Ambiguity Simulations
- I've been hoping to do some psychological simulations but haven't
found the time.
- You could get a net to categorise stimuli
(in say 15 cycles or 150 ms.).
- It could then categorise an ambiguous stimuli in say 22 cycles.
- I think this mimics someones experimental data (Marslen-Wilson?).
- I had an RA start to do this, but there wasn't much progress.
- One thing that would need to be investigated is the
speed of ignition as our small nets (400 or 900) tend to ignite
in less than 7 cycles.
- It seems you can reasonably ignore say 200 ms of the data is
input/output parts of the system.
- Eventually this will have to be resolved but not until we
actually get IO working.
