Stanley Klein, Professor

Closed (1) Using electrocorticogram data to explore neural dynamics

Applications for Spring 2017 are now closed for this project.

The electrocorticogram (ECoG) is a method for measuring high resolution brain activity in awake humans. Involves multiple electrodes being placed on the brain before surgery to determine optimal surgical decisions. We recommend URAP applicant to google Electrocorticography to learn more. The late Walter Freeman left some ECoG data, taken straight from the brain during surgery. It allows exploration about how brain waves like alpha, gamma and so on affect neural firing. Students are invited to see how such cortical processes can interact with dynamics at the single neuron level to exemplify a polyrhythmic brain. We will also explore new methods for analyzing nonlinear processing that takes place between adjacent cortical regions.


After reading several articles on the topic, the undergraduate would collaborate with Sean O’Nuallain and Stanley Klein on developing models of neural processing that would enable neurons to carry out Fourier processing of stimuli.

Qualifications: The student would need to be familiar with Fourier processing and be able to program in Matlab. Prior knowledge of Matlab isn’t needed if the student is familiar with Python’s NumPy.

Weekly Hours: to be negotiated

Related website: http://cornea.berkeley.edu

Closed (2) Getting real neurons to do Fourier/Gabor transforms

Applications for Spring 2017 are now closed for this project.

General description of the project (the concept)
While the current neural net approaches of deep learning seem impressive, we believe along with many in the community that simply calling them “convolutional nets” is more accurate. They are not neural, assuming a linear model of the neuron on which they impose non-linearity in a randomly-chosen transfer function. We propose a neuron that is a harmonic oscillator. Similarly, their model parses photos from the web that already exploit Western laws of perspective. We propose that no such assumption is valid, and all that exists out there is a power spectrum.

In that we follow the great American scientists David Bohm and Karl Pribram whose metaphysics and epistemology of perception is explored at length in our Feb 2016 book. We attach a summary of this approach; the first task of the student is to implement the system described in the attached slides about the Bohm/Pribram model

It involves resonate and fire models of the neuron based on the harmonic oscillator which are available open source in:

http://freesourcecode.net/matlabprojects/65239/fitting-critically-damped-simple-harmonic-oscillator-in-matlab#.VmcQHG4U6pp

http://www.compadre.org/osp/items/detail.cfm?ID=8290


Qualifications: The undergraduate would collaborate with Stanley Klein and Sean O’Nuallain on developing models of neural processing that would enable neurons to carry out Fourier processing of stimuli. The student would need to be familiar with Fourier processing and be able to program in Matlab. Prior knowledge of Matlab isn’t needed if the student is familiar with Python’s NumPy.

Weekly Hours: to be negotiated

Related website: http://cornea.berkeley.edu