My name is Trent N. Cash and I am a Postdoctoral Scholar in the School of Psychology at the University of Waterloo in Ontario, Canada. I earned a joint Ph.D. from the Department of Psychology and the Department of Social and Decision Sciences at Carnegie Mellon University. My research largely focuses on the role of higher-order reasoning - particularly metacognition - in judgment, decision making, and learning.
In one line of research, I focus on the role of metacognition in value-based, multi-attribute choice decisions, such as choosing colleges, buying houses, or selecting romantic partners. I am particularly interested in understanding the degree to which decision makers have metacognitive knowledge of the factors that influence these decisions and exploring contexts that may promote or inhibit this knowledge. I primarily assess metacognitive knowledge using a novel paradigm I created, called the Knowledge of Weights (KoW) paradigm. This line of research was the focus of my doctoral program and was funded by a Doctoral Dissertation Research Improvement Grant (#2333553) from the National Science Foundation. During my postdoctoral career, I am studying how AI can be used to improve metacognitive knowledge in real-world contexts, such as healthcare. This research is funded by the Lupina Foundation for Health, Society, and Technology.
In a second line of research, I study the metacognitive capacities of large language model chatbots (LLMs), such as ChatGPT. To do so, I ask LLMs to complete experimental tasks from the metacognition literature and compare their performance to that of humans. For example, I recently published a paper demonstrating that LLMs are capable of making confidence judgments that are about as well-calibrated as those made by humans. I also have a project under review comparing the performance of LLMs on the KoW paradigm to that of humans and have published a theoretical piece arguing that LLMs lack access to fluency-based cues that support human metacognition.
In a third line of research, I study how interactions with AI affect human cognition and metacognition. For example, I recently published a study demonstrating that LLMs can help students learn how to write argumentative essays by suggesting revisions and providing feedback. Better yet, the students enjoyed learning with the LLMs and felt a greater sense of self-efficacy for using LLMs in their future work. In another paper, I reviewed existing evidence about the impact that offloading to AI will have on our cognitive abilities. I argued that while our skills (e.g., flying a plane) may be at risk, our basic cognitive abilities (e.g., working memory) are likely to be more resilient. I also argue for the importance of staying in the cognitive loop.
In a final line of research, I study student learning and development. In my first-ever academic paper, I found that gifted students in pull-out classrooms reported better psychological well-being than gifted students in self-contained classrooms. In a more recent paper, I shared data from a 4-year longitudinal study (data collected from 2019-2022) demonstrating that gifted students were no more resilient, and in some cases more vulnerable, to the negative effects of the COVID-19 pandemic than non-identified students.
In one line of research, I focus on the role of metacognition in value-based, multi-attribute choice decisions, such as choosing colleges, buying houses, or selecting romantic partners. I am particularly interested in understanding the degree to which decision makers have metacognitive knowledge of the factors that influence these decisions and exploring contexts that may promote or inhibit this knowledge. I primarily assess metacognitive knowledge using a novel paradigm I created, called the Knowledge of Weights (KoW) paradigm. This line of research was the focus of my doctoral program and was funded by a Doctoral Dissertation Research Improvement Grant (#2333553) from the National Science Foundation. During my postdoctoral career, I am studying how AI can be used to improve metacognitive knowledge in real-world contexts, such as healthcare. This research is funded by the Lupina Foundation for Health, Society, and Technology.
In a second line of research, I study the metacognitive capacities of large language model chatbots (LLMs), such as ChatGPT. To do so, I ask LLMs to complete experimental tasks from the metacognition literature and compare their performance to that of humans. For example, I recently published a paper demonstrating that LLMs are capable of making confidence judgments that are about as well-calibrated as those made by humans. I also have a project under review comparing the performance of LLMs on the KoW paradigm to that of humans and have published a theoretical piece arguing that LLMs lack access to fluency-based cues that support human metacognition.
In a third line of research, I study how interactions with AI affect human cognition and metacognition. For example, I recently published a study demonstrating that LLMs can help students learn how to write argumentative essays by suggesting revisions and providing feedback. Better yet, the students enjoyed learning with the LLMs and felt a greater sense of self-efficacy for using LLMs in their future work. In another paper, I reviewed existing evidence about the impact that offloading to AI will have on our cognitive abilities. I argued that while our skills (e.g., flying a plane) may be at risk, our basic cognitive abilities (e.g., working memory) are likely to be more resilient. I also argue for the importance of staying in the cognitive loop.
In a final line of research, I study student learning and development. In my first-ever academic paper, I found that gifted students in pull-out classrooms reported better psychological well-being than gifted students in self-contained classrooms. In a more recent paper, I shared data from a 4-year longitudinal study (data collected from 2019-2022) demonstrating that gifted students were no more resilient, and in some cases more vulnerable, to the negative effects of the COVID-19 pandemic than non-identified students.
To learn more about the research I am working on right now - including some projects that were not described here - check out my Publications.