spheres. Matrices: Female Printmakers in Revolutionary France is organized along two axes: chronologically and biographically. Moving chronologically from 1789 to 1848, each chapter centers on a particular political era and foregrounds three or four artists. Chapter 1, “Revolution, 1789-1799,” examines female artists of the first French Revolution, highlighting the work of engravers Angélique Allais Briceau and Emira Sergent Marceau and foregrounds how their work engaged with the body. Their engagement of print’s bodily nature to manipulate the human figure, Chapter 1 argues, constituted rethinkings of the Revolutionary ideal of republican motherhood in a reconceptualization of female citizenship. The massive reproductive print projects of the Napoleonic era – the Description de l’Égypte and Annales du Musée – are the focus of Chapter 2, “Empire: 1800-1815.” Female printmakers like Thérèse and Marguerite Lingée and Marie Pauline Soyer replicated paintings at the Salon, Napoleon’s plunder, and findings of his Egyptian campaign. Angélique Allais Briceau managed a workshop of female colorists that illuminated the engravings. Chapter 2 rethinks ‘reproductive’ print – prints that replicate another medium often seen as lesser and feminine – to argue how female printmakers promoted and troubled the Empire’s own politics of reproduction. The third chapter, “Restoration: 1815-1830,” reclaims female artists like Anne Emilie and Victorie Constance Bès’ and Joséphine Formentin’s role in the development of lithography, invented in 1796. Putting period anatomical and lithographic treatises in conversation, this chapter investigates the shared language of hypersensitivity within print and gender discourse and how it enabled female artists to problematize gender biases and in turn, ideas of artistic genius tied to the artist’s hand. Lithography’s ties to history preoccupy Chapter 4, “Revolution Again: 1830-1848,” focused on female lithographers like Joséphine Formentin, Cécile Marchand, and the Marie Marguerite Delpech who worked during the revolutionary moment of 1830. This chapter will explore how female lithographers used their medium’s privileged relationship time and history in conversation with a new gendered historical consciousness. Developing what I have termed an aesthetics of inscription, female lithographers experimented with lithography’s relationship to gendered history to inscribe women and themselves into history. The dissertation concludes with an appendix of the 286 female printmakers I have recorded in my research.

Water is essential to the biological processes required for life. However, the same properties that make water so ubiquitous in biology—namely its polarity and ability to form stable hydrogen bonding networks—also make it a poor solvent for nonpolar gases relative to many organic solvents. As a result, increasing the gas capacity of aqueous systems could have crucial impacts in medicine or for technologies requiring efficient transport of gaseous reactants, such as fuel cells or bioreactors. This dissertation investigates aqueous porous liquids, termed microporous water, as a strategy for improving gas solubility in water. Specifically, it describes two approaches to synthesizing microporous materials for use in microporous water and evaluates the feasibility of using the resulting dispersions in biomedical applications.
Chapter One introduces the importance of solvating gases within a liquid medium for industrial applications. Zeolites and metal–organic frameworks are presented as ideal gas sorbents, and field of porous liquids is subsequently discussed to illustrate how these materials have been incorporated into a liquid medium to result in free-flowing liquids with permanent microporosity. Chapter Two proposes a thermodynamic strategy towards creating porous liquids in water whereby microporous nanocrystals with hydrophobic pores resistant to water intrusion and hydrophilic external surfaces can be stably dispersed in water while retaining the porosity of the solid-state porous material. Initial dispersions utilizing pure-silica zeolites and hydrophobic metal–organic frameworks are introduced. Next, we describe the synthesis and characterization of high-silica ZSM-5 nanoparticles for microporous water, yielding a dispersion with record-high oxygen capacities. Finally, we discuss efforts to reproducibly synthesize ZSM-5 with lower Si/Al ratios to probe the effects of framework composition on gas uptake and water intrusion behavior. Chapter Three introduces the utility of encapsulated microbubbles for ultrasound-based medical diagnostics and, in turn, the disadvantages of using these materials compared to more stable microporous frameworks. We discuss our collaborative efforts to quantify the nonlinear cavitation behavior of various microporous water samples and the potential towards using these materials as noninvasive pressure sensors. Afterwards, we describe additional investigations to examine the suitability of aqueous porous liquids as ultrasound contrast agents. Chapter Four addresses the challenge of improving biocompatibility in metal–organic framework nanoparticles. Factors influencing the biocompatibility and hydrophobicity of these materials are considered and we describe our synthetic efforts towards constructing frameworks with biocompatible components. Specifically, we explore strategies to introduce hydrophobicity in an existing magnesium-based framework by utilizing ligands with hydrophobic moieties.