Changing Fatigue
- One way to achieve this is to change the fatigue mechanism.
- In our original fatigue mechanism, each neuron had a fatigue
value associated with it.
- For the neuron to fire it had to have activity greater than
the threshold + fatigue.
- If it fired fatigue went up by one constant.
- If it didn't fire fatigue went down by another constant (but never
less than 0).
- However, following Kaplan, Sontag and Chown's (1991) mathematical
model of CA behaviour, we forked this into two types of decay.
- The first was long term, and the second short term.
- The short term decay constants were large and lead to
neural firing oscillation.
- The long term decay constants were small but the fatigue constant
was much larger (relatively) than the recovery constant.
- This meant that overall neural activity of a CA went down over
time.
- I was really pleased with this result as it fit in with my
thinking about the model.
- However, Peter decided to try another mechanism.
