a
M
Project in Archaeology

ARCHAEOLOGY AND MATERIALS SCIENCE: The Red, Etched Stones from Ancient Caesarea: Copies or Trade?

TEAM

Michelle Aluko, Sarah Bansal, Gavin M. Brunelli, Thapati Chinthapatla, Zachary Cruz, Randy Fylstra, Arabella Go, Tucker Gregory, Emma Haig, Henry Lee, Kandace Po, Charles Quinn

Advisor: Dr. Maria Masucci

Assistant: Daylin Evans and David Hwang

ABSTRACT

Etched carnelian beads are among the rarest ornamental artifacts of the ancient world, widely regarded as products of specialized South Asian bead-making traditions. While these beads have been recovered from various archaeological sites throughout Asia, Northern Africa, and the Middle East, there is no published archaeological record of these artifacts being found in Caesarea Maritima, a major Mediterranean Roman port city. This study integrates archaeometry, experimental archaeology, and geochemical analysis techniques to investigate eight carnelian beads from Caesarea Maritima, including two etched artifacts. ATR-FT-IR was used to verify mineral composition. From there, microscopic examination of drill-holes and their impressions along with quantitative Drill Hole Taper Value analysis were used to compare drilling techniques. Stylistic comparisons with published etching patterns were then conducted alongside experimental attempts to recreate the etching process. ATR-FT-IR confirmed that the artifacts were genuine carnelian. Morphological analyses revealed substantial differences in all but three of the beads, indicating that most beads were likely produced in various workshops as opposed to a single production center; the three (non-etched) beads with closely related manufacturing characteristics were likely products of the same workshop. Stylistic evidence further suggests that one etched artifact displayed characteristics of Indus-related production while the other etched artifact exhibited features related to Near Eastern or Anatolian tradition. Drill hole analyses corroborate these conclusions based on chronological and geographical specific techniques. This study’s findings support that these artifacts represent an ancient trade network connection between Caesarea Maritima, Indus peoples, and Anatolian peoples, respectively. This study provides insight into technological diffusion and long-distance exchange routes in the ancient Mediterranean.

Final Paper >

Presentation >

Project in Psychology

THE EFFECT OF INTELLIGENCE ON SUSCEPTIBILITY TO COGNITIVE ILLUSIONS

TEAM

Sophia Abdul, Adelyn Davis, Grace Fine, Elias Kang, Yashasvi Kompella, Isabella McGowan, Aaryan Nagaria, Sharon Park, Anushka Patchigolla, Laksh Patel, Dhiya Selvakumar, Rickey Su, Brooke Valentine

Advisor: Patrick Dolan

Assistants: Natalie Bailey

ABSTRACT

Cognitive illusions reveal errors in how individuals perceive, remember, and judge information by showing how the brain uses prior knowledge or mental shortcuts instead of deductive reasoning. These illusions offer a means of investigating processes related to fluid intelligence, the ability to reason and solve new problems independent of prior knowledge. This study examines whether academically high-achieving scholars of the Governor’s School of New Jersey Program in the Sciences are less prone to cognitive illusions compared to online participants. The study hypothesized that the scholars are less susceptible to cognitive illusions in comparison to the online participants. Using a 15-task survey, participants were tested on their performance across various cognitive illusions. This survey was administered to 47 scholars and 185 online participants through Qualtrics, with the online participants being recruited through CloudResearch Connect. The data indicated that scholars were less susceptible to the illusions involving perception, attention, and judgment and decision-making. However, scholars were more susceptible to illusions that included memory and self-perception. These findings suggest that susceptibility to cognitive illusions varies depending on mental processes and the task presented. Overall, the data suggests that scholars were less vulnerable to cognitive illusions pertaining to analytical and deliberate reasoning, partially supporting the original hypothesis.

Final Paper >

Presentation >

Project in Computer Science

MULTIFACETED EVALUATION OF LARGE LANGUAGE MODELS’ POTENTIAL FOR SCIENTIFIC DISCOVERY

TEAM

Olivia Barsemian, Sophie Chen, XiaoQing Cui, Sahasra Datla, Daniel Egusquiza Bronstein, Sreethan Gangavarapu, Justin Lambert, John Loatman, Nathaniel Nadler, Shivan Shrinivas, Michael Xu, Adhithya Yuvaraj

Advisor: Dr. Minjoon Kouh

Assistant: John McLaughlin, Giorgi Tsnobiladze

ABSTRACT

Large language models (LLMs) have become increasingly popular in scientific research, yet their ability to autonomously perform each individual task involved in scientific discovery remains uncertain. We developed four proof-of-concept benchmarks that evaluated LLMs’ abilities in recognizing patterns in integer sequences, understanding whether given information is sufficient to make a conclusion, identifying human errors in lab reports, and deducing implications of alternative physical laws. We identify the disconnect between model accuracy and confidence and the consequences of model failure as key considerations when integrating LLMs into research processes. Ultimately, we conclude that the central concern of LLM benchmarking in a scientific discovery context is not determining whether models can serve as a scientist, but rather discerning which components of scientific reasoning can be performed reliably, the conditions under which they can be performed, and the areas in which failures occur.

Final Paper >

Presentation >

Project in Molecular Biology

Comparative Effects of Lipid Transfection Reagents on Transfection Efficiency and IL-1 Cytokine Induction in H1299 Cells

TEAM 

Kevin Aby, Armaan Balani, Zachary Caldera, Jake Caravello, Sophie Gang, Christine Johansen, Megan Kim, Kabir Kalsi, Stephanie Liao, Sam Rosen, Tianyi Iris Shen, Anya Verma

Advisor: Dr. Brianne Barker

Assistant: Brenna Hezel, Emily Shirk

ABSTRACT

Lipid-based transfection systems are widely used to deliver nucleic acids directly into cells, yet its effect on innate immune responses is still poorly understood. Host innate immune pathways recognize foreign biomolecules, often categorized as microbe-associated molecular patterns (MAMPs), via pattern recognition receptors (PRRs) and restrict transgene expression. While previous research has primarily focused on nucleic acid-mediated innate immune activation, emerging evidence suggests that lipid transfection carriers themselves may also stimulate innate immune signaling. However, the underlying mechanisms—how commercial lipid reagents differ in inducing inflammatory cytokine cascades independent of nucleic acid delivery—remain unclear. We evaluated whether different commercial transfection carriers trigger intrinsic IL-1 family cytokine responses independently of nucleic acid cargo, and whether this response varies across reagents. H1299 cells were treated with Lipofectamine 2000 and Polyethylenimine (PEI) in the presence and absence of plasmid DNA. Transfection efficiency was quantified via ampicillin-resistance (ampR) gene delivery, and immune activation was measured using qPCR for IL-1α and IL-1β. Both reagents induced significant cytokine expression even in the absence of DNA, showing that delivery vehicles activate innate immune pathways. However, Lipofectamine 2000 achieved a higher transfection efficiency while inducing a significantly weaker inflammatory response compared to PEI. When combined with DNA, PEI vastly upregulated IL-1α and IL-1ß. These findings show that the chemical composition of the delivery particle plays a significant role in determining cellular response. Therefore, understanding each reagents’ individual DNA transfection rate and the reagent-induced innate immune cytokine response cascade is critical for designing transfection reagents tailored for specific therapeutic applications, such as minimizing innate immune inflammation for gene therapies or using it to strengthen vaccine-induced immunity.

Final Paper >

Presentation >

Project in Geographic Information System (GIS)

MAPPING SOCIAL AND ENVIRONMENTAL HEALTH VULNERABILITY IN MADISON, NEW JERSEY

TEAM

Elias Chang, Rai Chowdhury, Zihan Chu, Shiven Das, Ethan Dougherty, Avaneesh Gore, Brianna Jimenez, Luke Nathan, David Pevzner, Evie Shen, Mia Teng

Advisor: Han Yan, Ph.D.

Assistant: Sofia D’Amelio

ABSTRACT 

An individual’s heat-related illness risk is largely dependent on their socioeconomic status and surrounding environment in which they live. Acting as a feedback mechanism, social demographics and environment are highly interconnected. Public health can be assessed by evaluating spatial relationships between social and environmental determinants. Within our research, we investigated heat-related public health risks across the four census tracts of the borough of Madison, New Jersey. A nested-scale temperature investigation was performed to assess the relationship between location, tree canopy coverage, and surface materials and temperature and perceived heat. We found that nearly all categories tested showed statistically significant associations. Additional spatial data were assessed by evaluating maps and aggregating social and heat data to create vulnerability and risk indexes. Significant disparities were discovered between Madison’s four census tracts. Notable findings were the high health vulnerability of Census Tract 429 and the low health vulnerability of Census Tract 431. Future studies can use our disparity findings to target the most effective ways of addressing and amending health risks.   

Final Paper >

Presentation >