[Paleopsych] Intelligent Bacteria

Eshel Ben-Jacob eshel at physics.ucsd.edu
Sat Apr 23 17:00:52 UTC 2005

Thanks for the complements and the thoughtful messages.

I do agree with Hawking. Currently we are in the process of quantifying the ideas of 

1. Latent information embedded in the complexity of the environment.

Attached is one of our paper that quantifies the notion of structural complexity and regularity.
The idea is that in daily life we intuitively mix the notions related to order vs. disorder and complexity in the sense of capacity for information. In regularity we refer to the axis from pure random sequence to purely periodic one. In terms of complexity both corresponds to the limit of vanishing complexity. Maximum complexity is usually at the crossing from disorder to order ( some refer to it as the edge of chaos) but on the regular side. So you have nested structure of variations ( variations within variations or variations on all time scales) but embedded within some level of regularity. The idea of the definition came from our study of real (not modelled) neural networks and recorded brain activity. We also learned from the networks ( and my own experience during my years in the Navy intelligence) that to extract the latent information by searching for hidden correlations one uses the following principle:

2. The correlation for correlation principle.

In analogy of using periodic signal for detecting periodic signal ( like in radio - a word you like) you have to be able to generate a spectrum of possible frequencies until you overlap with the external one. In a similar manner to identify hidden correlations you have to generate a large spectrum of possible correlations. 

Passing by I should mention that the ability to find hidden correlations is what we do in science.
To quote Einstein :

"It is a glorious feeling to recognize 
the unification of a complex of phenomena 
that appear to direct sense experience 
as completely separate things". 
(In many cases we refer to a genius as one who can reveal hidden correlations that other overlook. Well at times they cross the edge and find correlations that do not exist but this is another story.)

The next principle I proposed for the bacteria (and any other system) is:

3. The principle of matched complexity. The system needs an internal level of complexity which is sufficiently high in order to extract latent information from the external complexity.

I view this matched complexity principle the driving force of evolution that explain the ever increasing level of complexity. In a nutshell the idea is: A single bacterium needs some level of complexity to detect the complexity of the surrounding environment and over the time window between replication. To glean more information the bacteria form cooperative behaviour and generate complex colonies. However for that each individual bacterium needs a higher level of internal complexity for communication and to cope with its external environment which is now has higher level of complexity - the environment becomes both the outside and the rest of the colony. To solve the paradox self-organization leads to the formation of functional modules and spatio-temporal patterns. And than .. (for next time).

Additional essential point in my mind which is missed both in the new paper and in hawking book and most of the research on neural networks is the fact that organisms are beyond computers!. A simple statement that refers to the fact that man made computers are subject to the limitation of Godel theorem. And they are based on the idea of digital (Turing machine) computation. I suggest (as is now starts to be realised) that also the brain uses analogue computation provided by the glia cells that are 90% present of the cells in the cortex. We now know that they regulate the synaptic connections the soma excitability and correlations between the neurons by generation of chemical waves. I would argue that the distributed information processing (that includes also self-reference elements - the system changes its self according to the computation) is what sustains the cognitive functions of the brain.

This brings us back to the bacteria.

As Steve said they do a different kind of information processing (distributed) than the current models of neural networks. I would add that the brain does similar kind as well as the immune system (which is the other cognitive functioning system of our body).

There is much more but I do not have time to write it all down in papers ( we also do experiments in recording of brain activity using fMRI EEG and ECoG and it takes much effort to analyse and draw conclusions.

Happy Passover, Eshel

Eshel Ben-Jacob. 
Professor of Physics
The Maguy-Glass Professor                  
in Physics of Complex Systems      

eshel at tamar.tau.ac.il    ebenjacob at ucsd.edu
Home Page: http://star.tau.ac.il/~eshel/
Visit http://physicaplus.org.il - PhysicaPlus
the online magazine of the Israel Physical Society 

School of Physics and Astronomy           10/2004 -10/2005                                                
Tel Aviv University, 69978 Tel Aviv, Israel      Center for Theoretical Biological Physics 
Tel 972-3-640 7845/7604 (Fax) -6425787      University of California San Diego   
                                                                                La Jolla, CA 92093-0354 USA 
                                                                    Tel (office) 1-858-534 0524 (Fax) -534 7697

  ----- Original Message ----- 
  From: HowlBloom at aol.com 
  To: paleopsych at paleopsych.org 
  Sent: Saturday, April 23, 2005 8:02 AM
  Subject: Re: [Paleopsych] Intelligent Bacteria

  Eshel--Thanks for the papers.  And you're very, very right.  You didn't get the credit you deserve for your work, which in many ways is light years beyond most of the research that's cited in the World Science article on Intelligent Bacteria.

  And, Todd, many thanks for posting the article.  It supports the underlying arguments of The Lucifer Principle and Global Brain--that all of us individual social animals from bacteria to humans are modules in a collective intelligence that follows the laws of a neural net.

  The article you posted even supports the notion of "inner judges" and "self-destruct mechanisms" when it says, "some network elements can boost the strength of their own interactions."  This vaguely implies that network elements can also turn their strength down.

  It's unfortunate that Jeff Hawkins' model of the way the brain works hasn't been added to the concept of the neural net.  Hawkins says that individual modules and groupings of modules in a learning machine have to extract the repeating patterns in their environment. They have to spot repeating themes, repeating strings of signals that come in one note at a time like music.  Hawkins compares these temporal sequences, these strings of beads strung out on the thread of time, to songs.  

  When a neuron or a neural grouping gets the hang of one of these songs, it names that tune, sends the name upward to higher layers of cells, then watches out for weirdness, for signs in the stream of inputs flowing past that hint that the tune it called out was not the right one after all.  As long as the melody goes the way it should, the grouping of cells keeps quiet and lets the higher layers of cortical cells go about their business, confident that their inferiors have got a handle on the key facts of the moment.

  When the tune shows signs that it's NOT the one the lower cells named, then the mistaken cells send up distress signals and bring the higher cortical elements in to help figure out just what tune it is.  Once that puzzle is solved and the tune has been properly re-identified, the higher level cells are free to go about more lofty business--like thinking.

  A practical example.  You're laying in bed with the lights out and the window open, pondering Descartes and Pascal.  You know the room well, so there's norhing going on to distract you.  The closet door is slightly open. A gust of wind slips comes through the window. You suddenly notice a  really weird shadow moving where the shadow of the closet door should be.  But it looks nothing at all like the proper shadow of a closet door.  You're alarmed.  You drop your airy thinking and try to figure out just what in the world may be intruding on you-- a break-in artist, a Munster, a monster, or any of a dozen other frightening possibilities that flick through your brain.  Your hair stands up on the back of your neck.  You are scared witless, but you force yourself to go over to the closet door to check.  It turns out that someone .left a bathrobe on a hanger dangling from the top of the closet door and a bag you've never seen before leaning against the door's edge.  The bag and the wind-swayed bathrobe have made the shadow of a very strange creature, of a bizarre bigfoot or worse, of something you've never seen or even imagined before.

  Now that you've named that tune (bag, bathrobe, wind, and door), the lower levels of your cortex can go back to silently looking for other potential oddities, leaving the upper layers of your cortex free to agonize over how powerless mankind seems in the face of Pascal's immense, empty universe.  And other brain bits can try unsuccesfully to console you with the meager fact that you think, therefore you am.

  This picture, badly as I've put it, adds a bit more depth to the elements of the neural net that were first explicated back in the 1980s, the model I've used since 1986.  Hawkins can upgrade your view of learning machines whether you're using my quintet of learning machine elements or Klaas J. Hellingwerf's quartet of properties of a neural net.  

  What I've left out is something Hawins mentions only in passing--lateral inhibition, the competition that uptweaks some elements and down-tweaks others.  Lateral inhibition is important because it's one of the wrinkles of Hawkins' system in which I suspect the inner judges and resource shifters--the windfalls that hit those who've got a handle on the problem and the horrors that descend on those who don't get it--are hidden.

  My quintet of learning machine elements, by the way, is:

  Conformity Enforcers
  Diversity Generators
  Inner Judges
  Resource Shifters
  Intergroup Tournaments.

  Hellingwerf's four elements of a neural net are:

  multiple sub-systems that work in parallel.

  components that carry out logical operations

  auto-amplification (inner judges)



  The odd thing is that these lists of characteristic are not mutually exclusive, they're additive.  Each grabs a handful of the skin of a very big elephant.  It's an elephant I suspect Eshel has often gotten both arms at least half way around.


  In a message dated 4/22/2005 7:13:27 P.M. Pacific Standard Time, eshel at physics.ucsd.edu writes:
    Hi to all,
    The new paper in trends in microibiology is quite interesting but is limited 
    in scope (and references) it does not give a reference to our paper on 
    Bacterial intelligence published in Trends just 8 months ago. He also does 
    not give reference to any of Bassler papers.
    Attached are both papers. All the best, Eshel

  Howard Bloom
  Author of The Lucifer Principle: A Scientific Expedition Into the Forces of History and Global Brain: The Evolution of Mass Mind From The Big Bang to the 21st Century
  Visiting Scholar-Graduate Psychology Department, New York University; Core Faculty Member, The Graduate Institute
  Founder: International Paleopsychology Project; founding board member: Epic of Evolution Society; founding board member, The Darwin Project; founder: The Big Bang Tango Media Lab; member: New York Academy of Sciences, American Association for the Advancement of Science, American Psychological Society, Academy of Political Science, Human Behavior and Evolution Society, International Society for Human Ethology; advisory board member: Youthactivism.org; executive editor -- New Paradigm book series.
  For information on The International Paleopsychology Project, see: www.paleopsych.org
  for two chapters from 
  The Lucifer Principle: A Scientific Expedition Into the Forces of History, see www.howardbloom.net/lucifer
  For information on Global Brain: The Evolution of Mass Mind from the Big Bang to the 21st Century, see www.howardbloom.net

  This Mail Was Scanned By Mail-seCure System
  at the Tel-Aviv University CC.


  paleopsych mailing list
  paleopsych at paleopsych.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/paleopsych/attachments/20050423/8b1bdecd/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Functional-Holography-Published.pdf
Type: application/pdf
Size: 597169 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/paleopsych/attachments/20050423/8b1bdecd/attachment.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: CELLCOM_Hulata_ver2.pdf
Type: application/pdf
Size: 431060 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/paleopsych/attachments/20050423/8b1bdecd/attachment-0001.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Hidden-Manifolds-Warwick.pdf
Type: application/pdf
Size: 1028902 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/paleopsych/attachments/20050423/8b1bdecd/attachment-0002.pdf>

More information about the paleopsych mailing list