Anatomical organization of the hippocampus

The hippocampal formation has a unique anatomical organization in that the connectivity between adjacent hippocampal regions is almost exclusively unidirectional. The majority of neocortical input to the hippocampus comes in through the superficial layers of the entorhinal cortex and connections proceed through the dentate gyrus, to CA3 and on to CA1 (the hippocampus proper), and then to the subiculum. Nearly all neocortically bound outputs from the hippocampus originate in CA1 and the subiculum and target cells in the deep layers of the entorhinal cortex, which projects both to numerous neocortical regions as well as to back to the superficial layers of the entorhinal cortex. Our research uses that organization to compare patterns of activity across regions and to use the similarities and differences among the patterns to identify the transformations that occur in the hippocampal circuit.

An animal model for hippocampal function

Numerous researchers have shown that a human without a hippocampus is unable to form new memories of facts or events. In rodents these same structures play an essential role in animal's abilities to learn about and remember complex associations, including tasks where the animal must learn and remember information about a set of spatial cues in order to navigate through an environment. Event/fact memory in humans and spatial memory in rodents both require learning complex relationships, and that parallel strongly suggests that qualitatively similar processing occurs in the human and the rat hippocampus.

Previous studies have shown that neurons throughout the hippocampal formation show place specific firing patterns, where a given neuron is active only in a subregion of the animal's environment. While neurons throughout these areas show place specificity, the properties of the place code differs from regions to region. We have shown, for example, that neurons in the CA1 subregion tend to be active in small subregions of the environment, while downstream in the deep layers of the entorhinal cortex, neurons tend to be active across long, contiguous regions, suggesting that these cells may be involved in representing extended trajectories. In addition, deep entorhinal neurons appear to code for task related information, in that individual neurons are frequently in multiple, behaviorally related locations.

Learning in the hippocampus and cortex

Most previous focused on describing patterns of activity during well learned tasks, and we therefore know little about neural processing during learning. One prominent hypothesis states that learning takes place first in the hippocampus and over time information is transferred to neocortical regions in a process known as consolidation. We are currently examining the development of spatial and task related patterns of activity in CA1 and the entorhinal cortex to determine whether representations form at different rates in these structures.

That experiment, and related work currently underway, focuses on understanding the relationship between learning and changes in neural firing patterns in the hippocampal formation and in related cortical regions. Over the longer term we plan to examine learning related changes activity across a wide variety of regions in an effort to understand the interactions within the hippocampus and between the hippocampus and the cortex during learning and memory retrieval.

Our approach

Our investigations rely on the combination of spatial tasks, large scale multielectrode recordings from awake, behaving animals, advanced analytical techniques, and computational modeling. We utilize custom large scale (128 channel) real-time Linux recording systems that allow us to record both cellular activity and local field potentials (also known as EEG) from up to 32 independently moveable four wire electrodes (tetrodes). These tetrodes are generally targeted at two or more structures, allowing us to simultaneously monitor patterns of activity across multiple brain regions. We record activity while animals perform tasks that require them to learn about and remember the spatial relationships among various locations and the behavioral significance of those locations.

Once the data are collected, we analyze them using both standard techniques as well as advanced algorithms developed in collaboration with Dr. Emery Brown of Massachusetts General Hospital and the Harvard / M.I.T. Health Sciences and Technology Program. Learning is thought to involve complex changes in the relationships among external stimuli, neural activity, and behavior. Studying the neural correlates learning is difficult, however, as the standard analyses 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.

We have therefore 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. We are also working with Dr. Brown and colleagues to develop new algorithms and models that will allow us to represent the dynamics of populations of simultaneously recorded neurons and the relationship between those dynamics and behavior.

Finally, once sufficient data have been collected, it becomes possible to generate computational models of the system. Our current focus is on the data collection and analysis, as there are many fundamental issues that have not been addressed experimentally, but over the longer term we are interested in creating accurate and well constrained models.

Publications

Cheng S, Frank LM (2008) New experiences enhance coordinated neural activity in the hippocampus. Neuron. Jan 24;57(2):303-13.

Frank LM, Brown EN, Stanley GB. (2006) Hippocampal and Cortical Place Cell Plasticity: Implications for Episodic Memory. Hippocampus. 16(9):775-84.

Frank LM, Stanley GB, Brown EN. (2004) Hippocampal plasticity across multiple days of exposure to novel environments. Journal of Neuroscience, Sep 1;24(35):7681-9.

Eden UT, Frank LM, Barbieri R, Solo V, Brown EN (2004) Dynamic analysis of neural coding by point process adaptive filtering. Neural Computation, 16:971-998.

Barbieri R, Frank LM, Nguyen DP, Quirk MC, Solo V, Wilson MA, Brown EN (2004) Dynamic analyses of information encoding in neural ensembles. Neural Computation, 16: 277-307.

Smith AC, Frank LM, Wirth S, Yanike M, Hu D, Kubota Y, Graybiel AM, Suzuki WA, Brown EN (2004) Dynamic analysis of learning in behavioral experiments. Journal of Neuroscience, 24: 447-461.

Nathe AR, Frank LM (2003) Making space for rats: from synapse to place code. Neuron, 39: 730-731.

Frank LM, Brown EN (2003) Persistent activity and memory in the entorhinal cortex. Trends in Neurosciences, 26: 400-401.

Wirth S, Yanike M, Frank LM, Smith AC, Brown EN, Suzuki WA (2003) Single neurons in the monkey hippocampus and learning of new associations. Science, 300: 1578-1581.

Nguyen DP, Frank LM, Brown EN (2003) An application of reversible-jump Markov chain Monte Carlo to spike classification of multi-unit extracellular recordings. Network, 14: 61-82.

Frank LM, Eden, UT, Wilson, MA, Brown, EN. (2002) Contrasting patterns of receptive field plasticity in the hippocampus and the entorhinal cortex: an adaptive filtering approach. Journal of Neuroscience. May 1, 22(9).

Frank LM, Brown, EN, Wilson, MA. (2002) Entorhinal place cells: Trajectory encoding. In The Neural Basis of Navigation: Evidence from Single Cell Recording. (P. Sharp Ed.) Kluwer Press. pp. 97-116.

Brown EN, Barbieri R, Ventura V, Kass R, Frank LM. (2002) A note on the time-rescaling theorem and its implications for neural data analysis. Neural Computation. 14(2):325-46.

Brown EN, Nguyen DP, Frank LM, Wilson MA, Solo V (2001) An analysis of neural receptive field dynamics by point process adaptive filtering. Proceedings of the National Academy of Sciences, 98(21): 12261-12266.

Frank LM, Brown EN, and Wilson MA (2001) A comparison of the firing properties of putative excitatory and inhibitory neurons from CA1 and the entorhinal cortex. Journal of Neurophysiology, 86(4): 2029-2049.

Barbieri R, Frank LM, Quirk MC, Wilson MA, Brown EN (2001) Diagnostic methods for statistical models of place cell spiking activity. Neurocomputing, 38 (4):1087-1093.

Barbieri, R, Quirk MC, Frank LM , Wilson MA , Brown EN (2001) Construction and analysis of non-Poisson stimulus response models of neural spike train activity. Journal of Neuroscience Methods, 105: 25-37, 2001.

Frank LM, Brown EN, and Wilson MA (2000) Trajectory encoding in the hippocampus and entorhinal cortex. Neuron, 27: 169-178.

Barbieri R, Frank LM, Quirk MC, Wilson MA, Brown EN (2000) A time-dependent analysis of spatial information encoding in the rat hippocampus. Neurocomputing, 32: 629-635.

Brown EN, Frank LM, Tang D, Quirk MC, Wilson MA (1998) A statistical paradigm for neural spike train decoding applied to position prediction from ensemble firing patterns of rat hippocampal place cells. Journal of Neuroscience, 18: 7411-7425.