You are more up to date and grounded in this than I am, so I appreciate
the update and share the suspicion or question of how something grounded
in spatial sensorimotor feedback can evolve or exapt into a more general
intelligence, but it doesn't seem entirely out of range... just a bit of
an as-yet-detailed details missing.
It does seem like a plausible story for sensorimotor grounding,
compositional object modeling, and local model voting but the open task
is to show how that grows into valenced allostasis, temporal-causal
depth, cross-domain abstraction, and self-regulating attention.
My original throwdown mostly only was intended to reference the
precedent for a mode of local-to-global phase transition. This
abstraction (l-g-p-t) is a nice articulation of the general abstraction
which I tripped over trying to formulate this response. That alone
feels worth the cost of entry into this discussion.
Glad to see another FriAM convo which pushes me in (yet) new ways I
already found interesting...
On 7/8/26 8:15 am, glen wrote:
I was introduced to Monty
<https://github.com/thousandbrainsproject/tbp.monty> something like
last year or the year before maybe. It was a presentation and I've
forgotten who invited them to talk. It was intriguing as an
alternative to deep learning world modelers.
But I was always a bit leery of the CMP in the sense that it's
grounded in *space*. While it seems obvious that *animals* (including
robots with onboard computers, but not robots tethered to their
computers) are grounded in 3D space (+ time), it's not at all obvious
that things like interoception is so grounded. So, given others may
know more about Monty or the whole program of 1kbrains, can we
formulate CMP hypotheses based on something akin to "poses", but in
other homeostaic/allostatic "spaces"?
To be clear, it seems to me that such "spaces" can be inferred by the
deep learning modelers like VLA or JEPA. But Monty seems a bit
top-down ... like it's imposing how *we* work onto the artificial
organisms we might one day build. (That's not pejorative ... just
evidence of a slightly different purpose.)
.- .-.. .-.. / ..-. --- --- - . .-. ... / .- .-. . / .-- .-. --- -. --. / ...
--- -- . / .- .-. . / ..- ... . ..-. ..- .-..
FRIAM Applied Complexity Group listserv
Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoom
https://bit.ly/virtualfriam
to (un)subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
FRIAM-COMIC http://friam-comic.blogspot.com/
archives: 5/2017 thru present https://redfish.com/pipermail/friam_redfish.com/
1/2003 thru 6/2021 http://friam.383.s1.nabble.com/
