Senior Project 2024

Author: Kaiki Chiu
Title: The Role of Adaptive Cognition in Feedback and Judgment Accuracy
Advisor: Professor John Wilcox

Abstract:
Learning from past mistakes through feedback is an intuitive strategy for improving future performance. However, research indicates that individuals vary in their responsiveness to feedback interventions. This study aimed to investigate the general effects of feedback on subsequent performance outcomes and explore the role of adaptive cognition traits, such as growth mindset, active open-mindedness, and infallibility mindset, in moderating these effects. A survey (n=92) was conducted to examine the impact of feedback on calibration. Results revealed that feedback generally led to improvements in calibration by reducing overconfidence, and no significant moderation effect was detected for the role of adaptive cognition traits on post-feedback calibration improvement. However, the findings do not conclusively support or refute the existence of these potential moderation effects, as the study might not have had sufficient statistical power to detect them. Despite the lack of significant results, the direction of the relationship between adaptive cognition variables and calibration aligns with the hypotheses, indicating a potential effect. To this extent, future research endeavors should consider employing larger sample sizes to enhance statistical power and investigate potential moderators of feedback effectiveness in more detail.

Author: Stella Dong
Title: The Value of What’s To Come: Information-Seeking Behavior Around Aversive vs. Reward Outcome Anticipation
Advisor: Professor Kiyohito Iigaya

Abstract:
Alterations in anticipation, in both positive and negative conditions, are prevalent among individuals with psychiatric disorders. Anticipation can be formalized through an anticipatory utility computation derived from a behavioral economic model, where the total experience of a reward is the sum of the consumption of that reward plus anticipating that reward coming up in the near future (Loewenstein, 1987).

While previous studies have found that advanced information of rewarding outcomes boosts the overall reward that the experience offers (Iigaya et al., 2016), this effect of information is not clear in the negative domain. i.e. when the outcome is aversive. Therefore, this study tries to investigate the information-seeking behavior in an aversive condition and how this behavior varies across frames.

Many studies have reported that the sympathetic system supports the stress response to aversive events, and the mobilization of the body toward appetitive or rewarding stimuli. Therefore, we included physiological measures in the study to uncover physiological signs reflecting dread and other mechanisms during the anticipation period.

Our findings reveal information-seeking behavior in aversive conditions to be complex and highly variable, which can be influenced by individual framings and personality traits, highlighting the multifaceted nature of anticipatory experiences and validating the impact of information on these experiences through behavioral and physiological measures such as pupil size and heart rate.

Author: Abigail Ganz
Title: Are We Addicted to Our Playlists? The Possibility of and Potential Problems with Reliance on Music in the Background of Our Lives
Advisor: Professor Mariusz S. Kozak

Abstract:
With the advent of recorded sound and personal listening technologies, we can easily play music as a backing track to our entire lives: during our commutes and working hours, our leisure time, and even while we sleep. This study was interested in exploring whether or not our modern listening habits have created a reliance on music for navigating our daily lives, and if so, whether this reliance may have negative impacts. Semi-structured interviews were conducted with participants from both Generation X and Generation Z in order to explore modern listening habits and identify any generational differences in these habits. The interviews suggest that while listening to music as a backing track to our lives can have many positive effects across generations, this habit is not as innocuous as it may initially seem and music reliance is commonplace. However, despite the potential negative consequences of music reliance, it appears that this reliance is not accurately defined as an addiction and that music in general can bring an immense amount of positivity to people’s lives. Therefore, the study suggests that one should focus on engaging with music responsibly and thinking critically in order to develop a healthy relationship with the music in their life. 

Author: María del Valle García López
Title: Musical Primitive: Refining the Account of Auditory Perceptual Organization through Assays of Perceptual Bistability in Sine-Wave Speech
Advisor: Professor Robert Remez

Abstract:
This senior thesis aims to refine accounts of auditory perceptual organization by examining instances of concurrent perceptual bistability. Perceptual bistability refers to an instance in which a single sensory sample is included in two sensory contours, leading to an experience of two distinct perceptual impressions. Applying a technique of frequency forcing to one of the tone analogs in a synthetic sine-wave sentence led that acoustic element to evoke a phonetic impression simultaneously with a melodic impression. This is achieved by shifting the sine-wave's frequency to align with the closest musical note while preserving the acoustic structure crucial for the perception of consonants and vowels. As such, a musicalized sine-wave sentence elicits concurrent experiences of musical and phonetic organization. This phenomenon exposes limitations in the prevalent account of perceptual organization, Auditory Scene Analysis (Bregman, 1990), which relies on similarity-based Gestalt principles to explain the formation of sensory contours in primary perception. This approach fails to explain how inherently dissimilar acoustic elements can be perceptually grouped into a coherent sensory contour. Additionally, concurrent bistability is unique to audition without a visual equivalent, indicating that current auditory models derived from vision are inadequate. The perceptual measures of this phenomenon calibrate the resources supporting the perception of speech and music, moving towards a more comprehensive causal framework.

Author: Anna Li
Title: Memory and Misinformation: Examining Accessibility and Perception of Truth
Advisor: Professor Janet Metcalfe

Abstract:
This study aimed to challenge the position that people determine the truth of their information based on accessibility. We investigated how participants recalled true and false feedback and how they ascribed truth to it. Participants answered general information questions and were randomly assigned true or false feedback, and were later asked to either recall the information or determine the true information. We considered recall ability as accessibility, and we compared recall ability with truth determinations. Participants demonstrated different recall abilities and truth judgments based on the correctness of their original answers. Participants who were initially correct showed a higher ability to recall true feedback compared to false feedback. When participants were originally wrong, there was no significant difference between the rates in which they recalled and ascribed truth to true and false information. These findings provide further nuance to the accessibility position and lay foundations for areas of future research.

Author: Sasha Newman-Oktan
Title: Shifting Frames, Sharpening Decisions: Investigating the Impact of Metacognitive Framing on Information Seeking Behavior and Decision-Making Accuracy Under Uncertainty
Advisor: Professor Jacqueline Gottlieb

Abstract:
In situations where knowledge is limited, it is beneficial for an individual to seek additional information to reduce their uncertainty and gain a more precise understanding of the world (Schulz et al., 2020). We must often demand-information without trial and error learning or the luxury of time to inform our decisions, but how do individuals decide which specific sources of information to pay attention to, to reduce their uncertainty when multiple options are ambiguous (Gottlieb, 2023)? For example, before running out of the house, I will quickly decide to check the weather forecast. For that, I must imagine what I would learn if I were to check the forecast and how certain (or uncertain) I may become relative to right now. This kind of judgment of my uncertainty, both present and future, is called metacognition (Metcalfe, 2009). Our study looks at how this type of judgment affects choices of study in a strategic student situation in which participants prepare for a test with two problems, but only have time to “study” for one. Our experiment investigates how individuals navigate this decision, particularly when confronted with limited knowledge and uncertain options, specifically with respect to their information-seeking behaviors. We also investigate how their metacognitive judgments shape information-seeking and subsequent decision-making strategies. The experiment includes a metacognitive learning intervention framed either positively or negatively to prompt participants to reflect on their approach to task-solving investigation. We analyze participants' choices and their information-seeking preferences between two ambiguous options, considering their interaction with the intervention.

Our findings reveal shift trends in decision-making and information-seeking behavior across both framing conditions. In both the positive and negative frame conditions, there was a pronounced increase in Optimal Informative Choices, indicative of heightened decision-making efficacy. The Negative Frame (NF) condition elicited substantial changes in participants' strategies, particularly in their sensitivity to Probability Gain (PG). Participants exposed to the NF frame exhibited a greater inclination towards adopting PG-insensitive strategies and demonstrated a decreased likelihood of switching from PG-sensitive to PG-insensitive strategies. Notably, the metacognitive learning intervention appears to trend toward being more effective for nudging participants with adequate understanding, but not participants who complete the first block of tasks with a sub-optimal strategy. This work is a foundational step toward understanding the cognitive mechanisms underlying orientation of attention and expected information gain (EIG). It also addresses how metacognitive framing manipulations provide insights into information-seeking processes. These findings contribute to a deeper understanding of how contextual cues influence information-seeking behavior and have implications for learning as well as optimizing decision-making interventions in real-world settings, particularly in education.

Author: Chukwuma Orji
Title: Do I Understand My Chatbot? Investigating Linguistic Stance in Human-Chatbot Conversations
Advisor: Professor William Foley

Abstract:
In recent years, the usage of conversational artificial intelligence (AI) applications has significantly increased. Thus, the need for an understanding of how users communicate with them has become paramount, to maximize their utility. The concept of stancetaking has been studied in human-chatbot conversations, but there is little work done to understand it broadly. This thesis investigates stancetaking strategies and the relevant conversation mechanisms in human-chatbot conversations, as well as the influence of social characteristics on these communicative strategies. This study synthesizes existing literature on conversation analysis to propose a model of stancetaking features, containing both stancetaking strategies and conversation mechanisms, categorized into three domains of conversation. It then investigates the social characteristics which influence these domains, based on a model from previous research. Through a comprehensive review and analysis, the thesis discusses the interplay between conversation aspects, linguistic stancetaking, and the anthropomorphic qualities of CAs. By bridging these fields of study, this study presents findings which can be used to inform the development of conversational agents and enhance user experience in human-CA conversations.

Author: Yige Sun
Title: The Ontogeny of Disgust in Early Childhood
Advisor: Professor John Glendinning

Abstract:
The most recent review on evolutionary theory and the development of disgust is from Joshua Rottman et al in The Development of Disgust, 2019, claiming that disgust onset does not fully emerge until after the immediate post-weaning period, suggesting a later developmental onset around five years of age, which challenges the theory that disgust evolved as a defense mechanism acting as the behavioral immune system. This claim that disgust’s development is significantly influenced by cultural and social learning highlights its flexibility and the significant role of environmental factors in shaping disgust responses. This thesis aims to present an updated review of the ontogeny of the disgust response from infancy to early childhood to emphasize the importance of the role of culture and environment and examine how the development of disgust responses contributes to the rich cultural diversity and malleability of the human diet.

Author: Esther Thielking
Title: Alpha and Beta Oscillations Are Linked to Neuronal Excitability Independently of Oculomotor Activity During Free-Viewing
Advisor: Professor Luca Iemi

Abstract:
A prominent hypothesis in cognitive neuroscience suggests that neural oscillations regulate the excitability state of the neural system, thereby influencing cognition and behavior through a process known as functional inhibition. However, support for this hypothesis comes primarily from experiments that analyze brain activity during fixation periods, wherein participants avoid eye movements. It remains unknown whether functional inhibition persists in real-world conditions where neural activity is also shaped by oculomotor activity. To address this, we analyzed the relationship between oscillatory power and neuronal excitability during a free-viewing paradigm that promoted oculomotor activity. Using intracranial electroencephalography (iEEG, n = 2652), we estimated fluctuations of oscillatory power in the alpha (7-14 Hz) and beta (14-30 Hz) bands, as well as broadband high-frequency activity (BHA, 70-150 Hz) to index neuronal excitability. Simultaneously, we recorded changes in saccade rate and pupil size using eye tracking and pupillometry to assess oculomotor activity. Our results showed that excitability was suppressed during moments of strong oscillatory power in both alpha and beta bands, consistent with functional inhibition. Moreover, we identified fluctuations of neural activity that were both dependent on and independent of oculomotor activity during free-viewing. Importantly, we demonstrated that the hallmark of functional inhibition - the negative relationship between oscillatory power and excitability - persists even when considering fluctuations in neural activity that were dependent and independent of oculomotor activity. Taken together, these findings confirm and extend the functional inhibition account of alpha and beta oscillations during natural viewing conditions, motivating future investigations in more ecological settings.

Author: Cairo Yépez
Title: Neural Compression and Deindividuation Within Race-Based Memory Task
Advisor: Professor Jon Freeman

Abstract:
This project investigates the cognitive and neural mechanisms underlying the brain’s tendency to perceive outgroup members as less individuated compared to ingroup members - a phenomenon known as outgroup deindividuation. Using a combination of neuroimaging and behavioral tests, we search for neural correlates of categorical processing versus individuated processing. With a behavioral experimental paradigm to test memory for people of different races, we find evidence replicating prior work that participants are more likely to mistake individuals of the same race category rather than individuals of different race categories. Furthermore, with functional magnetic resonance imaging (fMRI) data collected during the task, we present evidence supporting our hypothesis that neural representations in correct trials are higher dimensional than either between-category or within-category errors in regions that were significantly activated during correct memory retrievals. Furthermore, in regions significantly activated during within-category errors, no statistically significant difference was found in neural compression in either within-category errors, between-category errors, or correct responses. We investigate the biases involved in deindividuation as part of a greater effort to understand the cognitive and neural mechanisms underlying individuation and categorization.

Author: Hannah Tongxin Zeng
Title: A Novel Board Game Paradigm for Studying the Acquisition of Overlapping Schemas
Advisor: Professor Christopher A. Baldassano

Abstract:
How does the brain acquire multiple overlapping schemas, the structured bodies of knowledge supporting our categorization and anticipation about what will come? We investigated this question using a novel board game paradigm where participants learned two games, Four-in-a-row and Knobby, with shared board configurations but differing win conditions requiring distinct optimal strategies. We conducted analysis on both game-level and move-level to compare effectiveness of schema acquisition in blocked and interleaved learning conditions (N=26). Several key findings emerged: 1) Blocked learning facilitated better overall performance, suggesting minimal interference between schema. 2) A primacy effect, where the first schema learned gained a substantial and persistent advantage. 3) Blocked learning renders better separation between schema, which is reflected as better moves to the specific game. Additionally, the study highlights the utility of game-based learning environments in cognitive research, offering a dynamic and engaging platform that closely mirrors real-world learning scenarios. This research contributes to our understanding of how different learning strategies affect the retention and application of new knowledge, particularly in settings where multiple schemas are involved.

Author: Ioanna Zervaki
Title: Awareness of Imprecision in Number Averaging Tasks
Advisor: Professor Michael Woodford

Abstract:
Everyday decision-making requires the rapid assessment and averaging of various criteria. An increasing body of literature uses number averaging tasks to investigate human information integration Spitzer, Waschke, and Summerfield (2017), Clarenau et al. (2022). This paper's contribution to number averaging tasks is two fold. Firstly, it isolates the influence of context on decision accuracy by standardizing the numerical representations and varying only the contextual pairings. Secondly, it distinguishes the effects of bias and variability on human errors picked up by number averaging tasks by eliciting a measure of subjective uncertainty through a confidence scale. We find compelling evidence context effects persist even when presented with the same numerical representations and that confidence is a robust predictor of accuracy. This paper highlights how cognitive noise affects tasks requiring sequential value integration such as stock selection on fluctuating returns or legal judgments assessing a defendant's. It also refines Bayesian models that assume subjective awareness of decision quality.