[Back to Research]
As a grad student, I worked as the teaching assistant for three classes at Princeton, and loved all of them.
I also ran a few workshops for undergraduate and fellow grad students on Matlab, and the MVPA toolbox. [I'll try and dig up my notes on these. TODO]
Introduction to connectionist models
Ken Norman‘s PSY330 (Spring 2006) class, ‘Introduction to connectionist models: bridging between brain and mind’.
This class contained grads and undergrads from Psychology, Molecular Biology, Aerospace Engineering, Economics and Linguistics, some of whom had never done any kind of programming, let alone computational modeling before. It was a huge amount of work – for me, for him, and for his students. But the end result was worth it. Every one of the students produced a detailed and original neural network model of some aspect of cognition that fascinated them, ranging from classical conditioning to music to mania to bilingualism.
Ken transported the class from the lowest-level subneuronal machinery to the highest-level dynamics of recurrent networks with thousands of interacting units, showing how computational models can unify evidence from various fields to explain complex, counterintuitive cognitive phenomena. I study this stuff, so I’m bound to get excited about it, but I really believe that in just a few months, the students who took that class grew significantly more powerful intuitions and tools for thinking about thinking.
From words to idioms to grammar
Psychology of language