Name: Yunxuan Zheng

School of Psychology – Ph.D. Dissertation Proposal Meeting

Date: Wednesday, November 5th, 2025

Time: 10:00am

Zoom link: https://gatech.zoom.us/j/92736325662

Dissertation Chair/Advisor: 

Dobromir Rahnev, Ph.D. (Georgia Institute of Technology)

Dissertation Committee Members: 

Rick Thomas, Ph.D. (Georgia Institute of Technology)

Robert Wilson, Ph.D. (Georgia Institute of Technology)

Jorge Morales, Ph.D. (Northeastern University)
Brian Odegaard, Ph.D. (University of Florida)

Title: A common computational basis for perceptual decisions and metacognitive judgments

Abstract: Type-1 perceptual decision making refers to judging the identity of sensory stimuli, whereas Type-2 perceptual metacognition involves judging the accuracy of those decisions. A central question is whether these judgments are generated by the same cognitive system or by two distinct systems. This thesis addresses this issue through three experiments. In Experiment 1, we applied a computational model to data from a 2-alternative forced-choice and a discrimination task to separate sensory, decisional, and metacognitive noise. We confirmed decisional noise in Type-1 judgments beyond sensory noise, and identified additional metacognitive noise unique to Type-2 judgments. In Experiment 2, we manipulated expectation and reward structure in the Type-1 judgment to match its complexity with the Type-2 judgment. Under the conditions showing matched complexity, we found the noise magnitude in Type-1 and Type-2 judgments was also similar, suggesting that the extra metacognitive noise may be a product of task demands rather than a separate system.  In Experiment 3, subjects further provided Type-3 judgments (certainty about their Type-2 judgments). We found that combining 2-point Type-2 and Type-3 scales was equivalent to a single 4-point confidence rating, with no evidence of additional computational noise in Type-3 judgments—indicating Type-2 and Type-3 judgments likely arise from the same system. This result further questions a simple division between a single Type-1 system and a separate system for all higher-level evaluations. Together, these findings suggest that perceptual decision making and metacognition likely share a common computational basis and operate on the same underlying signals, challenging strong two-system accounts of cognitive architecture