Yeah, that would be good. But I can't shake the idea that our obsession with
animals, and animals with neural structures, is preempting our study of
sensation. It's akin to identifying AI with language.
For example,
Making friends with a tiny spider
https://youtu.be/i6ucE1cfzmE?si=r-aPULyyw6X-5M0S
Even though the spider has no language, it's literally trivial for *us* to
mentalize, both about its knowledge and its beliefs. That
triviality/convenience gets in the way, prevents understanding. What's more
interesting, I think, is our ability (if it exists at all) to mentalize, say,
paramecia. What knowledge or beliefs can we expect such animals to have? Or,
even worse, to what extent can we mentalize plants? Does my western red cedar
*hate* that stupid cherry tree that's all up in its space? Or maybe it's happy
with the cherry tree for holding together the nearby turf, abating the erosion
down the already steep slope ... and helping to retain water in the very rocky
soil ... given the cedar doesn't have a tap root.
Such bias is also akin to problems with utopian visions like this:
https://www.whitehouse.gov/wp-content/uploads/2026/07/Science-A-New-Golden-Age.pdf
Such "moonshots" might be fizzy generators for lots of cool stuff. Or, more
likely, suck up all the money preventing very expensive and difficult basic research. As
with every such effort, it's ripe for grifting. What's the word for it? Goodhart's Law?
When the pot of money is really large, it draws more corruption ... players that game the
system ... optimizers. Although I like connectomics, I'm always worried the target is so
high it attracts petitio principii.
On 7/30/26 9:10 AM, Marcus Daniels wrote:
Neuralink in wildlife would provide an interesting capability. Like opossums
and birds.
*From: *Friam <[email protected]> on behalf of cody dooderson
<[email protected]>
*Date: *Wednesday, July 29, 2026 at 11:05 AM
*To: *The Friday Morning Applied Complexity Coffee Group <[email protected]>
*Subject: *Re: [FRIAM] crickets
I had a similar conversation about senses the other day, inspired by
https://ai-2027.com/ <https://ai-2027.com/>, where we speculated on when robots
might take over the planet.
We speculated that the rise of robots is far out because of the primitive
robotic senses compared to humans and other animals. We can feel heat, cold,
pressure, and pain across almost our entire bodies. In that respect, robots are
relatively blind. Even if they have heat sensors, they lack the spatial
resolution of the sensors in human skin.
For instance, last night in the pitch dark, I stepped on what is likely the
sharpest Hot Wheels car ever created. It rolled under my bare foot and nearly
took me down, but yelling a loud curse word helped me keep my balance. I doubt
a robot would have that kind of sensory feedback in its foot. I haven't looked
up what pain sensors currently exist for robots, but I speculate that they are
primitive. For instance an autonomous vehicle like a Waymo relies mostly on
vision, albeit very sophisticated vision like lidar. It cannot feel the pointy
parts of a bicycle as it rolls over it.
_ Cody Smith _
[email protected] <mailto:[email protected]>
On Mon, Jul 27, 2026 at 10:49 AM glen <[email protected]
<mailto:[email protected]>> wrote:
I had an interesting argument with a friend Friday at the pub. He speculated that there is a
(much) larger diversity of audio sensors/processors across biology than of light/vision
sensors/processors. It wasn't a very concrete speculation because we went 'round a few times about
vision versus light and *use* of audio versus *use* of light, etc. It even descended into a
conversation about the jargonality of "radial" versus "bilateral" animals. I
normally look this stuff up a bit before posting about it. But I haven't yet and I likely won't get
the chance until my next episode of insomnia.
But my *doubt* was expressed from the perspective of serial versus parallel
processing. I tried to counter-speculate that vision (either in animals with
multiple eyes or a small number of eyes, but multiple sensors like rods/cones
laid out in a plane) seems to often be parallel. And that audio seems (mostly?)
sequential. Granted, *fusion* takes place with both (hence the discussion of
bilateral animals). But not only do animals have binocular (or more) vision,
but we also have parallel sensors like retinae.
This led to my raising the degrees of freedom from scaffolding question:
When iterating off a scaffold, have the degrees of freedom increased or
decreased from the state before/without the scaffold? If one argues the DoF are
fewer atop the scaffold, then it might make sense that vision apparatuses
exhibit less diversity.
Of course, our stack of nonsense had piled very high at this point. So we
moved on to arguing about the uses of fernet in various cocktails ...
On 7/24/26 8:04 PM, Eric Charles wrote:
>
> "Anyway, my point is that if we thought of these subnetworks/circuits as separate ganglia that could have been located closer to the sensorimotor manifold they manage, then interoception kindasorta reduces away."
>
> Humans have fairly simple retinas. Roughly 3 layers. Built backwards. 3 types of color detectors, if you are lucky. The three-layers-built-backwards part is conserved across pretty much all vertebrates, but in terms of vertebrate mammal retinal complexity is mid. Birds and reptiles have much more "processing" complexity in the eyes, as well as more types of color receptors. Some birds have 10 times the number of retinal neurons we have in the equivalent space. In addition to lots of critters having 4 or 5 types of cones, some amphibians even get an extra type of rod.
>
> And then if you get into invertebrates there is a ton more that can happen at the back of the eye, and many more types of receptors (Mantis Shrimp have the currently known record, with 12 types of color receptors). You also can get eyes that are not built backwards, i.e., where the light receptors are actually in the front.
>
> I suspect, historically (i.e., across evolutionary time), we lost a lot of that retinal complexity when we were small, nocturnal animals with very little reliance on color vision. When we went back to day-time living, and ripe-fruit targeting, the complexity ended up in the back of the brain rather than re-evolving into the retina. Hard to say why, but it seems to have worked out well for us.
>
> How this relates to interoception I'm unsure.
>
> Best,
> Eric
>
>
> On Thu, Jul 16, 2026 at 11:55 AM glen <[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>> wrote:
>
> Yes, the way the active inference people (in particular - much less so than the predictive processing people) write/talk is either simply very dense *or* laden with bias.
>
> But re: your cross-trophy (?) comment - I can't help but think we're a lot more like octupusses than we want to admit. Each anatomical region of the brain might be thought of as a semi-independent ganglion. And maybe it just so happens our ganglia are clustered in our head/CNS for convenience or some sort of energy minimization ... or maybe risk minimization (which is why the meningeal lymphatic architecture paper was included in my list). Sure, *some* newer regions may only be 'efficient' because our ganglia are co-located. But we might also be able to smear those out, distribute them through (an artificial) organism with other pathways.
>
> Anyway, my point is that if we thought of these subnetworks/circuits as separate ganglia that could have been located closer to the sensorimotor manifold they manage, then interoception kindasorta reduces away.
>
> On 7/11/26 3:33 PM, Eric Charles wrote:
> > I skimmed the predictive processing chapter. My main thought was
wondering if it still seemed interesting if translated into less jargon-ey English.
> >
> > My second thought was: Doesn't Gibson help with this? Isn't the "loop"
problem a "problem" at least in part because you are trying to find the loops inside the
organism instead of in the larger organism-environment system?
> >
> > /If/ there is specification in the ambient energy, /then /the loop can extend outside the organism, and
exist at whatever scale the relevant patterns exist at. Yes, you need sufficient neuronal support for the organism to do
their part of the loop... at least with us very-neuronal organisms... but that type of "complexity" is different
than the complexity one might imagine if they thought _all_ the work had to happen inside the head. For example, recognizing
that one is accelerating towards the ground --- because the rate of optic expansion is accelerating --- could be much easier
than trying to "predict" whether one is accelerating towards the ground by comparing a series of snapshot images
and trying to neuronally create an internal model of everything happening around you. You can offload most of that by not
trying to "model" things that are directly perceivable. And that can extend as far as we can extend
"perceivable."
> >
> > Re the first thought.... Here is a quote:
> >
> > The second type of predictive circuitmight support the
sequencing and arbitration of behaviors. Simple solutions to these problemsmight
appeal to generative models for sequential dynamics (Parr et al., 2023). For example,
locomotion behavior in C. elegans might be supported by sequential generativemodels
that encode expected transitions among locomotion behaviors(e.g., forward and
backward movements, left and right turns), with decision points corresponding to
bifurcations in the dynamical sequences (Kato et al., 2015). In turn, the transitions
prioritized at bifurcation points might depend partially on external sensation and
partially on internal (e.g., interoceptive) sensations reporting impending drives and
needs, which leads us to the next point.
> >
> >
> > So far as I can tell, that means something like:
> >
> > A second thing well-developed neuron clumps do is ensure movements happen in
order, so as to form coherent "behaviors." We can understand how this happens in several ways,
some of which Ph.D.-level scientists call "simple." Roundworms, for example, swim in
predictable ways, sometimes changing direction. Those turns sometimes depend on external factors, at
least in part.
> >
> >
> > Seriously... this might be my weird form of getting old and yelling at kids to get off
my lawn... but... Ugh... A lot of that might as well be pretentious continental philosophers stringing
jargon together. "The epoch of the logos thus debases writing considered as mediation of mediation
and as a fall into the exteriority of meaning. To this epoch belongs the difference between signified and
signifier, or at least the strange separation of their 'parallelism,' and the exteriority, however
extenuated, of the one to the other." Orwell could take at least half the sentences as examples of
bad writing, and not be wrong. Even in my attempt at a translation, the word "ensure" is highly
suspect.
> >
> > Best,
> > Eric
> >
> > <mailto:[email protected] <mailto:[email protected]>
<mailto:[email protected] <mailto:[email protected]>>>
> >
> >
> > On Tue, Jul 7, 2026 at 10:35 AM glen <[email protected] <mailto:[email protected]>
<mailto:[email protected] <mailto:[email protected]>> <mailto:[email protected]
<mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>>> wrote:
> >
> > Of course. You have a knack for pushing my buttons. 8^D
> >
> > What irritates me about all this active inference and predictive
processing advocacy [⛧] is well-represented in the title of that chapter "From
Sensorimotor Skills to Higher Cognition". [grrrr] The reason I took the time to download
it and start skimming it was my hope for a thorough *composition* from the very small-fast
feedback loops to the large-slow ones. There are a lot of citations. So maybe the clues are in
there. But I'm lazy.
> >
> > What I *want* ... what I really really want is evidence of
predictive processing in a minimal model organism like C. Elegans or Drosophilia.
Such exist [1-5]! But now we need something like connectome (or simpler?) circuits in
more complex organisms that show how small-fast predictive processing composes into
large-slow predictive processing. Does the model work at *all* scales? Only some
scales? Is it like a percolating stew of predictions, some of which are suppressed by
the larger circuits?
> >
> > Speaking of which, I discovered this book just last night:
> >
> >
https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952
<https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952>
<https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952
<https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952>>
<https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952
<https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952>
<https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952
<https://bookshop.org/p/books/from-human-reasoning-to-belief-an-empirical-
account-joshua-mugg/de6c8394b4e24d99?ean=9781032736952>>>
> >
> > But as always, it's silly to keep buying books I'll never read. I post
it here in the hopes that you readers out there might read it and tell me what it says ... or
maybe I'll buy the epub and feed it to Claude ... or maybe it's read it already? I haven't
checked. You'll remember we've had such arguments before, when you claimed I *must* believe in
the floor in order to get out of bed in the morning. And my counter was that it is my *doubt*
about the existence of the floor that allows me to get out of bed. IDK if Mugg's "DJ
mixing board" model fits one of our stances better. But I do like it better than the
overly simplistic fast vs slow thinking model.
> >
> >
> > [1] Dimakou A, Pezzulo G, Zangrossi A, Corbetta M. The
predictive nature of spontaneous brain activity across scales and species. Neuron.
Published online March 1, 2025. doi:10.1016/j.neuron.2025.02.009
> > [2] Kaplan H, Nichols A, Zimmer M. Sensorimotor integration in
Caenorhabditis elegans: a reappraisal towards dynamic and distributed computations.
Philosophical Transactions of the Royal Society B: Biological Sciences. 2018;373.
doi:10.1098/rstb.2017.0371
> > [3] Kim A, Fitzgerald J, Maimon G. Cellular evidence for
efference copy in Drosophila visuomotor processing. Nature neuroscience.
2015;18:1247-1255. doi:10.1038/nn.4083
> > [4] Lin A, Witvliet D, Hernandez-Nunez L, Linderman S, Samuel
A, Venkatachalam V. Imaging whole-brain activity to understand behavior. Nature
reviews Physics. 2022;4:292-305. doi:10.1038/s42254-022-00430-w
> > [5] Wang S, Segev I, Borst A, Palmer S. Maximally efficient
prediction in the early fly visual system may support evasive flight maneuvers. PLoS
Computational Biology. 2019;17. doi:10.1371/journal.pcbi.1008965
> >
> >
> > [⛧] It seems to me that most of the peri-Friston work borders
on advocacy of the model(s) as opposed to challenging them. But I'm not a scholar. So
my scope is very small.
> >
> > On 7/6/26 8:15 PM, Nicholas Thompson wrote:
> > > Hi, Glen,
> > >
> > > I liked the predictive processing thing. It coheres with
an idea I have been kicking around of late. People tend to think of cognitive processes
as putting us in touch with the world as it is. Then we look at that represented world
and make decisions about the future. Wouldn't it make more sense for cognitive
processes to put us in touch with the world as it is going to be? To translate that back
into monist talk, we live in a world of successive anticipations. As I get more frail,
I become aware of all the hard work my cerebellum must be doing to anticipate the
consequences of any action I might take that changes my center of gravity. A delayed
prediction can lead to my taking actions that compound a balance prediction and send me
to the floor. it's like I am doing judo to myself.
> > >
> > > Is that annoying enough to feed the beast?
> > >
> > > Nick
> > >
> > > On Mon, Jul 6, 2026 at 6:31 PM glen <[email protected] <mailto:[email protected]> <mailto:[email protected]
<mailto:[email protected]>> <mailto:[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>>
<mailto:[email protected] <mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>> <mailto:[email protected]
<mailto:[email protected]> <mailto:[email protected] <mailto:[email protected]>>>>> wrote:
> > >
> > > It's so dead, here, I figure it can't hurt to post
arbitrary nonsense I've run across lately:
> > >
> > > Meningeal lymphatic architecture and drainage dynamics
surrounding the human middle meningeal artery
> > > https://doi.org/10.1016/j.isci.2025.113693 <https://doi.org/10.1016/j.isci.2025.113693> <https://doi.org/10.1016/j.isci.2025.113693
<https://doi.org/10.1016/j.isci.2025.113693>> <https://doi.org/10.1016/j.isci.2025.113693 <https://doi.org/10.1016/j.isci.2025.113693>
<https://doi.org/10.1016/j.isci.2025.113693 <https://doi.org/10.1016/j.isci.2025.113693>>> <https://doi.org/10.1016/j.isci.2025.113693
<https://doi.org/10.1016/j.isci.2025.113693> <https://doi.org/10.1016/j.isci.2025.113693 <https://doi.org/10.1016/j.isci.2025.113693>>
<https://doi.org/10.1016/j.isci.2025.113693 <https://doi.org/10.1016/j.isci.2025.113693> <https://doi.org/10.1016/j.isci.2025.113693
<https://doi.org/10.1016/j.isci.2025.113693>>>>
> > >
> > > Constructing a lower-bound estimate of the global
number of insect species on a hyperdiverse empirical foundation
> > > https://www.pnas.org/doi/10.1073/pnas.2524283123 <https://www.pnas.org/doi/10.1073/pnas.2524283123>
<https://www.pnas.org/doi/10.1073/pnas.2524283123 <https://www.pnas.org/doi/10.1073/pnas.2524283123>> <https://www.pnas.org/doi/10.1073/pnas.2524283123
<https://www.pnas.org/doi/10.1073/pnas.2524283123> <https://www.pnas.org/doi/10.1073/pnas.2524283123 <https://www.pnas.org/doi/10.1073/pnas.2524283123>>>
<https://www.pnas.org/doi/10.1073/pnas.2524283123 <https://www.pnas.org/doi/10.1073/pnas.2524283123> <https://www.pnas.org/doi/10.1073/pnas.2524283123
<https://www.pnas.org/doi/10.1073/pnas.2524283123>> <https://www.pnas.org/doi/10.1073/pnas.2524283123 <https://www.pnas.org/doi/10.1073/pnas.2524283123>
<https://www.pnas.org/doi/10.1073/pnas.2524283123 <https://www.pnas.org/doi/10.1073/pnas.2524283123>>>>
> > >
> > > Predictive Processing: From Sensorimotor Skills to
Higher Cognition
> > > https://doi.org/10.7551/mitpress/15999.003.0011 <https://doi.org/10.7551/mitpress/15999.003.0011>
<https://doi.org/10.7551/mitpress/15999.003.0011 <https://doi.org/10.7551/mitpress/15999.003.0011>> <https://doi.org/10.7551/mitpress/15999.003.0011
<https://doi.org/10.7551/mitpress/15999.003.0011> <https://doi.org/10.7551/mitpress/15999.003.0011 <https://doi.org/10.7551/mitpress/15999.003.0011>>>
<https://doi.org/10.7551/mitpress/15999.003.0011 <https://doi.org/10.7551/mitpress/15999.003.0011> <https://doi.org/10.7551/mitpress/15999.003.0011
<https://doi.org/10.7551/mitpress/15999.003.0011>> <https://doi.org/10.7551/mitpress/15999.003.0011 <https://doi.org/10.7551/mitpress/15999.003.0011>
<https://doi.org/10.7551/mitpress/15999.003.0011 <https://doi.org/10.7551/mitpress/15999.003.0011>>>>
> > >
> > > As always, I'm reading them in fitful bursts,
interleaved across each other and all the other open tabs and crap strewn about my desk.
So .... grain of salt and all.
> > >
>
>
--
8647 ⊥ ɐןןǝdoɹ ǝ uǝןƃ
ὅτε oi μὲν ἄλλοι κύνες τοὺς ἐχϑροὺς δάκνουσιν, ἐγὰ δὲ τοὺς φίλους, ἵνα σώσω.
.- .-.. .-.. / ..-. --- --- - . .-. ... / .- .-. . / .-- .-. --- -. --. / ...
--- -- . / .- .-. . / ..- ... . ..-. ..- .-..
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/
