[From Bruce Abbott (941227.1250 EST)]
Bill Powers --
Bill, I have some good news. The revamping of our animal care facility's
ventilation system that I mentioned earlier will not take place until next
fall. This means that we are clear to order rats and get down to business.
Now would be a good time to nail down what data we are going to collect and
when.
My apparatus includes a digital interface between computer and chamber; it has
only four TTL-style optically-isolated inputs, but I can gang up to three of
the interfaces together if we need more. I also have a game port that can
provide four additional digital inputs and four crude A-to-D inputs.
In the past I have written programs in Turbo Pascal to run the experiment.
The interfaces do not have interrupt capability so I use polling to determine
when an input has occurred. I reprogram the DOS clock for 100 Hz operation; a
procedure at the start of the main loop waits for a clock "tick" and then
exits; all loop operations must be completed within that 100th second or the
system falls behind real time. The computer is an 8088 10 MHz so this is a
serious constraint. However, all the programs I have written thus far have
managed to complete the loop on time (I avoid most screen I/O to achieve
this).
I assume we will be time-stamping each response and pellet delivery. I hope
to have some mechanisms in place to determine whether the rat is stationed in
front of the lever at any given moment or visiting the food cup (probably an
IR beam & sensor) but will need to construct those circuits. Given the data
rate, it would probably be a good idea to write everything to disk from time
to time.
How does this sound to you as a start? I'm thinking that we can at least get
our own data to replace Motheral's while I'm working on creating a more PCT-
compatible test environment, one in which the animal can live and which will
require no deprivation of necessities.
Regards,
Bruce