[FRIAM] Causal emergence

[David Eric Smith]

Thank you for the pointers, Jochen,

I didn’t know of this, but once I saw that it branches from Tononi, much context became available.

It seems to me that in the original PNAS article, they set up a strawman (at the level of the abstract) of how science uses micro/macro distinctions.  As if the only reason to use the macroscale description is either ignorance or laziness.  The reason I say that is a strawman is that even very non-specialist people have run across the dictum “Don’t fit to noise!” at one or another time.  Often that Is applied to degrees of freedom in the state that turn out to be dynamically undetermined by the relevant boundary conditions.  But it can refer equally to parameters in an inferred causal model on the microscale that are not robust or pertinent to the control feedbacks that hold the whole system together.  In such cases one would associate the macroscale not with a deterministic function (a summary statistic) of a more-valuable microscopically resolved datum, but rather one would associate the macroscale with a distribution over any or all of states, trajectories, or generating models at the microscale.  Then, fitting a causal model at the microscale from samples that are too-short, and getting a model that predicts, on longer terms, trajectories and states that are rare or unrepresentative of the actual distribution, amounts to missing information rather than overestimating it.  

Still, it will be good if possible to go through the PNAS and see what they build, and whether any of it is useful or has anything novel that derives from the other technical constructions of EIT.

Eric



> On Feb 10, 2022, at 12:28 PM, Jochen Fromm <jofr at cas-group.net> wrote:
> 
> There is a new preprint paper from Erik Hoel about what he calls "Causal emergence"
> https://arxiv.org/abs/2202.01854
> 
> It is an extension of his earlier work published here in PNAS about the same topic
> https://www.pnas.org/content/110/49/19790 <https://linkprotect.cudasvc.com/url?a=https%3a%2f%2fwww.pnas.org%2fcontent%2f110%2f49%2f19790&c=E,1,AypxgMmbiLk19jlFt4fwpP1NpZtqi2vVmRcek7n-FQOHJ5Vsbu3LZCjIiIKhS_uhbcAQgo4dHJAOv8OnZNbKqs21nDPfHiuGfOMkHJqBEYD9DUJlREMC1w3jpLNw&typo=1&ancr_add=1>
> 
> What do you think of it? On first sight he seems to model that aggregates of parts (for instance neurons) can influence other aggregates, which seems rather obvious to me, or do I miss something? 
> 
> -J.
> 
> 
> .-- .- -. - / .- -.-. - .. --- -. ..--.. / -.-. --- -. .--- ..- --. .- - .
> FRIAM Applied Complexity Group listserv
> Zoom Fridays 9:30a-12p Mtn UTC-6  https://linkprotect.cudasvc.com/url?a=https%3a%2f%2f%2f%2fbit.ly%2fvirtualfriam&c=E,1,9ax20xHa8aCSmY1dh-C_v8vNVFGWmr3kaeiyzGNEXuGkj6SD73a0J1Iu2da5KbV8XH0HF6dztsWEdDM1kLakSQEhHDadbYYQo8iroK4s_RtUtCPrNMU5AtgpwZs,&typo=1
> un/subscribe https://linkprotect.cudasvc.com/url?a=http%3a%2f%2fredfish.com%2fmailman%2flistinfo%2ffriam_redfish.com&c=E,1,rJpr0SOQpju7SnE0f7KRzgAkhPoC5UPbUzM6SxrmSzqtI5jVfn0fb9YlIbrLMy1wc4ou6YAL57yeD5Jm-3S36nnQX8zoq0J-YMvNmYKgu017lTf0&typo=1
> FRIAM-COMIC https://linkprotect.cudasvc.com/url?a=http%3a%2f%2ffriam-comic.blogspot.com%2f&c=E,1,4Isg63Y1Z2gIZSq-xGb1AVUkyqcPPKo0wH5O5Zv83RvBWotxGs5D9UiyelDf8xgqn54KEaQfd0NriAvswaMojtULc2-suaZYS30Qm8jRUi9laJLsvgrs&typo=1
> archives:
> 5/2017 thru present https://linkprotect.cudasvc.com/url?a=https%3a%2f%2fredfish.com%2fpipermail%2ffriam_redfish.com%2f&c=E,1,kxq5l0d1Dg0z6Fv7n7YQoD4flE-RyvIEIasuQmuZo8K1484AXPwcfYWS8zuwiiGgFcVXk_awIzd0kEQ4S422bZNMYq97Y9A4JDPCgk01DCMtJn5u&typo=1
> 1/2003 thru 6/2021  http://friam.383.s1.nabble.com/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20220211/b6b91a52/attachment.html>