Research

Williams is a powerhouse of scientific research and a national leader in the training of future scientists. Faculty and students collaborate on original research, contributing meaningfully to the creation of new knowledge.

At Williams, teaching and research come together in the context of a broad liberal arts education that emphasizes critical thinking and exploring questions from all sides. The college ranks near the top among liberal arts colleges in faculty research supported by the National Science Foundation, and both faculty and student research receive extraordinary support from the college.

For a look at faculty research activity and faculty-student collaborations, please see our annual Report of Science.

Students can pursue research early in their careers, and in a variety of ways. During the semester, many students spend 5-10 hours per week in labs as paid research assistants. During Winter Study in January, students can pursue independent research on campus or in the field, or collaborate with faculty members and other students. And in the summer, nearly 200 students work with faculty on campus as fully funded research fellows, and still more travel to conduct research nearby and abroad.

A sampling of recent graduate theses:

  • Planetary Nebulae as Tracers of the Chemical History of the Andromeda Galaxy
    Kerrin G. Hensley

    This thesis presents analysis of emission-line spectra of eight outer disk planetary nebulae in the Andromeda Galaxy (M31) obtained with the 10.4-meter Gran Telescopio Canarias. The galactocentric radii of these eight planetary nebulae meet or exceed the radii of previously studied objects in M31. We observe unexpectedly high oxygen abundances of these outer disk objects and a relatively shallow oxygen abundance gradient. Possible reasons for these findings, such as an interaction between M31 and neighboring galaxy M33 about 2-3 billion years ago, are discussed.

    Measurement of the Hyperfine Structure of the 7P1/2 state and 8P1/2 state in 205Tl and 203Tl
    Gabrielle D. Vukasin

    We report a final value of the hyperfine splitting of the 7P1/2 state of 205Tl and 203Tl made using a two-step excitation. Our final values are 2173.3(8) MHz and 2153.2(7) MHz respectively. We also measured the isotope shift of the 7S1/2 → 7P1/2 transition to be 534.4(9) MHz. These experimental hyperfine splitting values are ≈ 20 MHz larger than those measured by another group in 1988 [1]. Our values bring the experimental values closer to the theoretical values published in 2001 [2]. Our data consists of spectra taken by scanning the second-step laser 6 GHz. For precise measurement of these spectra, we stabilize the first-step excitation using a method called laser locking. Using the same experimental layout, we are now working to measure the hyperfine splitting of the 8P1/2 state of both isotopes.

  • Identification of discrete, intermingled hypocretin neuron subpopulations
    Manasi Iyer

    Hypocretins (Hcrts) are neuropeptides exclusively expressed in a population of neurons in the hypothalamus. Hcrt neurons have been shown to play a role in both addiction and arousal behaviors.  Recently, it has been hypothesized that there are actually two distinct subpopulations of Hcrt neurons that each project to downstream nuclei that either affect addiction or arousal. However, direct anatomical proof of the existence of these two subpopulations remains elusive. Manasi’s thesis directly tested the hypothesis that Hcrt neurons differentially project to either addiction or arousal neurons using injections of fluorescent retrograde tracers in mice.  She found that there was no overlap between Hcrt neurons that projected to arousal centers versus Hcrt neurons that projected to addiction centers.  Interestingly, she also found that these subpopulations of Hcrt neurons are intermingled and are not biased in the medial/lateral or anterior/posterior direction. This research suggests a mechanism by which Hcrt neurons can influence both addiction- and arousal-related behaviors.

    Variation in the benefits of ant-tending to the treehopper Publilia concava
    Eric Hagen

    Mutualism is a widespread phenomenon in nature, but the dynamics are still poorly understood both in general and for specific mutualisms. Here I use model selection to look at how variation in levels of ant tending affects the treehopper Publilia concava survivorship. Publilia concava is an insect in the family Membracidae which feeds on the phloem of tall goldenrod (Solidago altissima) and excretes a sugary waste-product called honeydew. Ants collect this honeydew and in turn provide predator defense, feeding facilitation, and other benefits. Results here show that show that the magnitude of net-benefit to treehoppers varies among study site locations as well as temporally within a single season and across seasons. For treehoppers in 2013 the best model explains these differences with shifts in predation type and pressure.

  • Synthesis and Characterization of Phenylenevinylene Oligomers
    and Functionalized Aromatic Systems
    Peter L. Clement

    Our group has synthesized a library of phenylenevinylene oligomers and functionalized anthracenes, perylenes, and pyrenes. While functionalization in previous work was centered on the incorporation of electron-withdrawing moieties (Br, CN, CHO, BMes2) to conjugated systems, this year’s work incorporates electron-donating moieties, in the form of diphenylamine groups, into conjugated systems. Diphenylamine functionalized anthracene, perylene and pyrene were all synthesized. Additionally, the characterization of the library of compounds synthesized was expanded from UV/Vis and fluorescence spectroscopies to include characterization by cyclic voltammetry (CV). Using a combination of UV/Vis and CV, electron-withdrawing groups were shown to primarily decrease the energy of the LUMO energy while electron-donating groups were shown to primarily increase the energy of the HOMO energy.

    Activation of σU and the Putative Role of PrsU in Streptomyces coelicolor
    Jessica Monterrosa Mena

    Streptomyces coelicolor is a model species of gram positive-soil dwelling bacteria, with a complex life cycle and multicellular differentiation. This study focused on the role of stress response sigma factor σU, in the life cycle of Streptomyces coelicolor. Disruption of the gene encoding RsuA, a protein believed to be a negative regulator of the sigma factor of σU, inhibits development and antibiotic production in S. coelicolor. Due to homology between the B. subtilis σW regulon and σU, and observations about acidic environmental stressors activating σU, the PrsU protein is hypothesized to be part of the SigU system. Prsu is hypothesized to be involved in detecting acid or comparable environmental stressors and respond by degrading RsuA, thereby activating σU to participate in gene transcription. To investigate whether PrsU is involved in regulating the σU system, we used mutant strains to determine how mutations of the SigU system affect antibiotic synthesis, strain sensitivity to acid, and σU activation by acid stressors.

    In this study we found differences in acid sensitivity and antibiotic production between sigU, rsuA, and prsU mutants. In antibiotic assays, Bld, sigU-rsuA and prsU mutants demonstrated different secretion of undecylprodigiosin and actinorhodin antibiotics compared to the WT strain. In acid sensitivity assays, sigU-rsuA and prsU mutants showed increased sensitivity to a variety of organic acids including acetic acid and oxalic acid, but diminished sensitivity to citric acid as compared to the WT. Finally, luciferase reporter assays suggested that σU activity in WT strains can be induced by acid, but results are inclusive due to methodological problems with control strains. The results of this study support the hypothesis that PrsU is involved in activation of the SigU system, but suggests that complicated mechanisms are at work in σU activation.

  • Decoupling and Coalescing Race Checks
    Parker S. Finch

    Much of our computing infrastructure utilizes multicore processors and multiprocessor hardware. Such systems can concurrently execute many software threads of control to improve responsiveness and performance, but the potential for unintentional interference between concurrent threads makes it difficult to ensure the reliability of multithreaded software. Automated tools to identify concurrency errors have the potential to make this task easier. Perhaps the most fundamental concurrency error is a race condition. A race condition consists of two threads accessing (and at least one modifying) the same piece of data at the same time. While dynamic data race detection algorithms have improved in recent years, the overhead of race detection in array-intensive programs remains prohibitive. One promising insight is that arrays are often accessed via common patterns that enable compression of the information maintained by a dynamic race detector, as well as a reduction in the number of checks performed. However, a purely dynamic implementation of this compression technique failed to realize a corresponding decrease in run-time overhead due to the cost of inferring those access patterns at run time. We explore statically annotating programs with the array access patterns that appear at run time to eliminate the need to infer them. Our results show that this can effectively reduce the run-time overhead of race detection on programs with identifiable array access patterns.

    Multi-Class Feature Selection
    Joshua E. Geller

    Feature selection involves identifying an “optimal” set of features in order to make a classification task more efficient or effective. There are three major approaches to feature selection in the context of classifier learning. One of these - the wrapper method - is a general class of algorithms that can be applied to any underlying classifier, but the choice of the classifier learner impacts the features selected. An interesting issue arises when the goal is multi-class classification but the underlying classifier learner is fundamentally designed for two-class problems, as is the case for support vector machines (SVMs). These types of classifiers handle multi-class tasks with an ensemble of binary classifiers. Current multi-class feature selection approaches operate on the black box level, using one uniform feature set for all underlying binary classifiers. I propose a framework for multi-class feature selection wrapper methods that includes the black box approach, but also introduces two additional multi-class feature selection approaches: the composite individual and the composite summed. I investigate a number of properties of this three-fold framework as well as multi-class feature selection in general.

  • For Meat’s Sake: Anticipated Hypocrisy and Environmental Information Avoidance
    Peter Drews

    Information avoidance has long been a topic of study in the health domain, yet it has rarely been applied to environmental psychology. In three studies, I examine whether feelings of anticipated hypocrisy may lead to avoidance of environmental information, in the form of meat-eating’s negative consequences for the environment. In Study 1, participants rated either images of appealing meat, disgusting meat, or cookies, and then read a paragraph about either meat’s negative environmental consequences or a control paragraph. I then had participants rate their preferences for viewing four videos, one of which related to environmental issues. In Study 2, participants rated either pictures of nature or of meat, and then either advocated for the environment or did nothing. They were then asked whether or not they wished to view a video linking meat-eating to environmental devastation. In Study 3, participants either advocated for the environment or did nothing. Then, some were asked whether they wished to see the same meat-eating video as in Study 2, while others were asked whether or not they wished to view a video about the damage international corporations do to the environment. These findings suggest that anticipated feelings of hypocrisy may lead people away from information about the environment.

    Fitting in to Feel Good: Explicit & Implicit Predictors of Conformity
    Fanny Mlawer

    The present study examined the unique and joint contributions that implicit and explicit measures of self-esteem (SE), need to belong (NTB), fear of negative evaluation (FNE), rejection sensitivity (RS) and importance of popularity (IMP) make to the prediction of behavioral conformity. In an initial testing session, participants completed explicit measures of the five constructs in addition to providing preliminary answers on a comic rating task and a dot discrimination task. In the second session, participants rerated a subset of the comics and dot discrimination items in the presence of supposed “prior peer ratings” before completing implicit tasks measuring the five constructs. Results revealed a significant three-way interaction between implicit FNE, explicit FNE and gender on self-reported conformity, as well as a two-way interaction between implicit and explicit self-esteem on behavioral conformity in the comic rating task. Specifically, we found that among males high in implicit FNE, those with corresponding high explicit FNE were significantly more likely to report succumbing to peer pressure than those with low social avoidance, and, among females low in implicit FNE, participants with high explicit FNE reported higher susceptibility to peer pressure. We also found that on the comic rating task among those low in implicit SE, individuals with correspondingly low explicit SE conformed significantly more to false peer feedback than those with higher levels of explicit SE.

 

Complete theses are available by department here.