Learning to Respond to Stimulus Pairs in Neural Networks
Author Information
Author(s): Bourjaily Mark A., Miller Paul, Gutkin Boris
Primary Institution: Brandeis University
Hypothesis
What synaptic plasticity rules can generate stimulus-pair specificity within a randomly connected network of spiking neurons?
Conclusion
The study found that long-term potentiation of inhibition is crucial for generating reliable decision-making in neural networks responding to stimulus pairs.
Supporting Evidence
- Long-term potentiation of inhibition was found to be the most beneficial plasticity rule.
- Selective neurons were present in initial networks, and appropriate plasticity mechanisms improved task performance.
- Networks trained with LTPi alone or in combination with STDP produced reliable decision-making across a wide range of networks.
Takeaway
This study shows that neurons in the brain need to learn to respond to specific pairs of things, like colors and shapes, to make good decisions, and that certain rules help them do this better.
Methodology
The study used computational models of spiking neuron networks to investigate how different synaptic plasticity rules affect stimulus-pair selectivity.
Limitations
The findings may not directly translate to biological systems due to the simplifications in the computational model.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website