## Bidirectional Completion Of Cell Assemblies In The Cortex (1998)

Venue: | In Computational Neuroscience: Trends in Research |

Citations: | 5 - 5 self |

### BibTeX

@INPROCEEDINGS{Sommer98bidirectionalcompletion,

author = {Friedrich T. Sommer and Thomas Wennekers and Günther Palm},

title = {Bidirectional Completion Of Cell Assemblies In The Cortex},

booktitle = {In Computational Neuroscience: Trends in Research},

year = {1998},

publisher = {Plenum Press}

}

### OpenURL

### Abstract

This paper examines the hypothesis that synaptic modification and activation flow in a reciprocal cortico-cortical pathway correspond to learning and retrieval in a bidirectional associative memory (BAM): Unidirectional activation flow may provide the fast estimation of stored information, whereas bidirectional activation flow might establish an improved recall mode. The idea is tested in a network of binary neurons where pairs of sparse memory patterns have been stored in bidirectional synapses by fast Hebbian learning (Willshaw model). We assume that cortical long-range connections shall be efficiently used, i.e., in many different heteroassociative projections corresponding in technical terms to a high memory load. While the straight-forward BAM extension of the Willshaw model does not improve the performance at high memory load, a new bidirectional recall method (CB-retrieval) is proposed accessing patterns with highly improved fault tolerance and also allowing segmentation of ambiguous input. The improved performance is demonstrated in simulations. The consequences and predictions of such a cortico-cortical pathway model are discussed. A brief outline of the relations between a theory of modular BAM operation and common ideas about cell assemblies is given. THE MODEL

### Citations

1579 | Neural networks and physical systems with emergent collective computational abilities - Hopfield - 1982 |

513 | Distributed hierarchical processing in the primate cerebral cortex
- Felleman, Essen
- 1991
(Show Context)
Citation Context ... a new retrieval scheme. We call it crosswise bidirectional (CB) retrieval since matrix rows and columns are evaluated simultaneously: y(r+1) j = H( X i2x(r) C ij [C T y(r \Gamma 1)] i \Gamma \Theta) =-=(5)-=- x(r+1) i = H( X j2y(r) C ij [Cx(r \Gamma 1)] j \Gamma \Theta 0 ) (6) For r = 0 pattern y(r\Gamma1) has to be replaced by the simple retrieval result H([Cx(0)]\Gamma`), for r ? 2 Boolean ANDing with r... |

253 | The Oraanization of Behaviour - Hebb - 1949 |

157 | Bidirectional Associative Memories - Kosko - 1988 |

98 | Non-holographic associative memory - Willshaw, Buneman, et al. - 1969 |

82 |
Anatomy of the cortex: Statistics and geometry
- Braitenberg, Schüz
- 1991
(Show Context)
Citation Context ...In the Willshaw model each neuron j forms the dendritic potential d(0) j := [Cx(0)] j = X i C ij x(0) i ; (2) and determines its activity value by threshold comparison y(1) j = H(d(0) j \Gamma `) 8j; =-=(3)-=- with the global threshold value ` and H(x) denoting the Heaviside function. In the following eqs. (2) and (3) are denoted as simple retrieval. In auto-associative memories usually iterative retrieval... |

77 | The Hebbian paradigm reintegrated: Local reverberations as internal representations
- Amit
- 1995
(Show Context)
Citation Context ...s are sparse, i.e., a; b !! n; m. For this case the Willshaw model 16, 18 is efficient, where only minimum synaptic depth is required by binary clipping of the outer product sum: C ij = sup x i y j : =-=(1)-=- During retrieval an associative memory maps an initial pattern x(0) closest to a memory pattern x onto a pattern close to y . In the Willshaw model each neuron j forms the dendritic potential d(0) j ... |

22 | Improved bidirectional retrieval of sparse patterns stored by hebbian learning - Sommer, Palm - 1999 |

11 |
Physiological mechanisms underlying longterm potentiation
- Gustafsson, Wigstrom
- 1988
(Show Context)
Citation Context ...ieval since matrix rows and columns are evaluated simultaneously: y(r+1) j = H( X i2x(r) C ij [C T y(r \Gamma 1)] i \Gamma \Theta) (5) x(r+1) i = H( X j2y(r) C ij [Cx(r \Gamma 1)] j \Gamma \Theta 0 ) =-=(6)-=- For r = 0 pattern y(r\Gamma1) has to be replaced by the simple retrieval result H([Cx(0)]\Gamma`), for r ? 2 Boolean ANDing with results from timestep r \Gamma 1 can be applied. RESULTS Memory Perfor... |

9 | Iterative retrieval of sparsely coded associative memory patterns - Schwenker, Sommer, et al. - 1996 |

4 |
Characterization of a calcium-dependent current generating a slow afterdepolarization of CA3 pyramidal cells in rat hippocampal slice cultures
- Caeser, Brown, et al.
- 1993
(Show Context)
Citation Context ... strategy or with a Boolean ANDing () in each component between the old and the new input pattern applied after the second iteration step 13 , i.e., rs1: x(r+1) j = x(r) jsH(d(r) j \Gamma \Theta) 8j: =-=(4)-=- Biologically, Boolean ANDing could be realized by after-depolarization effects, where the cell response is fascilitated by its recent previous activation (cell fascilitation) 4, 9, 2 . Crosswise Bidi... |

4 | Stability, capacity and statistical dynamics of second-order bidirectional associative memory - Leung, Chan, et al. - 1995 |

4 | Bidirectional retrieval from associative memory - Sommer, Palm - 1998 |

3 | Two coding stragegies for bidirectional associative memory - Wang, Cruz, et al. - 1990 |

2 | Novel lists of 7 \Sigma 2 known items can be reliably stored in an oscillatory short-term memory network: Interaction with long term memory - Lisman - 1996 |

2 | Patchiness as a means to get a message across - Schüz - 1994 |

1 |
Hippocamal two-stage learning and memory consolidation
- Bibbig, Wennekers
- 1996
(Show Context)
Citation Context ...memory maps an initial pattern x(0) closest to a memory pattern x onto a pattern close to y . In the Willshaw model each neuron j forms the dendritic potential d(0) j := [Cx(0)] j = X i C ij x(0) i ; =-=(2)-=- and determines its activity value by threshold comparison y(1) j = H(d(0) j \Gamma `) 8j; (3) with the global threshold value ` and H(x) denoting the Heaviside function. In the following eqs. (2) and... |