SSP Forum: Senior Honors Students, Class of 2022

Monday, June 6, 2022
Building 460, Room 126 (for current Stanford affiliates)
or Zoom (open to all) - see link in announcement

symsys bubbles logo

UPDATE: Based on new guidance from the School of Humanities & Sciences concerning the recent spike in Covid cases among students, this is now a hybrid in-person (for Stanford students, faculty, and staff only) and Zoom (open to everyone) event. See revised info below.

The
Symbolic Systems Forum
presents

Annual Presentation of Honors Projects
featuring
Senior Honors Students from the Class of 2022
Symbolic Systems Program

Monday, June 6, 2022
12:15-1:25pm Note later ending time
Margaret Jacks Hall (Bldg. 460), Room 126
NOTE: Due to Covid-19 restrictions on building occupancy, only current Stanford students, faculty, and staff may attend this event live

or
Join from PC, Mac, Linux, iOS or Android: https://stanford.zoom.us/j/99713091022?pwd=b0hna3VRUkFWWHl0TUEwcTRQYktw…
    Password: 352122
(Zoom attendance is open to all)

A RECORDING OF THIS SESSION WILL BE MADE AVAILABLE AFTERWARD FOR STANFORD-AUTHENTICATED USERS ONLY

SCHEDULE:

12:15 Josh Nkoy, Comparing Functional Dependences on Interventionist Counterfactual Causal Models (Honors Advisor: Thomas Icard, Philosophy; Second Reader: Johan van Benthem, Philosophy)
Abstract: Formalizing causality into terms of (mathematical) logic can allow us to determine with greater precision (1) when it is appropriate to draw causal relationships, (2) whether we can express all the nuanced types of causality, and (3) how we can define the constituent parts of the machinery that drives causal reasoning, among other objectives. While work on (1) and (2) is plentiful and exemplified by Pearl’s structural causal model (SCM), the use of functional dependence logics to describe reasoning over SCMs is a relatively new tactic toward (3). Shifting focus of these efforts from the global dependence logic DL to the local dependence logic LFD, I argue that using the more local LFD to analyze interactions between dependence and causality is more fundamental, fine-grained, and easily controllable. However, I also show that localizing dependence in combined functional-counterfactual dependence logics still leaves various metaphysical problems for analyzing causality that lie outside the scope of said localization. Consequently, I then propose interventional analysis of sets of causal models, rather than singular models, might help solve some of these issues. All throughout, I will show how even in these localized settings, the stratification of causal inference tasks encapsulated in the Pearl Causal Hierarchy hold — and that, in general, logic still backs that correlation (almost) never implies causation.

12:25 Chris Iyer, Towards a Multiple Systems Understanding of Race-Related Biases in Human Memory (Honors Advisor: Anthony D. Wagner, Psychology; Tyler Bonnen, Psychology)
Abstract: Human memory is biased. In memory research, bias is often studied as a monolith, operationalized in a single behavioral measurement. However, memory behaviors are supported by a concert of interacting systems, and understanding our biases requires disentangling the unique contributions of each system and how each one is shaped by social dynamics around us. Here, we begin to isolate the contributions of perceptual and memory-specific bias to episodic memory retrieval. To do so, we implement a behavioral experiment and construct a novel stimulus set of face images capturing both naturalistic variation and low-level control. Participants performed a 3-way concurrent face discrimination task, and subsequently a 2-alternative-forced-choice recognition memory test. Moreover, we tested participants on both previously seen images and images of previously seen faces from unstudied viewpoints, to test for race-related bias in the generalizability of memory representations. to assay race-related bias in the format of memory representations. We find evidence for canonical ingroup perceptual biases but (reverse-canonical) outgroup-favoring memory bias, with preliminary evidence that viewing time at encoding may explain some of this memory bias. Furthermore, we find no evidence for bias in participants’ memory generalizability to decisions on novel viewpoints. This work offers a more fine-grained understanding of how social biases shape different systems of human cognition and models new approaches to studying race-related biases with attention to the limitations of psychology and neuroscience, the social ramifications of race and racism, and the complexity of human cognition and behavior.

12:35 Bobby Sparks, Communication as Influence - Examining Preschool-Aged Children’s Ability to Infer Knowledge from Influence (Honors Advisor: Hyowon Gweon, Psychology; Second Reader: Aaron Chuey, Psychology)
Abstract: How do we know what others know? Prior work has examined how children use evidence about isolated agents, like their perception and actions, to infer what they know. However, we often reason about others in social contexts where agents interact with and influence each other. In these contexts, both a speaker’s communication and the way it causes a listener to behave can together license epistemic inferences about the speaker. The present studies investigated how preschool-aged children use two pieces of evidence to reason about what speakers know: changes in the outcomes of a listener’s actions following communication (Study 1) and changes in a listener’s actions themselves following communication (Study 2). In both studies, children observed two scenarios where a listener failed to activate a toy before succeeding. In Study 1, children observed a speaker produce nonsense language towards a listener after they failed but before they succeeded to activate a toy, as well as another speaker who spoke to a listener prior to initial failure. In Study 2, children observed a speaker speak to a listener before a distinct change in action, followed by success, as well as another speaker who spoke to a listener resulting in no distinct change in action, followed by success. In both studies, when asked which speaker knows how to make the toy work, children preferred the speaker who appeared to cause the listener to succeed (Study 1) or change their action (Study 2). These results suggest that preschool-aged children are sensitive to the way that speakers influence others via communication and can use evidence of that influence to infer what speakers know. More broadly, these studies highlight children’s ability to reason about the knowledge of one agent (a speaker) based primarily on evidence about another agent (a listener).

12:45 Mia Bahr, A New Hyperbilirubinemia Treatment Pathway: Improving Accessibility, Quality, and Equity of Care (Honors Advisor: Richard Fan, Urology; Second Reader: Michael Higgins, Center for Biomedical Informatics Research)
Abstract:  Neonatal hyperbilirubinemia is one of the most prevalent infant conditions impacting approximately 60% of newborns worldwide. However, prior studies have shown high variation in care practices across the United States despite the condition's prevalence. NICU overuse in non-emergent cases is one of the most common and stressful treatments for mothers and infants. The present research investigates the feasibility of benefits for patients and providers from the inclusion of a novel transcutaneous bilirubinometer in treating infants with the condition. Through interviews with American mothers of jaundiced infants and decision analysis performed on neonatal providers affiliated with Stanford University, I sought to understand and analyze current maternal needs in hyperbilirubinemia treatment and assess provider tolerance for novel options of care. I propose a novel treatment technology and seek to answer two questions: (1) Could the proposed technology potentially reduce core concerns with hyperbilirubinemia treatment among American mothers? (2) Could introducing the proposed technology be a feasible alternative to traditional NICU care (Study 2)? I find favorable evidence for both and discuss future directions for research.

12:55 Isabel Wang, Caregiver and Child Health Under Lockdown: Impacts of the COVID-19 Pandemic on Caregiver Outcomes, Caregiving Behaviors, and Health Services Utilization in Rural China (Honors Advisor: Scott Rozelle, Freeman Spogli Institute for International Studies and Stanford Institute for Economic Policy Research; Second Reader: Gary L. Darmstadt, Pediatrics)
Abstract: The COVID-19 pandemic has exacerbated societal and health inequities globally. In rural China, in particular, disparities in caregiver and child health already existed and likely dramatically shifted, in part due to the uniquely strict lockdown mandates. Caregiver outcomes, caregiver behaviors, and perinatal health services utilization seemingly changed immediately for the many under-resourced communities in rural China. By studying the behaviors and outcomes of families with children born before lockdown and children born during lockdown, I identify the crucial impacts of the COVID-19 pandemic on communities in rural China. I find that parental migration significantly slowed during the pandemic and related lockdowns, which likely contributed to the significantly improved mental health outcomes in caregivers through improved social support. Caregiving behaviors generally remained the same, but perinatal health services utilization suffered, particularly in postnatal care. This research has implications in public health initiatives and global health policies to improve the quality of life for the most vulnerable populations. Contributing knowledge through academic research to the literature on caregiver and child health is essential to promoting equity and transforming lives.

1:05 Yasmeena Khan, College Students Aren't As Liberal As You Think: Evidence for Political Moderation in Homogenous Groups (Honors Advisor: Alice Siu, Center for Deliberative Democracy; Second Reader: Larry Diamond, Freeman Spogli Institute for International Studies and Hoover Institution)
Abstract: As America grows ever more polarized, a question arises: are productive discussions even possible anymore? Due to the rising phenomenon of partisan sorting, more and more Americans are living in enclaves that share not just demographic features but are politically homogeneous (Brown and Enos 2021). The widely-accepted universal law of group polarization states that homogeneous groups always become more extreme after discussion (Sunstein 2002). If this law really is universal, it poses a serious problem for the efficacy of political conversations. This thesis aims to disprove the universality of this law by considering a study, “Shaping Our Future,” from the Stanford Center for Deliberative Democracy, undertaken with a group of highly homogeneous college students (n = 541, 75% Democratic, 11.5% Republican). By examining opinion change using pre- and post-study surveys, this thesis demonstrates that this homogeneous group moderated on a majority of proposals (decreased in ideological polarization) as well as grew warmer towards the other side (decreased in affective polarization). T-tests are used to prove statistical significance. Overall, I find that this group, despite its makeup, did meaningfully reduce polarization through discussion. I therefore offer a potential counterexample to the universal law of group polarization.

1:15 Jonah Wu, The Incoherence of Stakeholderism (Honors Advisor: Colleen Honigsberg, Law; Second Reader: Curtis Milhaupt, Law)
Abstract: Stakeholderism is a theory of corporate governance which states that directors manage the corporation in the interest of all stakeholders of the firm. This theory of corporate governance has gained increasing popularity and traction and corporate legal scholarship has sought to justify it. However, these justifications for stakeholderism are neither descriptively consistent with Delaware law nor accurate in their theories of the firm, as stakeholderists claim. Enlightened shareholder value stresses the lack of tradeoffs or the ubiquity of win-win scenarios between shareholders and stakeholders, but this is empirically dubious when considering the effect of pro-stakeholder policies on firm value. Despite the lack of justification, the desire for corporations to serve some public benefit is not going anywhere. I propose that shareholder welfare maximization is a more theoretically coherent means to justify corporate concern for social goods and that the corporate ecosystem could evolve to chiefly include shareholder value firms seeking to maximize shareholders’ economic interests, and shareholder welfare firms seeking to maximize both shareholder’s economic and social interests as their ultimate end.