causal clusters

Causal clusters

(Note - I've sent this to Francis Heylighen's gbrain listserv as well, so for those 
who are members of both groups, please excuse the duplication).

Bill Hibbard, a member of both this list and the Psyche-D list, suggested in an 
earlier email that I think more about the nature of information exchange between 
interacting components, as I had posited in my draft paper "Global Consciousness - 
More than a Metaphor" that "information exchanges" are common fundamental features of 
neural networks and human interactivity. Bill had raised the notion of the individual 
human capacity to solve what he calls the credit assignment problem (acting now for 
rewards later), and how this capacity may be observed in larger groups (this doesn't 
do justice to the concept, I know, but I'm attempting to paraphrase briefly), 
suggesting I consider this concept in developing notions of information exchanges in 
neural and human complex systems.

But in trying to clarify what I mean when I say that neuronal interaction and human 
interaction are both systems of information exchange, I have been developing an 
approach substantially different from Bill's, as it is based on actual physical 
information in the sense of physical causes and effects.  The brain is a system of 
neurons which physically respond in highly complex ways to electro-chemical 
interchanges. Similarly a human system is one consisting of complex individual and 
group physical causes and effects. The information exchange is that which causes 
actual physical responses in individuals and groups and which ultimately gives rise to 
self-organizing emergent phenomena.

The following is an overview of my current thoughts on "causal clusters" and "causal 
networks". I tentatively suggest that the basic features of such networks are similar 
to those observed in small-worlds networks as described by Barabasi, Watts and 
Strogatz, and others.

A few basic premises:

Premise 1: object A physically contacting object B can cause object B to move in a 
particular direction (and A to move in another direction too) -- the dynamics of 
motion of each object are governed by laws of physics;

Premise 2: person A can cause person B to move physically without actually contacting 
person B - such motions are governed by social rules of conduct (not physical laws), 
basic evolutionary drives, and basic survival instincts (such as avoiding walking into 
another person on a crowded street).

Premise 3: person A can cause person B to move physically by communication, whether by 
vocal communications or gestures; i.e. B may respond in some physical way to A's 
communication

Premise 4: if person A causes person B to move physically through any of the above 
means, B's movement may in turn cause person C to move, or respond in some physical 
fashion. Let A's indirect cause of C's movement be an "order 2 cause" (and A's cause 
of B's movement to be an "order 1 cause");

Premise 5: person A may indirectly cause physical responses to an order somewhere 
beyond an order 2 cause (i.e. orders 3 and above), but beyond an order 2 cause, the 
causal proximity diminishes rapidly and it becomes very difficult to establish a 
relationship between A and some action farther down the line of orders.

Premise 6: in spite of premise 5, A's actions have some degree of influence, albeit 
remote, on actions at orders greater than an order 2 cause, the degree of influence is 
just very difficult to quantify.

Causal clusters


A causal cluster involves a group of objects --in this discussion, people--who are 
involved in oscillating sequences of physical interactions; i.e. according to premise 
3 when one vocalizes a particular statement, that statement causes both an actual 
physical reponse in the listeners' brains, and a manifestation of their responses by 
vocal and gestural responses. Two people engaged in conversation rapidly shift between 
causing a response and responding -- this is what I mean by "oscillating sequences".

Scenario


Consider the following scenario: A, B, and C stand in a corner at a party. A makes a 
statement to B, causing B to respond. A and B are at an order 1 causal relationship 
(i.e. each one directly causes responses in the other). C also chooses to respond to A 
so at this moment A and C are in an order 1 causal relationship. Say that A's comment 
triggers B to say something to C, so B and C are now in an order 1 causal 
relationship, but A is now at an order 2 causal relationship to B and C's 
conversation. There may be continual oscillations between order 1 and order 2 
relationships among these three.

In another corner of the room D, E, and F, are engaging in similar oscillations 
between order 1 and order 2 causes and effects. In a burst of spontenaity, B shouts 
over to group D, E, and F, causing all three of them to look over. At this point B has 
an order one relationship with each of D, E, and F, and A, who triggered B's outburst, 
has an order 2 relationship to D, E, and F. Meanwhile, F's cellphone rings. F answers 
her phone and engages in an order 1 relationship with her friend G in a restaurant 
somewhere downtown. G is engaged in a similar set of oscillating causal orders with 
her friends H, I, and J, in the restaurant downtown.

F's cellphone call is thus a long-distance link between causal clusters (as was the 
brief exchange between groups ABC and DEF) and at this point the cluster may be 
thought of as a singular unit, such that causal cluster DEF has an order 1 
relationship with causal cluster GHIJ, because the intrusion has been a direct 
influence on the overall dynamics of cluster GHIJ.

In our currently highly connected world, via one of these long-distance links, my 
actions say here in Victoria, Canada, may have a direct causal (order 1) effect on 
someone who happens to be emailing me at the time, for example. One can imagine 
situations where, for example someone is speaking on a car phone to me here in 
Victoria, when the fellow in the car gets distracted by something I say (no doubt some 
sort of nonsense!) and thereby causes an accident (an order 2 effect).  In this 
scenario the fellow in the car and the drivers he affects by his irregular driving 
behaviour (resulting from his conversation with me) are part of the causal cluster.

I suggest tentatively that these causal clusters are more closely analogous to neurons 
than are individual humans, and it is this sort of information we are looking at when 
we look for the patterns of self-organizing phenomena which emerges from a currently 
very highly interconnected network of causal clusters.  And here, just as Watts and 
Strogatz demonstrated with other sorts of networks, the long-distance networks are 
what tightens the whole human network into a highly dynamic, continuously shifting 
network of causes and effects.  Thus a third dimension or layer, if you will, to any 
self-organizing patterns emerge.

What I mean by this "extra-dimension" or layer is that the self-organizing phenomena 
are no longer easily observable.  For example with an ant colony or a sheep herd, we 
can look from above to observe directly the actual self-organizing phenomena which 
results from the interactions of the individual ants or sheep. In the case of causal 
clusters highly interconnected via long distance connections, we are unable to observe 
in this fashion the precise nature of any self-organizing phenomena which arises: 
hence the added dimension to the self-organizing phenomenon.

Again, I tentatively hypothesize that these dynamic causal clusters manifest typical 
network features, such as hubs (the clusters themselves) or a power law,  (many small 
clusters and fewer larger clusters); and exhibit similar self-organizing patterns to 
what we see in the human brain.

This is an outline, and as these ideas are currently only in the throes of 
development, I will no doubt be making significant adjustments as I learn more (or 
even perhaps scrapping it altogether if necessary).

I would be grateful to anyone who can steer me to some relevant resourses or otherwise 
comment.

Hugh Trenchard, Victoria, Canada