Project 4

Models of neural system integration of sequence memory
Michael Hasselmo

Computational models of the hippocampal formation, entorhinal cortex and prefrontal cortex will be used to simulate memory-guided behavior in both rat and human behavioral tasks. These simulations will extend previous modeling work in this laboratory (Hasselmo et al., 2002b; Fransen et al., 2002; Koene et al., 2003; Hasselmo, 2005a; Koene and Hasselmo, 2005; Hasselmo and Eichenbaum, 2005) to generate predictions about physiological data in a range of different memory-guided tasks.

Simulations of memory-guided behavior of a virtual rat in a virtual spatial alternation task, with the activity of the simulation matching experimental data: A.) in the simulation, oscillations in magnitude of synaptic inputs to CA1 (top) match current source density data on theta rhythm (bottom - Brankack et al., 1993), B1.) Input representing place and reward causes activity in neuron populations representing prefrontal cortex (PFC), entorhinal cortex, dentate gyrus DG , CA3 and CA1. Output from PFC correctly guides next action in the spatial alternation task. B2.) simulated neuron in CA1 replicates selective splitter cell responses after right turn but not left turn trials (gray scale represents firing rate), C1.) CA1 place cells show theta phase precession (Skaggs et al., 1996), C2.) the model effectively simulates this theta phase precession phenomenon during virtual movement on a linear track

Selection of behavioral actions in the model depends on the encoding and context dependent retrieval of sequences of input stimuli (i.e. episodes). The models will generate predictions for specific research projects in the center, including:

  1. generation of predictions about the timing of neuron firing relative to stimuli and hippocampal theta rhythm during performance of the order recognition task with odors in rats (Eichenbaum project 2), and the magnitude of fMRI activation associated with correct versus incorrect performance of an order recognition task in humans (Stern project 1)
  2. generation of predictions about context-dependent properties and theta phase of neuronal firing in the odor sequence disambiguation task (Eichenbaum project 2), and the magnitude of fMRI activation associated with choice in a disambiguation task in humans (Stern project 1),
  3. generation of predictions about the delayed non-match to place (DNMP) task in the T-maze (White project 3), concerning the timing of splitter cell responses and sequence readout relative to theta rhythm and behavior and the disruption of behavioral responses caused by stimulation at different phases of theta during different task periods.
These simulations will use networks of neurons starting with threshold units and building to use of detailed compartmental biophysical simulations, to relate network dynamics to intrinsic currents and subclasses of neurons within the hippocampus and entorhinal cortex. The insights gained from this work will improve our understanding of the normal function of the hippocampus and associated cortex, potentially contributing to treatment of memory impairments in disorders such as Alzheimer's disease or schizophrenia.

Participating Institutions:
Boston University Center for Memory and Brain Department of Biomedical Engineering Department of Mathematics