Re: activation decay between events (leabra++)

["Randall C. O'Reilly" <oreilly@grey.colorado.edu>]

FYI you can also set trial_init to DECAY_STATE in which case it will
actually use that decay parameter you set in the hidden layer spec, if
you want a level of decay between all or nothing.

				- Randy

Matthijs van der Meer <M.vanderMeer@ed.ac.uk> writes:
> To answer my question, yes, I've overlooked the hidden "owner" field of
> the trial process, which in my case is called taBase:Actions, and
> editing this reveals the "trial init" field which can be set to [Do
> Nothing] instead of [Init State].
> 
> Sorry for bothering you all - at least it's in the archives now!
> 
> (Additional keywords: activation-based priming)
> 
> On Fri, 2002-08-09 at 00:02, Matthijs van der Meer wrote:
> > Dear PDP people,
> > 
> > I'm having trouble getting a particular leabra hidden layer to preserve
> > its activation between events. Setting the events.decay parameter in the
> > .specs to 0 (from the default 1) does not change anything. I've also
> > tried changing this parameter from a script, but it has the same lack of
> > effect.
> > 
> > Am I overlooking some other process/event/gotcha that somehow resets the
> > hidden layer's activation?