- Poster presentation
- Open Access
Spiking neural network models for memorizing sequences with forward and backward recall
BMC Neuroscience volume 10, Article number: P211 (2009)
It has been reported that when a rat reaches the end of a track, the hippocampal place cells, which spike sequentially during the run, can generate spikes in a short time window in the reverse order [1, 2]. Also, a forward replay has been recorded from the place cells in a short time interval prior to the run . A recently published model  uses phases in the theta cycle to code different locations of the rat. A disadvantage of such an approach is that a frequency of the theta rhythm changes in a wide range and it is not clear whether the phase coding can be reliable under condition of the variable theta-frequency. Here, we present an oscillatory neural network model that can memorize sequences by adapting their synaptic weights during the memorization period. The model includes Hodgkin-Huxley type spiking elements arranged in small groups with excitatory and inhibitory connections. These groups can generate a rhythmic activity in the theta range. Excitatory elements in different groups are coupled by plastic excitatory connections. We propose that each group of rhythmic elements includes two subgroups of excitatory neurons. For the first sub-group, the modification of inter-group excitatory connections is according to the Spike Timing Dependent Plasticity (STDP) learning rule and for the second group the modification is according to the anti-STDP learning rule [3, 4]. Also, these learning rules are adjusted to the cycles of the theta rhythm. This approach allows us to memorize locations in the track during the running stage and replay them in forward order by the first sub-network and in the reverse order by the second sub-network (Figure 1). To initiate forward (backward replay we use a group of neurons corresponding to the beginning (end) of the track. In both cases the duration of replay is short and takes one cycle of the theta rhythm, in accordance with experimental results [1, 2].
Foster DJ, Wilson MA: Reverse replay of behavioural sequences in hippocampal place cells during the awake state. Nature. 2006, 440: 680-683. 10.1038/nature04587.
Diba K, Buzsaki G: Forward and reverse hippocampal place-cell sequences during ripples. Nature Neuroscience. 2007, 10: 1241-1242. 10.1038/nn1961.
Koene RA, Hasselmo ME: Reversed and forward buffering of behavioural spike sequences enables retrospective and prospective retrieval in hippocampal regions CA3 and CA1. Neural Networks. 2008, 21: 276-288. 10.1016/j.neunet.2007.12.029.
Han VZ, Grant K, Bell CC: Reversible associative depression and nonassociative potentiation at a parallel fiber synapse. Neuron. 2000, 27: 611-622. 10.1016/S0896-6273(00)00070-2.
Rumsey CC, Abbott LF: Synaptic equalization by anti-STDP. Neurocomputing. 2004, 58–60: 359-364. 10.1016/j.neucom.2004.01.067.
This work is supported by UK EPSRC (Grant EP/D036364/1).
About this article
Cite this article
Chik, D., Borisyuk, R. Spiking neural network models for memorizing sequences with forward and backward recall. BMC Neurosci 10, P211 (2009). https://doi.org/10.1186/1471-2202-10-S1-P211
- Neural Network Model
- Learning Rule
- Synaptic Weight
- Place Cell
- Excitatory Neuron