Vocabulary Networks: Program-3
This simulation shows the effect of turning words ON
when the network is not in a stable attractor state.
It builds a network like the ones in Program-1, and lets
the network settle into a stable attactor state.
You can move the network out of its attractor state by
changing the value of the SPK parameter. This parameter
turns ON a number of words at update 99.
THINGS TO DO.
Find a network with a low level of activation. Set
SPK to a high value like 600, and set nEv to 0. How does
the network react to this spike?
Now experiment with the nEv and sEv parameters and see
if you can find a combination values that prevents the
network from falling back into its attractor state.
QUESTIONS TO ASK.
How long do the effects of a large spike last for?
Do spikes affect high or low level vocabularies more?
How do spikes interact with clusters of activation?
What implications does this set of behaviours have
for theories of L2 vocabulary acquisition?