Discussion of the Disambiguation Experiment
On the good side:
- This is a reasonable model for homonymous words.
- All senses are activated and one wins out.
- It's based on a neurally plausible model.
On the bad side:
- It doesn't agree with the Necker Cube. Why?
- The interface is not biologically viable
- Why is Learning turned off? CAs in long-term memory should be resilient
- The timing doesn't agree with psychological results
- The CANT model used doesn't agree with some known human parameters (e.g.
using 400 neurons)
