CNC Presents: Andrea Stocco, PhD (UW Psychology)

“Computational Phenotyping of Forgetting”

Abstract:

Forgetting is among the most salient aspects of long-term memory, yet it is not directly observable, posing a fundamental challenge for theory. In this talk, I argue that forgetting is best captured as a latent computational process that can be inferred through computational phenotyping—the use of formal models of cognition to estimate theoretically meaningful parameters from behavior. I will present a series of experiments showing that parameters governing forgetting can be reliably identified at the individual level, remain stable across time, and provide diagnostic leverage beyond surface performance measures.

I will then introduce preliminary EEG and fMRI evidence linking these computational phenotypes to neural dynamics, providing converging constraints on the underlying mechanisms. Together, these results inform a long-standing theoretical debate in memory research: whether forgetting reflects the decay of memory representations or a progressive failure of access to otherwise intact representations. More broadly, this work illustrates how computational phenotyping can serve as a bridge between cognitive theory,  neural data, and clinical applications, enabling stronger tests of mechanistic accounts of memory.