2004-03-31 #5

Adaptron Test Run-5 31st March, 2004

Random sequence #1 has shown a problem dealing with unexpected stimuli.

At 16 after the B we were expecting an X as before at 2/3 but got an A. If the A were stored in LTM at 17 because it was the unexpected result we would not start executing at 1 and 4 and would not recognize the AC at 4/5. So instead we don’t store the unexpected result and start executing at 1 and 4 to recognize 4/5. However when this strategy is used at 19 and 20 it is not until 21 that the pair is recognized. This is because the context stimuli of an active S-Habit are not in LTM for matching purposes. I am going to make a radical change in the code to put these context stimuli in LTM for matching purposes just as I do currently for R-Habits and then go back and update LTM if something gets recognized.