it's 'just' activation.
the 'excitation'/'inhibition' is all exquisitely 'engineered' with respect to
the single 'goal' of TD E/I-minimization.
it's =Topologically Distributed=. the Topology, inherent, is ordered with
respect to 'pain'.
everything's built 'on top of' 'pain' avoidance.
these things are discussed in AoK's 'short-paper' section (and have been
discussed in this, and other online 'places', ad nauseum.
all that's required is activation.
there isn't any 'magic'.
given activation, the neural topology produces TD E/I-minimization, which
simultaneously by-produces 'behavior'.
what i've been doing is pleading with folks to do the primary stuff as it can
be done, rather than resigning themselves to its blind automation.
the secondary products (by-products)... 'behavior'... will follow suit.
no more blind slaughter would be the longed-for result.
K. P. Collins
>Subject: Re: inhibitory LTP
>From: Lawrence Ryan loryan at sfu.ca>Date: Tue, 19 October 1999 11:47 AM EDT
>Message-id: <7ui3q4$fmm$1 at nnrp1.deja.com>
>>And what is the trigger?
>With excitatory synapses, it's a level of correlation between excitatory
>input and backprop (among other things, perhaps). What triggers
>inhibitory LTP? Simply a frequency of inhibitory input above a threshold
>level?
>>>Yes they do. One possible post-synaptic mechanism is increased >number
>>of GABA(A) receptors.
>[...]
