Washington University Medical Center
Washington University St. Louis: Open ScholarshipNot a member yet
26344 research outputs found
Sort by
Segmentation and Memory in Interactive, Inferential Events
No full textThough everyday life is continuous, people tend to understand and remember experiences as discrete events separated by boundaries. Event segmentation theory (EST) (Zacks et al., 2007) describes these event boundaries as being driven by prediction error. People have an internal model of how an event unfolds, uses this to form predictions, and then places a boundary when a new model is necessary. Prior studies suggest that episodic memory is enhanced for information occurring near event boundaries, though less is known about these effects in interactive environments. To assess this, I designed an experimental paradigm where participants uncover changes in the underlying situation through interaction, leading to event boundaries. In Experiment 1, I report the findings of a free recall study with the word rule inference task (WRIT). The primary finding being that while event boundaries structured recall, memory for items immediately following a boundary was impaired, and contextual certainty, rather than prediction error, was coupled with improved recall. Experiment 2 was developed to delineate whether the findings of the first experiment were due to effects of rule inference or temporal uncertainty. Temporal certainty improved overall recall, but the post-boundary deficit persisted unless participants were given full certainty over the rule sequence and the bounds of the event. Extending this approach to recognition memory (Experiment 3), I assess whether the effects of certainty are differential when one does not have to guide search using events. In Experiment 4, I consider whether the post-boundary memory deficit is linked to event model binding, attentional resource shifts, or a mixture of these processes. This is assessed by using temporal distance and order judgments to examine how items are represented closer or further from each other. Across these studies, there were two key deviations from event memory effects in passive paradigms. First, gains in contextual certainty regarding the active rule, rather than prediction error, were the primary predictor of memory performance. Second, memory for items experienced around event boundaries was rarely enhanced and was instead hindered for items encoded right after boundaries. These results challenge the emphasis on prediction error for memory effects in interactive contexts and suggest that the cognitive demands of active inference and model updating at event boundaries can impair, rather than enhance, encoding. The findings highlight how the cognitive cost of interaction and building certainty alter memory for events, indicating the need for more nuanced models. More specifically, models of event memory should incorporate costs of cognitive control that interact with encoding benefits, dampening them when control costs are high
Plant-Microbe Interactions Across Scales: Diversity, Disease, and Stress in Herbaceous Plant Communities
Get PDFPlants interact consequentially with microbes spanning mutualism to pathogenicity, with broader effects on large-scale plant community structure. These plant communties, in turn, shape below and aboveground microbial communities through a variety of mechanisms. However, plant diversity-microbial diversity relationships remain broadly inconsistent across environments and among different plant and microbial assemblages. In my dissertation, I work to bridge some of these gaps in plant-microbial ecology in three experimental chapters assessing relationships between plant diversity and microbial diversity, including the influence of belowground-aboveground microbiome interactions, the influence of drought stress, and incorporating different spatial and experimental scales. First, I explore the microbial underpinnings of a plant-soil feedback between Monarda fistulosa and Plantago lanceolata, which only experienced a negative plant-soil feedback under both drought and pathogen stress. I sequenced the soil microbial communities associated with the treatment experiencing a feedback to better understand the microbial drivers. Second, I scale up from a two-species system to a many-species mesocosm experimental system, with the aim to understand how microbial communities are influenced by multiple co-occurring species and interactions with soil inoculation and drought stress. Finally, I assess the relationship between wild plant diversity and aboveground pathogen and herbivory damage, exploring how and whether dilution effects operate in Missouri glade restoration sites. Together, I find that plant species assemblages and environmental stress in the form of drought are important shapers of microbial community composition. Interpreting the ecological mechanisms underpinning microbial community responses to plant diversity differences woud benefit from higher taxonomic resolution of under-described microbial taxa, and higher-detail functional annotations. My research contributes to our understanding that multiple factors simultaneously and interactively shape patterns in plant diversity-microbial diversity relationships
Auto-Enrollment, Auto-Escalation, and the Need for Retirement Plan Portability: Implications for SECURE 2.0
Get PDFThe SECURE 2.0 Act of 2022 aims to improve the retirement security of low-income U.S. households by making retirement saving easier, automatic, and more attractive. The act’s provisions focus on employer-sponsored retirement-savings plans. This brief examines how low-wage employees utilize employer-sponsored retirement-plan features, how they say they would respond to automatic contribution increases, and how job transitions impact their retirement savings decisions. Data come from the Workforce Economic Inclusion and Mobility survey, which collects data from a nationally representative sample of vulnerable workers in the United States.
This is the second research brief in a series examining the implications of the SECURE 2.0 Act for the retirement security of low-wage workers in the United States. All briefs in the Implications for SECURE 2.0 series can be found here
VECTOR: A 3U CubeSat for Educational and Algorithmic Development in Space-Based Optical Imaging
Get PDFVECTOR (Versatile Educational Controls Testbed for Optical Response) is a 3U CubeSat designed to serve as an educational platform and on-orbit laboratory, advancing the understanding of image processing, search, and control algorithms for a rapid-slewing optical follow-on telescope. It is designed as the first flight of a bus using a minimal electronic interface to support complex imaging missions in a 3U form factor, balancing competing requirements of modularity/ease of repair against SWaP (Size, Weight, and Power).
VECTOR will include a Canon lens, an optical CMOS sensor, and an FPGA for the imaging pipeline. Torque rods and large reaction wheels will be used to maximize agility. The detection software pipeline is designed to include modern software-based motion-blur correction with point-spread-functions estimated from IMU data and star-spread from the images. This configurable software pipeline enables different algorithms to be easily tested, an almost unprecedented spaceflight capability.
VECTOR is designed to detect transient events, including rapid responses to the NASA GCN and IPN, direct notifications of GRBs, FRBs, and AGNs, and events of public interest. When not observing targets, VECTOR will remove a star from its star map, create a probability density function for the removed star, and search for the “new” target. Leveraging its configurable FPGA imaging pipeline and control systems, it can test pointing/navigation algorithms under varying conditions. A guest investigator program invites students and the public to develop and submit algorithms, with this being considered for use by the WashU Electrical and Systems Engineering department
A Mechanistic Exploration of How Neighborhood Crime Exposure Relates to Neonatal Brain Function, Early Externalizing Behaviors, and Callous-Unemotional Traits
No full textPregnancy through early childhood is a period of tremendous neural and behavioral development. Neuronal birth and migration occur prenatally and result in the formation of neonatal functional and structural networks. These early brain networks are a building block for postnatal development wherein infants rapidly learn a variety of new skills, including socioemotional skills such as emotion regulation and moral understanding. During the acquisition of these skills, even typically developing toddlers can act aggressively or impulsively; however, toddlers with high levels of externalizing behaviors are more likely to develop clinical externalizing disorders later in childhood and adolescence. As such, it is crucial to understand the mechanisms that influence the development of early externalizing behaviors and underlying brain function. Prior work has demonstrated that exposure to adversity may alter developmental trajectories; yet, the specific influence of neighborhood crime exposure, which is thought to be a salient environmental threat, remains unclear. The first aim of this dissertation was to determine whether living in a high crime neighborhood prenatally was related to reductions in neonatal functional connectivity and whether maternal psychosocial stress mediated this relationship (Chapter 2). The second aim examined whether neighborhood crime exposure was also related to early externalizing behaviors and whether neonatal functional connectivity and/or parenting mediated this relationship (Chapter 3). The third aim asked whether callous-unemotional traits, which are present in a group of children with particularly severe and persistent externalizing behaviors, were related to neighborhood crime exposure (Chapter 4). The fourth and final aim was to examine what other mechanisms might influence the development of callous-unemotional traits by examining parenting behaviors and maternal emotional intelligence (Chapter 5). By addressing these aims, this dissertation increased knowledge of the mechanisms underlying neonatal functional connectivity, early externalizing behaviors, and callous-unemotional traits. In the future, these factors may be potential therapeutic targets that could lessen the burden of disruptive behaviors
The Conservative Case for Leaving Harvard Alone
No full textThe Supreme Court precedent allowing the IRS to revoke a university’s tax-exempt status is a textualist’s nightmare
Controlling Cross-Presentation of Integrin Binding and BMP Mimicking Peptides to Control Mesenchymal Stromal Cell Fate
Get PDFGrowth factors are attractive for their ability to drive proliferation, differentiation and other biological programs critical for tissue repair. In many cases, growth factors are used to improve cell survival and integration of transplanted cells such as Mesenchymal Stromal Cells (MSC) within host tissue. However, growth factors, either alone or in combination with cells like MSC, are typically administered at supraphysiological doses that can lead to dangerous off-target effects when the growth factors diffuse away from the site where they are originally delivered. Short peptide mimics of full-length peptides can be a promising alternative since they can be covalently conjugated to biomaterials and therefore localized to the site of interest. However, these short peptide “mimetics” often lack the potency of their full-length counterparts. In this dissertation, I sought to improve the potency of growth factor mimicking and cell adhesive peptides to encourage differentiation of MSC. I focused on a peptide that mimics the knuckle epitope (KE) of Bone Morphogenetic Protein 2 (BMP-2). KE is from the region of BMP2 thought to bind the BMP receptor. I used orthogonal click chemistries (strain promoted-azide alkyne cycloaddition (SPAAC) and maleimide-thiol) to graft either KE or integrin binding cyclo-RGD (cRGD) peptides to alginate hydrogels. I first optimized SPAAC for grafting cRGD to alginate hydrogels, and found that cRGD grafted in this manner, through a heterobifunctional crosslinker (BCN-amine) was more potent than cRGD that was directly grafted to alginate via carbodiimide chemistry. I then sought to increase the potency of the KE peptide. MSC expressed key markers of osteogenesis (RUNX2, Alkaline Phosphatase, ALP and osteocalcin, OCN) in a KE-dose dependent manner. Interestingly, when co-presented with cRGD, KE at a high concentration partially mimicked the activity of full length BMP-2. To investigate the possibility that the mechanism through which high KE dose induced MSC osteogenesis was through close proximity between KE and cRGD, I used the orthogonal chemistries to create alginate polymers grafted with both KE and cRGD (bivalent), forcing the peptides to be closer to one another than when the peptides were grafted to different alginate chains (monovalent) at a constant overall peptide concentration. In these studies, I found that close proximity between cRGD and KE can explain much, but not all, of the need for high KE dosage to induce MSC osteogenesis. The methodology and results gained these studies could be used as a guideline for future models, either as a platform for studying another kind of growth factor and/or integrin binding peptide, or as an application for safer growth factor delivery
SAFE or Not Safe? Perspectives of MO Gender-Affirming Care Providers in Response to Anti-Trans Care Legislation
Get PDFReceiving gender-affirming care is a life-saving intervention for many, yet the current wave of transphobic rhetoric and anti-trans legislation, such as the Missouri (MO) Senate Bill 49 “Save Adolescents From Experimentation” (SAFE) Act, continue to restrict trans individuals from accessing gender-affirming care. Previous literature depicts medical providers as enactors of medical power and authority, producing knowledges of trans subjectivities, constructing and controlling trans bodies, and gatekeeping medical care from trans individuals; however, the changing roles, labor, and acts of resistance trans care providers perform under the recent context of heightened social surveillance and legal restrictions have yet to be documented. This study analyzes the nuanced positionality of MO gender-affirming care providers through semi-structured interviews, demonstrating how the current medical institution reproduces imbalanced patient-provider power dynamics on the basis of empirical objectivity and places a double bind on providers between quietly maintaining trans care and vocally advocating for trans rights. Many providers of trans care also exist at the intersection of multiple disciplines, roles, and identities, demonstrating the necessity for reproductive justice frameworks in future research to critically evaluate the systems that produce intersectional harm for non-White, non-cis, non-heterosexual, physically and/or mentally disabled people
Quantifying Noise in Optical Redox Ratio Imaging
Get PDFCharacterizing cellular metabolism through noninvasive imaging of metabolic cofactors NAD(P)H and FAD has become a powerful diagnostic tool, with the Optical Redox Ratio (ORR) serving as a key quantitative metric. However, widespread adoption of ORR imaging is hindered by the lack of standardized image quality requirements. This study addresses that gap by analytically deriving expressions for the variance of ORR measurements as a function of photon shot noise and dark current contributions. Using statistical conditioning, we modeled the expected pixelwise ORR variance and verified the model through simulated image sets and experimental data. We observed that variance decreases with higher photon counts and frame averaging, aligning with theoretical predictions. The results showed strong correlation between analytical and empirical variance, even under varying noise conditions. This work provides a quantitative foundation for evaluating ORR image quality and can inform future acquisition protocols and data processing. Ongoing efforts include developing a MATLAB application for variance analysis and preparing a manuscript to disseminate these findings
Training Safety Control Filters Using High-dimensional and Un-labeled Data
Get PDFSynthesizing control policies that preserve the safety of autonomous systems is a challenge that remains to be solved. Towards that goal, control barrier functions (CBFs) have been developed as mathematical constructs that can be used in real-time to correct safety-violating nominal actions to ones which preserve the safety of control systems. However, synthesizing CBFs using correct-by-construction methods has not been scalable. Instead, recent research has proposed data-driven approaches for learning CBFs in the form of neural networks. Two main challenges face such approaches: (1) labeling states as unsafe or safe ones requires the knowledge of the states in the backward reachable set of the failure set--the true dynamics-dependent unsafe set, and (2) in the case of systems with high-dimensional observations, such as images and point clouds, enormous amount of data is needed to train these neural observation-based CBFs, which is expensive to obtain in robotic domains. We tackle the first challenge by using inverse constraint learning to infer a neural classifier that defines the backward reachable set from expert trajectories and use it to label sampled states. This method outperforms baselines and performs comparably to a CBF trained with ground truth labels in four environments. We tackle the second challenge by using existing vision models which are pre-trained on large and diverse datasets as frozen perception backbones on top of which latent dynamics and neural observation-based CBFs are trained. Our experimental results indicate that the resulting filters are competitive with those that have access to the ground truth state