Coming from the cerebellum, epileptiform isn't that much of a problem.
Long-term plasticity of interneuron inputs is though and not using a
GABA-A blocker should lead to very complex results (Mittmann and
Häusser, 2007 goes into that direction). Personally, I think that the
ideal situation is to record from pairs of identified neurons - keeping
the obvious problems with that in the back of my mind of course ;)
Cheers, Christian