Slowness and Spike-Timing–Dependent Plasticity
Author Information
Author(s): Henning Sprekeler, Christian Michaelis, Laurenz Wiskott
Primary Institution: Humboldt-Universität zu Berlin
Hypothesis
How can the 'temporal stability' or 'slowness' approach be implemented within biologically realistic spike-based learning rules?
Conclusion
The study shows that spike-timing–dependent plasticity can implement the slowness principle, allowing neurons to learn invariant representations.
Supporting Evidence
- Neurons can learn to recognize objects despite changes in context by focusing on slowly varying aspects of input signals.
- Spike-timing–dependent plasticity can be interpreted as an implementation of the slowness principle.
- The learning window for STDP is influenced by the shape of the excitatory postsynaptic potential.
Takeaway
This study explains how neurons can learn to recognize objects by focusing on changes that happen slowly over time, using a special learning rule.
Methodology
The study uses mathematical modeling to analyze how slow feature analysis can be implemented in continuous and spiking model neurons.
Limitations
The model simplifies the behavior of real neurons and does not account for all complexities of synaptic plasticity.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website