Week 6: Neural Modelling Flashcards
1
Q
Types of networks
A
- Logical circuits (localist networks)
- Fully distributed 3-layer neural networks (auto-associative aka attractor, hetero-associative aka feedforward)
- Deep neural 6-layer
- Whole brain
2
Q
Types of neurons
A
McCulloch-Pitts
Mean field
Spiking
3
Q
McCulloch-Pitts neuron
A
Analogue input
Digital output
4
Q
Mean field neuron
A
Analogue input
Analogue output
5
Q
Spiking neuron
A
Analogue input
Outputs (analogue) frequency of discrete events
6
Q
7 constraints on models
A
1) Integration of modelling at different levels
2) Neuron models
3) Synaptic plasticity
4) Inhibition
5) Network area structure mimics cortical area structure
6) Within-area local connectivity
7) Between-area global connectivity
7
Q
Jumping links
A
Allow for reverberation and verbal working memory
8
Q
Connector hub areas
A
Cortical assemblies that are local hubs that connect to more distal modules. Like Tiverton
9
Q
Fast-mapping
A
New models show that Hebbian learning doesn’t always need repeated exposure to a word
