The Sloan Center at UCSF represents a wonderful interdisciplinary integration between quantitative folk like myself and the ongoing experimental research programs in the Keck center. To that end, I am currently involved in a couple of different projects.
It is becoming clear that in the higher stations of the auditory system, such as the inferior colliculus and auditory cortex A1, the response of neurons to various acoustic stimuli is essentially nonlinear. Although much of the auditory field still characterizes responses in terms of linear systems quantities, such as spectral response and transfer functions, the time has come to start studying nonlinear response functions of units in the auditory system. As a start, Christoph Schreiner and I are looking at the modulation response of neurons in the IC. The linear response of units to modulations is captured in the modulation transfer function (MTF). We created a novel form of stimulus, the perturbative m-sequence, to investigate the nonlinearities of the MTF itself. We found that the MTFs are indeed nonlinear, so that two modulation functions of the same carrier wave do not obey superposition in the response.
You can download the preprint here.
There is a basic problem in recording extracellularly from single neurons in the cortex: there are many neurons in the vicinity! Thus, a voltage trace may show action potential spikes from not just one cell, but from many cells; furthermore, these spikes may overlap. When the spikes do not overlap (as may happen in more quiet areas of the brain), it is possible to separate the spikes from each cell according to shape and size. Methods to classify spikes include template matching and adaptive clustering techniques. When the spikes overlap, however, these methods break down and a more sophisticated analysis is called for. I am working with Ken Miller's group to sort spikes using 1) Bayesian analysis for estimating overlap decompositions of superposed spikes, and 2) multi-electrode probes. Bayesian analysis for estimating overlap decompositions was pioneered by Michael Lewicki, and I am adapting and extending some of his code to handle multi-electrode inputs.
In the process of creating the sorter application, I created specPerl, a GUI builder for Tkperl.
Each lab in the Keck Center uses a different format for storing neural data and associated experimental conditions, which prevents people from using the best tools for the job. A group of us decided to remedy this situation by creating a common file format for all the labs to use. We created a detailed specification (which I co-authored) and are now in the process of creating and adapting tools to use SPIFF. These tools will be useful in the collaborations engendered in a upcoming Program Project grant.
Check out my Personal information and links .
Office: Health Sciences East room HSE-806A