-
Technology Overview
Our work relies on a combination of technologies, including large scale multielectrode recording, optogenetic and electrical manipulations, advanced analytical techniques and computational modeling to understand the neural bases of learning and memory
Multielectrode recording and real time feedback
We believe that understanding information processing in neural circuits requires interacting with the circuit on the millisecond and tens of milliseconds timescales relevant for neural computations. We have therefore worked in collaboration with Jim MacArthur of the Harvard University Instrument Design Laboratory to develop the NSpike data acquistion system. This system, described in detail at nspike.sourceforge.net, includes both custom designed, DSP based hardware that permits extremely flexible data acquisition and real-time feedback as well as linux based open source software package. This system is also in use in other labs, and anyone interested in the hardware is welcome to contact Dr. Frank at loren _at_ phy.ucsf.edu.
Optogenetic and electrical manipulations
While multielectrode recording provides important insights into the properties of neural circuits, targeted interventions are required to determine which patterns of neural activity or plasticity underlie the ability to learn and remember. We are using electrical stimulation to activate or inactivate parts of the hippocampal circuit, but electrical stimulation has a number of important limitations. In particular, electrical stimulation approaches make it very difficult to target specific cell types or to induce physiologically realistic patterns of activity in the hippocampus.
With that in mind we have developed a multielectrode - fiber optic microdrive array that allows us to deliver light to specific parts of the brain through a movable optical fiber. In collaboration with Dr. Karl Deisseroth and colleagues, we have used viral approaches to express optically sensative channels in specific neurons within the hippocampus. Given the continued rapid advances in viral and optogenetic technologies, these techniques offer the potential for an unprecendented level of control over neural circuits in awake, behaving animals. We are therefore incorporating these approaches into many of our recording studies.
Expression of ChannelRhodopsin limited to dentate gyrus granule cells and their mossy fiber projections to CA3.
A number of people have asked us for information on the equipment we are using, and that information can be found here.
Analytical techniques
Large scale multielectrode recording in awake, behaving animals produces tremendously rich and complex data sets, and we use a variety of approaches to understand these data. As these data come from experiments where animals learn, these approaches have to take into account the fact that learning involves complex changes in the relationships among external stimuli, neural activity, and behavior. These complexities make it difficult to apply standard appraoches which require averaging across multiple trials or multiple neurons to produce accurate results. Unfortunately, averaging makes it impossible to identify the fast changes in neural representations that are thought to occur during learning.
To overcome these difficultiles we have developed adaptive estimation algorithms that allow us to describe the changing relationships between neural firing rates and a set of other variables, including, but not limited to, the animal's position in space and the temporal structure of the neuron's spike train. These algorithms do not require binning over time or space. Instead, they combine information about previous activity with new information to produce accurate instantaneous estimates of the underlying neural representation.
Example video showing adaptive estimation of the development of
phase precession in a CA1 pyramidal cell. Complete Caption.
Computational models
Our main focus is on data collection and analysis, as we feel that creating accurate models that explain how the hippocampus interacts with the cortex to support learning and memory requires a much deeper understanding of the relevant neural circuits than is currently available. Nonetheless, we are actively developing simple models to help explain the transformation of spatial information across the hippocampal circuit. As the data become available we will expand these models to help us get a handle on the complex neural phenomena we are studying.