Neuroscience 2005 Abstract
| Presentation Number: | 198.12 |
|---|---|
| Abstract Title: | What produces attractor effects in the rat hippocampus? |
| Authors: |
Papp, G.*1
; Treves, A.1,2
1Cognitive Neuroscience, SISSA, Trieste, Italy 2Norway, via Beirut 2, 34014, |
| Primary Theme and Topics |
Cognition and Behavior - Animal Cognition and Behavior -- Learning and memory: Physiology and imaging |
| Session: |
198. Place Cells III Poster |
| Presentation Time: | Sunday, November 13, 2005 11:00 AM-12:00 PM |
| Location: | Washington Convention Center - Hall A-C, Board # MM26 |
| Keywords: | HIPPOCAMPUS, SPATIAL MEMORY |
Spatial locations are thought to be encoded in the rat hippocampus as positions on 2D continuous attractors, each representing (a chart of) a distinct evironment. In a simple recurrent neural network with several discrete attractor states, population activity tends to align with the closest attractor, so that slightly crossing the boundary between two basins of attraction can lead to sudden transitions in the final state. Leutgeb et al (SfN 2004) aimed to detect such attractor states in CA3 activity by first training rats to forage in a square and in a circle box, and the testing them in 5 intermediate morph shapes. Unexpectedly, activity in the morphs changed gradually, interpolating between the representations of the extremes.
We ran computer simulations on a reduced hippocampal model (Treves, Hippocampus 2004) to test the hypothesis that the initially well separated attractors at the extremes might have been stretched by ongoing plasticity to accomodate intermediate shapes. In the simulations, we could test the virtual rat while blocking plasticity, and again after unblocking it. Our results show that positional information is indeed slightly more accurate at the extremes (E1, E7) than in the morphs (E2-6), before learning the morphs (dashed line). After reinstating associative plasticity the effect disappears (solid line). The U-shape is more salient in CA1 than in CA3, and it is not matched by a sigmoid in the relative similarity of the representations. We hypothesize that the 2D nature of the continuous charts favours smooth interpolation between distinct environments, unless they are fully orthogonalized by ad hoc pretraining, as in (Wills et al, Science 2005).
We ran computer simulations on a reduced hippocampal model (Treves, Hippocampus 2004) to test the hypothesis that the initially well separated attractors at the extremes might have been stretched by ongoing plasticity to accomodate intermediate shapes. In the simulations, we could test the virtual rat while blocking plasticity, and again after unblocking it. Our results show that positional information is indeed slightly more accurate at the extremes (E1, E7) than in the morphs (E2-6), before learning the morphs (dashed line). After reinstating associative plasticity the effect disappears (solid line). The U-shape is more salient in CA1 than in CA3, and it is not matched by a sigmoid in the relative similarity of the representations. We hypothesize that the 2D nature of the continuous charts favours smooth interpolation between distinct environments, unless they are fully orthogonalized by ad hoc pretraining, as in (Wills et al, Science 2005).
Sample Citation:
[Authors]. [Abstract Title]. Program No. XXX.XX. 2005 Neuroscience Meeting Planner. Washington, DC: Society for Neuroscience, 2005. Online.
Copyright © 2005-2025 Society for Neuroscience; all rights reserved. Permission to republish any abstract or part of any abstract in any form must be obtained in writing by SfN office prior to publication.
