22484 research outputs found
Sort by
A Powers Framework for Mental Action
No full textMental actions are things we do with our minds. Consider inferring, deliberating, imagining, remembering, calculating, and so on. I introduce a non-reductive alternative to standard causalist accounts of mental action that understands such action in terms of dispositions for performing mental actions. I call this alternative the powers framework. On the powers framework, habitual and skillful mental actions are themselves infused with practical intelligence by being expressions of the agent’s rational tendencies and capacities, respectively. The intelligence exemplified in the performance of habitual and skillful mental actions stems from the agent’s having shaped the corresponding tendencies and capacities through training and practice. In this way, mental habits and skills are ‘second nature’ to us. I substantiate the powers framework by giving an account of imagining as a type of skillful mental action. In particular, I argue that imagination is a power to construct representations and select their contents as a means to performing learned behaviors like pretending, engaging with fiction, predicting others’ behavior, reasoning about possibility and necessity, reasoning hypothetically or counterfactually about contingent matters of fact, and even imagining for its own sake. I extend the account of imagining to episodic remembering. I argue that such remembering, considered as a mental action, is a kind of imagining by virtue of the agent’s constructing a representation and selecting its content as a means of performing the learned behavior of navigating her personal past
Computational Models of Political Learning and Belief Polarization: From Individuals through Social Networks to Disruptive Systems
No full textThis project examines the dynamic interplay of information dissemination, belief formation, and social interactions within the political communication landscape representing the new digital media environment, particularly focusing on social media. Employing agent-based modeling, this research explores how cognitive biases, social network structures, and the presence of misinformation influence political belief formation in digital environments.
The study is structured into four essays: the first provides a theoretical agenda and methodological justification, while the subsequent three modeling essays build upon one another to enrich our understanding of political learning. The baseline model investigates how individual cognitive biases affect the assessment of information source credibility. The second model expands this model to incorporate the influences of social network structures, examining how interactions within these networks impact belief formation and consensus. The last essay explores the effects of disinformation, introduced by an adversarial jammer, on social learning and consensus building.
Key findings reveal that while cognitive biases significantly shape information credibility assessments, the structure and interconnectivity of social networks play critical roles in mitigating misleading information and facilitating consensus. Furthermore, the simulation results indicate that networks characterized by diverse and robust social ties can effectively dilute the impacts of disinformation.
This dissertation contributes to the fields of political communication and computational social science by demonstrating how agent-based simulations can provide nuanced insights into the complex interactions of individual behaviors and macro-level network effects. It also highlights practical implications for designing interventions to counter mis(dis)information and enhance the resilience of public discourse in the digital age. This research highlights the necessity of integrating both psychological and social dimensions to foster a well-informed public capable of navigating the modern political information ecosystem
The Role of Silence in Salvatore Sciarrino’s Infinito Nero, and a Series of Original Compositions Based on Silence
No full textThis dissertation focuses on the multifaceted role of silence in the structure of Salvatore Sciarrino's (b.1947) composition, Infinito Nero. The study employs a graphic technique I call "outsidegraphy" to explore how silence is intentionally incorporated and represented in the score. This method condenses the musical elements, strategically removing rests and staves, providing a comprehensive visual overview that emphasizes the significance of silence in the composition. Outsidegraphy centers on perception, offering a silence-focused lens to interpret the composer's intentions. The resulting single-image representation of the score integrates both temporal and spatial elements, presenting a holistic image of the piece's silence. Salvatore Sciarrino's compositions, particularly Infinito Nero, challenge conventional notions of sound and silence. Sonic elements are deliberately placed at or below the perceptual threshold, creating a deliberate distance between the audience and the sound platform. This intentional distancing encourages listeners to actively engage with hidden intricacies of the music, akin to scrutinizing a seemingly blank canvas to reveal concealed patterns and colors. The dissertation also explores the concept of "self-erasing boxes" in Sciarrino's compositions, highlighting the transient nature of certain sounds. These sounds, meticulously notated in the score, lack stability, lingering at the threshold of perception. The fragility of these sounds, influenced by factors like azzerare dynamics and the absence of recognizable musical cells, contributes to Sciarrino's intentional creation of transient sonic elements.
The original compositions include A Miniature Opera: Censorship and The Unnoticed Dance No. 1 & 2, exploring the convergence of sociopolitical and musical silence. In the exploration of censorship, the musical elements illustrate the process through reduction, converting sounds into noises, and gradually introducing silence. The title, A Miniature Opera: Censorship, signifies metaphorical censorship, where all elements of the opera undergo suppression, leaving only four instrumentalist characters on the stage. Amidst these alterations, a singular action endures—the foot strike, symbolizing a potent form of protest. Within The Unnoticed Dances, the established theme undergoes an elimination process, not solely in sounds but also in the visual aspect of the score
Preterm Delivery and Cardiovascular Health in Women of Reproductive Age
No full textPregnancy is considered a stress test for cardiovascular risk in women of childbearing age. Women who experience preterm delivery are seen as failing to adapt to the dramatic physiological changes during pregnancy (i.e., increased insulin resistance, increased triglycerides, and increased cardiac output with accompanying reduced systemic vascular resistance and blood pressure) and have been found to have a higher risk of cardiovascular disease (CVD) later in life. However, the mechanisms linking these two conditions remain unclear. Accumulating evidence indicates that preterm delivery is associated with cardiometabolic risk factors before, during, and after pregnancy, suggesting a shared pathway between preterm delivery and CVD. My dissertation uses longitudinal data to explore why preterm delivery is associated with an elevated risk for CVD from a life course perspective, considering traditional cardiometabolic risk factors at different periods relative to pregnancy.
In the first paper, I investigated whether the unfavorable changes in the cardiometabolic profile associated with preterm delivery begin before, during, or after childbearing. I found that preterm delivery was associated with unfavorable patterns of change in diastolic blood pressure and adiposity that originate during the childbearing years and persist or worsen later in life.
In the second paper, I examined how cardiometabolic risk factors and preterm delivery relate to coronary artery calcification. Compared to women with term delivery, impaired fasting glucose and elevated blood pressure are more detrimental to the development of subclinical atherosclerosis among those with preterm delivery.
In the third paper, I tested the hypothesis that changes in blood lipids from during pregnancy to years after pregnancy may vary based on delivery outcomes. I found that women with spontaneous preterm delivery experience blunted changes in lipids from mid- to post-pregnancy, potentially reflecting impaired ability to adapt and recover from pregnancy.
By examining cardiometabolic risk factor changes across different periods relative to pregnancy, this dissertation enhances our understanding of the mechanistic pathways between preterm delivery and future maternal CVD. This is of public health significance because women with a history of preterm delivery could benefit from closer monitoring of cardiometabolic risk factors for better risk stratification and preventative intervention for early CVD
Identification and characterization of TANGO2 protein and its interaction with metabolites
No full textPathogenic variants in the Transport and Golgi Organization 2 Homolog (TANGO2) gene lead to an autosomal recessive disorder associated with TANGO2 deficiency. The disorder can exhibit a spectrum of clinical presentations, varying in severity. Patients may experience rhabdomyolysis, acute metabolic crises, cardiomyopathy, and/or cardiac arrhythmias. Studies suggest TANGO2’s involvement in mitochondrial dysfunction and endoplasmic reticulum/Golgi trafficking, yet its exact function remains unknown. Currently, there is no known effective treatment for TANGO2 Deficiency Disorder (TDD). Recent research revealed that pantothenic acid (vitamin B5) supplementation can ameliorate clinical symptoms and rescue the phenotype in TANGO2 mutant Drosophila, though the mechanism remains unclear. Until recently, there was no confirmed crystal structure of the TANGO2 protein. I hypothesize that the determination of the TANGO2 protein three-dimensional structure and its tissue expression can shed light on its function and interaction with other proteins/metabolites.
TANGO2 cDNA with a C-terminal His-tag coding motif and wildtype TANGO2 cDNA without tag were cloned in pET-27b(+) expression plasmids. BL21, containing a plasmid expressing GroEL and GroES chaperonins, bacterial cells were transformed using both plasmids via electroporation. The TANGO2 C-His Tagged protein was purified by fast protein liquid chromatography (FPLC) using the ÄKTA Pure chromatography system, a HisTrap column, and a MonoQ column. The wildtype without tag TANGO2 was purified by ammonium sulfate fractionation, a DEAE Sepharose column using two different buffer conditions, and a MonoQ column. Elution from the column was followed using SDS-PAGE/western blot. The level of purity was assessed using SDS-PAGE gel and protein bands visualized by Coomassie blue staining. Western blotting was used for surveying TANGO2 expression in mouse tissues.
The purified C-His tagged TANGO2 protein was used to obtain the TANGO2 crystal structure as well as screening potential metabolites that interact with the TANGO2 protein. The screening of mice tissues showed universal TANGO2 expression in all tissues as well as no discernable difference between males’ and females’ tissues. The heart and brain tissue had the highest TANGO2 expression levels. This research holds public health significance because it can advance our knowledge of TANGO2 structure/function and guide future efforts to develop an effective treatment for TDD patients
Strategic Alliances and Capability Evolution: Two Essays about Big Oil Going Green
No full textThis dissertation investigates the role of network dynamics and strategic capabilities in guiding the oil and gas industry’s transition toward sustainable energy practices. It addresses the incumbent's dilemma of whether to explore new technological frontiers or adapt existing operations to meet evolving environmental and market demands. Through a detailed analysis of alliances and joint ventures from the SDC Platinum database spanning 1982 to 2022, the 2 essays explore how firms can leverage their positions within an industry’s alliance networks to foster innovation and adapt strategically. The first essay analyzes how firms utilize their network positions to enhance strategic capabilities and respond to environmental pressures, finding that reducing reliance on core capabilities adversely affects the transition to greener energy. The second essay employs Temporal Exponential Random Graph Models (TERGMs) to examine the impact of broader network structures on strategic capabilities, revealing that open and dense networks enhance firms' strategic flexibility and innovation potential. Collectively, the findings demonstrate that maintaining a balance between exploring new capabilities and sustaining core operations is crucial for enabling firms to navigate the complexities of shifting to sustainable practices effectively. This work advances theory-building in strategic management by illustrating the dynamic interplay between network embeddedness and capability development during industry transitions. It provides practical insights for industry leaders on structuring strategic alliances to enhance innovation while preserving operational stability. For policymakers, the dissertation underscores the importance of supporting policies that facilitate both technological innovation and the optimization of existing infrastructures to accelerate sustainable transformations within the energy sector
Cell-type specific investigation of molecular rhythms in the prefrontal cortex
No full textCircadian rhythms are ~24-hour rhythms in physiological processes that are generated by a transcriptional-translational feedback loop called the molecular clock. The molecular clock then generates rhythmic expression of downstream targets, including up to 43% of protein coding genes in mice. Recently, studies have shown that molecular rhythms are extensively altered in the prefrontal cortex (PFC) of individuals with psychiatric diseases; however, the contribution of cell-type specific rhythms remains unknown. Using RNA sequencing, we examine translatomic rhythms in two cell types in the mouse PFC, parvalbumin (PV) and pyramidal cells, both of which have been heavily implicated in schizophrenia. We further assess the consequences of molecular clock dysfunction on PV cell properties as well as on psychiatric disease relevant behavior. Finally, we expand our analysis to compare transcriptomic rhythms between the mouse PFC and two psychiatric disease relevant subregions of the PFC in humans. We find that while the molecular clock oscillates in synchrony between PV and pyramidal cells, nearly twice as many transcripts are rhythmic in pyramidal cells than in PV cells. Moreover, rhythmic transcripts in pyramidal cells show broad overlap between sexes while rhythmic transcripts in PV cells are largely distinct between sexes, consistent with the observed sex-specific diurnal changes in PV cell electrophysiological properties. Interestingly, processes associated with mitochondrial function are highly enriched for rhythmic transcripts in PV cells from males. Furthermore, disruption of the molecular clock leads to reduced PV expression and increased excitability of PV cells, as well as deficits in cognitive flexibility. In our cross-species analysis, we find that while rhythms in core
v
molecular clock components are conserved across species in the PFC, the difference in peak times between the mouse PFC and human PFC subregions differs by sex. Nevertheless, overall transcriptomic rhythms are largely unique between the mouse PFC and human PFC. As up to 80% of drug targets show rhythms in gene expression, this work provides critical insight into cross species translation of preclinical findings. Moreover, our cell-type specific findings suggest that there are sex and cell-type specific vulnerabilities to circadian rhythm disruption, which may have implications for psychiatric diseases such as schizophrenia
FPGA Acceleration of k-mer Counting using On-Chip HBM2 and oneAPI
No full textCounting substrings of an arbitrary length k (k -mers) is the single most time-consuming step of de novo genome sequencing. Sequencing machines generate large quantities of data (>100s of GBs per genome). Processing this genetic information requires frequent memory accesses into data structures considerably larger than available cache, leading to a memory-bound runtime. Stemming from the gap between processor and memory speed, this bottleneck can be alleviated through alternative computing architectures. Recent FPGA devices, equipped with on-chip High-Bandwidth Memory (HBM), enable custom architectures to employ high-capacity, high-bandwidth memory to address memory-bound tasks. This research investigates accelerating k-mers counting with one such device, the BittWare 520N-MX, a Stratix 10 FPGA with 16 GB of on-chip HBM2. The architecture was designed using Intel’s oneAPI framework. The accelerator architecture leverages inherent parallelism in the algorithm via multiple parallel hash functions, along with partitioning data structures across multiple memory banks, and employing multiple independent parallel processing pipelines on the device to maximize throughput. The accelerator achieves 57.98M k-mers per second, 3.80× more than the throughput-optimized CPU version and 5.85× more than the original CPU app. This was done despite the clock speeds in the oneAPI design falling well below the board’s maximum frequency. Multiple methods of improving the clock speeds were attempted but were ultimately unsuccessful. OneAPI was able to achieve speedup over the CPU using the FPGA equipped with the on-chip HBM2, but there is the potential for additional performance improvement with higher FPGA clock speeds
Efficient Hardware and Software Design for On-Device Learning of Video Streams
No full textVideo detection has been widely applied in edge devices with heterogeneous accelerators for video pre-processing and real-time inference. Our work aims at video detection in an elderly care robotic application. When an elderly falls, a DNN model is necessary to both recognize the action and localize the temporal position where the action happens. Such a task is called temporal action localization (TAL). Currently, a TAL DNN is first trained on the centralized cloud and then deployed to edge devices. However, when devices come to a new environment, it is necessary to use online streaming data to update the pre-trained model to improve accuracy. To adapt to new environments with less time consumption while protecting users’ privacy, it is desirable for the models to continuously and directly learn from local data on the device.
However, existing software and hardware systems are mainly designed for inference, while training is not the main concern. To enable efficient on-device learning for video learning, three main challenges need to be solved. The first is how to develop practical algorithms on the software side that are feasible to improve the TAL model directly from on-device a single long video stream rather than pre-divided video datasets collected in the cloud. The second challenge is how to implement the algorithm on individual resource-limited edge devices with severe computation complexity and intra-device communication bottlenecks. Third, we need to address how heterogeneous accelerators on the devices collaborate in the training process for efficient training with computation complexity across different accelerators and the inter-device communication bottleneck. The first challenge is from the software perspective, while the latter two challenges are from the hardware perspective.
To address the first challenge, we have developed a weakly-supervised on-device learning framework for streaming videos that contains all class actions and does not require any laborious manual pre-segmentation. To solve the second challenge, we have developed an efficient DNN training accelerator that can achieve end-to-end training on a single resource-limited low-power edge-level device. For the last challenge, a multi-accelerator training algorithm is proposed to enable efficient CNN training on edge devices with heterogeneous training accelerators
Investigating and Improving Student Understanding of Mechanics, Electricity and Magnetism, Quantum Mechanics, and Thermodynamics using Conceptual Surveys
No full textValidated conceptual multiple-choice surveys administered before and after instruction in relevant concepts can be useful tools to gauge the effectiveness of curricula and pedagogical strategies. Here we discuss the use of four different validated surveys to investigate student understanding: The Energy and Momentum Conceptual Survey (EMCS), the Conceptual Survey of Electricity and Magnetism (CSEM), the Quantum Mechanics Formalisms and Postulates Survey (QMFPS), and the Survey of Thermodynamic Processes and First and Second Laws-Long (STPFaSL-Long). The EMCS and the CSEM were used to investigate progression in student understanding of introductory-level physics concepts by administering them to both introductory and advanced level students as a pre- and a post-test. The cross-sectional performance of students on these introductory level physics concepts reveals which concepts remain challenging for students past their first year of physics and how the most common incorrect answers may evolve on various questions from introductory to advanced levels. The QMFPS was used to investigate co-construction and construction of knowledge in advanced quantum mechanics courses. Students were asked to take the QMFPS individually and then in a group of 2-3 with no feedback on their initial performance. Co-construction occurred when all students in a group originally answered a question incorrectly on their own but answered it correctly as a group after discussion. Construction occurred when only one student answered a question correctly individually, but the group answered it correctly. By comparing construction and co-construction of knowledge, we were able to determine which concepts were difficult and which concepts were easier for students so that they could answer them correctly by working together without feedback from the instructor. Lastly, we discuss the development and validation of the STPFaSL-Long survey instrument. This survey instrument was administered as a pre- and post-test to introductory-level students, upper-division students in an upper-level thermodynamics course, and graduate students across many different universities. We discuss the context dependance of student responses along with student difficulties with thermodynamic variables and the first and second laws of thermodynamics. We used both written data and interviews from the STPFaSL survey instrument to investigate these issues