Weakening memories by half-remembering them
For my PhD thesis, I worked on a series of behavioral and fMRI memory experiments to understand a little more about forgetting, called ‘Weakening memories by half-remembering them’.
fMRI and multi-voxel pattern analysis (MVPA)
I was heavily involved in efforts to analyze fMRI neuroimaging data multivariately. The basic idea is that we should be able to use machine learning tools to learn how the patterns in your brain activity indicate what you’re thinking.
I led development on the Princeton MVPA toolbox for Matlab, an open source package to facilitate these kinds of analyses.
- our TiCS review for more information on this kind of analysis
- my software page for more info on the toolbox
- Brain orchestras and fMRI analyses – for a more figurative explanation of this kind of analysis
With Ken Norman and Ehren Newman, I worked on a neural network learning algorithm that works very well at learning to pick apart similar patterns from one another, and captures a series of fiendish and counter-intuitive behavioural findings that I helped simulate.
This was the inspiration for a lot of the PhD work on ‘weakening memories by half-remembering them’.
Pittsburgh EBC fMRI analysis competition
Similarity structure and spatial working memory
As part of my master’s thesis, I worked (with Ken Norman) on a novel multivariate method for exploratory fMRI analysis. It calculates the isomorphism between the pattern of activity in a brain region and the patterns predicted by rich psychological models. In my thesis, I show that we can better predict which location someone is covertly attending to with this multivariate ‘similarity structure’ method than with standard or univariate measures on a spatial working memory dataset. I also attempt to relate this to my work on temporal context and memory.
Context and free recall
With Danny Oppenheimer and Anouk Schneider, I won the Society for Judgment and Decision Making’s Hillel Einhorn Award for the best paper for a young investigator, for our work on VAMP, a Voting Agent Model of Preferences.
With Agatha Lenartowicz and others, I worked on using multi-voxel pattern analyses of fMRI to investigate theories of task-switching.