Thread by @evolvingcalm

@normonics' Intro to Applied Complexity #ACS101 #SpringA2021 Highlights

Session 11: The Organisation and Dynamics of Living Agents

Thread/

https://twitter.com/evolvingcalm/status/1368826090085486594

Intuitively much of the real world is alive, whatever that might mean.

The kind of complexity that we observe in living systems is really an order of magnitude (however you might want to measure that) more complex than anything we see in the 'merely' physical world

1/n

Physical systems can display non-linear behaviours, pattern forming processes, etc.

When we get to the realm of the living almost none of our reductionistic concepts are too helpful, outside of very narrow contexts

2/n

Beyond the fundamentals of complexity (self-organisation, emergence, pattern-formation, scaling etc.)
living systems provide us with the most direct empirical access to vast complexity.

3/n

Individual agents are a unit of life.

All our considerations depend on the existence of living agents that are situated in some context, on which they both depend on and can be harmed by.

4/n

Focus on the organisation of a living system and agent rather than the particular material they are made of.

What is it about their organisation/relations that compel us to call something alive?

(As so many have wondered before)

5/n

Autopoiesis: via Maturana/Varela's - 'The Tree of Knowledge'

"Living beings are characterised in that, literally, they are continually self-producing"

Avery important organisational property for a system to have.

6/n

Fractal and chaotic patterns serve functional roles in an human/animal/organism like how the surface area of lungs form, heart rate variability etc.

"Physiology may prove to be one of the richest laboratories for the study of chaos and fractals."

7/n

"In living systems the whole generates the parts. The parts do not exist a priori."

Reductionism has no good way of accounting for things like that.

thesideview.co/journal/genera…

8/n

2nd Law of Thermodynamics

There are grand claims that entropy is always increasing, the universe is heading to a heat death, things are ordered and they're going towards disorder and that's just the way it is.

9/n

What generally isn't mentioned is the 2nd Law of Thermodynamics does not apply necessarily to all systems.

It only applies to isolated systems.

10/n

Isolated Systems:

Energy or matter can't transfer, there's no interaction with systems outside of it - exchanges no matter and no energy.

(pretty rough when you find out)

But we have other kinds of systems (e.g. closed).

11/n

Open/Dissipative Systems:

The entropy of an open system, by virtue of its exchange of matter and energy with the environment, can reduce entropy locally.

(Putting aside any global considerations one way or the other).

12/n

A vortex is the prototypical kind of dissipative system.

Due to their physical energetic closure structure they persist - roughly by virtue of this closure dynamic.

Spontaneous behaviours, oscillation, chaotic behaviours, only come out of dissipative systems.

13/n

Flow energy and matter is essential to systems that become more ordered over time, or maintain their order over time.

They are not subject to the 2nd Law locally.

14/n

It's not necessarily so easy though to identify why a hurricane is a physical system - not a living system - in a crystal-clear way.

15/n

Pondering the origins of life:

Autocatalytic sets are often posited as a potential beginning of life. We assume and assert that we had an earth that originally had no life on it then at the some point somehow life emerged out of the physical and chemical dynamics.

16/n

Autocatalytic set:
Subset of chemicals and reactions that formed a closed structure with input from an external 'food source'.

Let it go and it will produce more and more over and over.

17/n

People like to think of autocatalytic sets as a potential origin of life because you don't DNA, RNA, etc. - just chemicals - and out of that you form a closed set of things that can grow themselves, reproduce themselves etc.

18/n

There's something special about closure.

(Thematically it is shows up in many contexts)

(Deterministic systems can continue on indefinitely if they have closure, if not there's something wrong.)

19/n

But autocatalytic sets are just chemicals so it's not really like living agents that we're talking about yet - just a set of chemicals and they just get amplified overtime if they have continuous access to a resource.

20/n

Autopoiesis: Self-production

The cell is posited as the archetypical autopoietic system.

The essential ingredient these have that 'just' an autocatalytic set doesn't have, is that they are also not just producing chemicals as such, but producing their own boundary.

21/n

The essential organising principle of living agents of a living unit:

Linked together recursively in such a way that the membrane that's being produced via the dynamics also bounds & limits the growth of the thing & maintains it all as a discrete unit pattern in space.

22/n

The boundary is essential:
-Differentiates between agent/environment.
-Limits the reaction.
-Maintains environmental conditions internally as separate from external.
-Ensures reactants/products are in contact with one another & this keeps happening.
-Selective Permeability.

23/n

Selective permeability:
The boundary e.g. (membrane) is both a product and participant.

If you have a boundary you might need a resource (food) but you might not want other things.

So the boundary needs to and can select what the system takes in and what it doesn't.

24/n

Other side of the coin is Allopoiesis - other production.
e.g. Ford car factory.

25/n

Autopoietic unity - closure involves generation of a membrane.

Where interdependence between system units is so strong, like a conversation (an interdependent behaviour, except conversation can end), that if both conversation members left they would disintegrate.

26/n

Are social systems autopoietic?

Are they not?

You can walk away from social structure, but could you persist?

In some ways perhaps a difficult question - not obvious.

27/n

We work with the toy examples.

We have to remind ourselves they are just that.

This is attempt to map the human metabolic pathways.

A crazy amount of processed interactions, chemical processes, etc.

Source: John Gall - Systems Bible

How do you get from something less complex to something more complex?

A challenging question but the short answer is evolution.

Despite the model the actual system is never simple in any sense.

29/n

Things need to survive if they're going to be considered life, or at least survive for sometime.

To persist they need to be viable in some sense.

30/n

Viability:

-Intrinsic viability - the structure of the autopoietic system is in fact autopoietic, it achieves closure, it's able to synthesise itself.

-Contextual viability - it could be viable in some contexts, but not necessarily in others.

31/n

Contextual viability is like a fish out of the water.

Obviously fish live in the sea, (they are very happy to do that) you take it out of the water and it's no longer viable.

32/n

Intrinsic unviability:
Doesn't have internal logic/coherence e.g. elephant allometry - diameter of legs relative to its body.

A monstrous elephant couldn't exist in any world in any environment.

It just can't exist because it doesn't have internal logical coherence.

33/n

Contextual unviability:

e.g. fish out of the water - it's just in the wrong context.

It doesn't have what it needs.

But it could, in principle, be viable.

34/n

One of the crucial roles of the membrane/boundary/autopoietic system is to have an internal environment that can be different from the external environment because a lot of the processes don't just occur no matter what - they need certain environmental conditions.

35/n

Homeostasis:
The system maintains some state or range of states inside of the system.

i.e. the environment is fluctuating so the organism is damping out the fluctuations.

36/n

Somehow homeostatic state maintenance occurs in the face of external fluctuations.

Some of it can be passive - the boundary just existing in many cases can partly achieve that.

In other cases an active process is needed that responds/adapts/reacts to environmental events.

37/n

The typical canonical example of Homeostasis is the thermostat.

Bi-directional stabilisation via negative feedback.

(Joe is designing/selling wood chopping robots. Maybe? Stay posted.)

38/n

Homeodynamics:
Keeping things in the right range or having some of the right properties dynamically.

(Maintaining appropriate internal conditions is not always so static it turns out.)

39/n

A sign of a healthy heart is that it is spontaneously changing its rate in these scale-free kind of patterns/self-similar kinds of patterns.

(Very different from the initial assumptions of homeostasis might have led you to believe.)

(Also what happened to this paper?)

40/n

Sometimes you're trying to keep things steady.

Sometimes things fluctuate & that can be a healthy sign.

Too much regularity can indicate something is wrong e.g. heart rate. If it becomes extremely stable/fluctuates very regularly - indication of impending failure.

41/n

Regularity is not always what you're after.

Often we are indeed leveraging variability and dynamics to keep things healthy.

42/n

Requisite Variety:
General idea - essential for homeostatic/control type mechanisms.

Environmental variety impinges on a system - if that needs damping out to keep essential variables/continue system viability then variables need at least = internal variety as external.

43/n

There is insufficient variety when there is some external variety that the system has no response to internally.

This is extremely obvious when you look at it and seems almost tautological but is very much ignored when you look at systems that most people generate & build.

43/n

It's hugely powerful if you keep this in mind when you're looking at the functions of systems and the failure of systems then thinking about where did this failure come from? Where do these errors come from?

Very often it's insufficient requisite variety.

44/n

One of the reasons we have so much internal complexity is because the environments we live are very complex -they have a lot of variety that we need to deal with and respond to.

45/n

If you only have one of a heating actuator or a cooling actuator you don't have enough variety to keep the room temperature between a reasonable value of too cold and too hot.

46/n

You can overshoot variety too.

The consequence of which is wasting the resources required to maintain it.

47/n

Functional properties:
What does a thing do - what's it for.

It's intuitive to us, we use things for things all the time
Almost a bit Aristotelian i.e. formal & final cause
Doesn't line up with so far discussed kinds of emergence
Far from being explainable via reductionism

48/n

Take the heart, which is a structure with some properties - physical and behavioural

Whole body is autopoietic system self-synthesising nature to the whole

For the whole body to find that autopoietic closure it needs the heart to do certain physical things.

49/n

The physical properties of the heart, even enumerated at length, by themselves don't tell you anything about what its functional properties are, and indeed alone you wouldn't necessarily know which of them are relating to the function it's fulfilling in the wider system.

50/n

We tend to talk about functional properties of the heart like it aids in delivering oxygen rich blood to all the body tissue - that's a functional property of the heart - it's fulfilling some role in a system that is in its context.

51/n

But the heart has a boundary of its own (not all the structures are within that boundary) & there's an 'inheritance' of a functional property by virtue of the context the heart sits in. We know the truth of this since you can actually go on living with an artificial heart.

52/n

An artificial heart is physically and materially very different from a natural heart - But you replace one with the other and its functional properties continue to be fulfilled, at least for the most part as we understand it, at least enough to keep you alive.

53/n

A functional property of a part of the system is not something that's *in* the part but in something about how the part & its physical properties fulfil some pattern in the larger system its in the context of.

(Note the relationship)

54/n

Relating multiple scopes:

-Scope of the heart proper - can discuss physical properties.

-Scope of the autopoietic system - can discuss properties & closure properties.

-Relationship of the scopes.

It's in this relation that the functional properties actually exist.

55/n

A functional property is somewhat strange.

The heart has a subset of physical properties & there are properties of the heart that are outside of it which is why we can totally replace heart with a totally different material system & still get the same functional results.

56/n

To call something properly functional it really needs to be inside of some context that typically has some closure property & where that piece of the system is helping to fulfil that closure property - but that closure property is not in that piece of the system itself.

57/n

You Cannot really talk about functional properties in a reductionistic way & it's interesting because biologists trip over themselves all the time trying to deal with this.

58/n

Biologists constantly refer to the functional property of a thing & then because they are interested in maintaining their reductionistic assumptions they say "it's an epiphenomenon", "it's an illusion", "it looks like a thing but it's not really a thing", etc.

59/n

As soon as you take a non-reductionistic approach all that is easily resolved - a functional property is a thing that lives in the relations of things.

Relations produce properties emergently - so it's a kind of emergent property that is based in the context it sits in.

60/n

Organisms need be robust so we often have functional redundancy e.g. a pair of kidneys.

If one goes you still have enough kidney to keep going.

61/n

There are microgranularities of redundancy - any given tissue is made up of many cells that are more/less the same.

If any single cell in your body dies no problem - there's no single cell in your body you need that multiple other pieces can't instead fulfil the function.

62/n

Functional Degeneracy: (I know a few people like this)

Very common in organisms - not as common in machines (although it does exist)

The idea that a given structure or set of structures can perform multiple functions depending on how they're relating to the context.

63/n

Functional Degeneracy Example:
Mouth, lips, tongue, teeth, palate, etc.

Is the mouth for eating? singing? talking? kissing?

Same structure, different context, different function.

64/n

Anticipation:
Organisms are highly anticipatory (different than prediction)

Acting in such a way that you're expecting a future.

A kind of causal entanglement across durations of time.
Can't be dealt with reductively - only looked at in terms of whole processes over time.
65/n

Anticipation Example:
Functional properties when they do their thing - when the heart squeezes and pumps blood - the blood that it delivers to the tissues hasn't gotten there yet.

When the heart contracts the blood will get to the tissues in the future.

66/n

Typical causality is generally thought of as past to future chain but with functional behaviours things are happening now because of what they cause to happen in the future.

67/n

We have these loopy closure structures such that the system and its structure persist overtime (and only by virtue of that closure).

Related to at least one of Aristotle's breakdown of causality

68/n

Of course in living systems sometimes it's not always clear how to get to the end - so the system is always actively changing what it's doing in order to get to some end.

(example: locking yourself out of the house)

69/n

Patterns not only have a scope in space but a scope in time.

You can indeed have causal structures that are bound up where in some sense future states of the system are effecting what's going on now.

Living systems - highly weird and non reductionistic.

70/n

Complex 'Adaptive' Systems:

Living systems actually adjust what they're doing in order to achieve future states.

You might call that adaption - which can be reactive but it also can be anticipatory.

71/n

But adapting to what? - adapting for what?

You could have a complex system that couldn't in any sense could be said to be adapting.

You need these functional & higher order type properties like autopoietic closure to even begin to discuss adapting.

72/n

You need goals and future states of the system to somehow be relevant for adaption to make any sense at all.

73/n

'Doing the same thing twice & expecting the same result is insanity'

(as Joe likes to say)

74/n

Inverse of the usual saying - which only makes sense if you're in a totally dead world.

When you're in a living world the system is never the same twice,

or rarely it is.

You can't expect it to be.

75/n

Another way to think about this is you can't expect two people to react the same to any given environmental perturbation, any given input, any given stimulus.

Complex agents in a complex system.

There's no fungibility of items for the "same class".

76/n

Whereas when we're talking about things like atoms etc. we expect them to pretty much behave the same - this hydrogen atom is interchangeable with that hydrogen atom, they're fungible there's not really much difference & once we get to living systems it's not true anymore.

77/n

We have this idea of classes of things - this is a person, - that's a person, but they don't even do the same thing when they're presented with the same thing, so we always have to keep that in mind.

Even the same person is not the same person at two moments in time.

78/n

A good example of doing the same thing twice and expecting the same result is insanity is telling a joke,

maybe it's a good joke, maybe they'll laugh,

then telling it again,

79/n

A good example of doing the same thing twice and expecting the same result is insanity is telling a joke,

maybe it's a good joke, maybe they'll laugh,

then telling it again,

79/n

(Joe's low entropy/low surprisal event)

(A bit like Déjà vu)

Constraints in Possibility Space:
(Joe describes his fantasy?)

If you take @HundredthIdiot and put him in a blender - Tim is no more.

80/n

In order to maintain closure properties relative to the total possibility space of the system we live in a very small constrained possibility space.

That's what these ideas are all about.

81/n

How to think about it in terms of the ensemble/the variety/the entropy - we have to maintain these very narrow bands of possibility space for structures to maintain themselves.

82/n

When we talk about closure structures, in say the body, these are nervous system, muscular system, immune system and they all depend on one another and if you exit this narrow band of possibility space you destroy this closure property.

83/n

Boundary vs Constraint:

A boundary is a spatial pattern.

A constraint is an information theoretic indication of a narrowed possibility space.

(Think about relating this to an example of two people carrying a couch down some stairs.)

84/n

Physical boundaries can serve as a constraint but so can nonphysical things like communication.

85/n

There's a lot of functional degeneracy when it comes to the boundaries of living systems.

There are a bunch of different functions that a boundary can help fulfil, and indeed, those can be articulated as constraints.

86/n

Multiscale Requisite Variety:
multiscale variety + requisite variety

Red: Independent
Blue: Coherent
Green: Multiscale (sometimes aka correlated)

87/n

Depending on how an environment is structured an independent system has enough variety to deal with small perturbations i.e. high complexity fine-granularity perturbations, but if a large scale event came along it can't do anything, it can't push back.

88/n

Vice versa for coherent behaviour - some large-scale event comes along & it might be able to deal with that, but it doesn't have the fine-grained complexity to deal with variety & small scales.

89/n

Something in-between coherent and independent might have enough in both.

Think of an environmental variety curve - to maintain viability system variety has to be greater than or equal to the variety at all scales of the environmental variety.

90/n

Generalisation of requisite variety to multiscale situation:

Organism are a great example
- You have the muscular system - you can punch a guy if you need to.
- You have the immune system, you can deal with viruses coursing through your veins.

You can't punch a virus.

91/n

You need not only the variety to match, but the scale and the variety to match, to be able to deal with/maintain the integrity of a system.

92/n

The immune system isn't fixed - it learns about new invaders.

Your muscles aren't fixed - you can lift weights, get stronger

We always have uncertainty about the environment,

different degrees of uncertainty.

but always uncertainty.

93/n

Biological systems are always set up to learn by probing and prodding the environment to find out what's actually going on out there.

You don't know a priori.

94/n

Evolution is an even more extreme pattern for discovering what the structure of the environment is.

The variety and the scales of variety are themselves dynamic.

95/n

But even at the level of the organism we don't know exactly what the structure of the environment is - how much variety do I need? at which scales do I need it?

Depends on the perturbations.

96/n

We probe the environment via stressors.

(a kind of antifragility if you will)

97/n

Meta-variety: Adaption

Maybe you're at one stage & you lift some weights - add a little bit in scale.

Literally when you lift weight you gain mass & part of the reason you gain mass is because of this issue of not wanting to suffer a scale/complexity profile trade-off.

98/n

I don't want to have to suffer a trade-off where I gain some strength, some muscle, and now I need have less of an immune system - less variety.

99/n

There's a very good reason you actually gain mass to get stronger - you're increasing the scale at which you behave without necessarily decreasing the complexity of your fine-grained responses.

100/n

The organism, the body, probes the environment via stressors - a kind of antifragility if you will.

101/n

Organisms come from somewhere - they have to develop - typically from a single cell.

102/n

A few things to note:
-the parts are differentiating internally from a whole autopoietic system, a whole system from the beginning.

Cell prototypically autopoietic, from the system -> parts, not coming together from the outside

Also leveraging a pattern forming process.

103/n

From a cell to a full organism this is a system growing in MASS - i.e. non stationary complexity profile, breaking trade off constraint.

104/n

Compare the mechanical e.g. engine.

A whole from parts

Not only shows you what the parts are & how they fit together also how you would construct it, each part produced then put together.

Different from system internally differentiating into parts as it forms & grows.

105/n

One of the only ways living systems achieve this is by leveraging pattern-forming processes of fractals, long-range inhibition close-range activation etc.

They use these to generate them in the right places such that they fulfil functional roles in the autopoietic system.

106/n

A pattern produced in the context of an organism can achieve functional properties, both scale free patterns and those with a characteristic scale - scale bound
Cheetah spots, Zebra stripes, lining of the intestine, lungs.

107/n

Lungs have huge area for gas exchange - size of a tennis court in surface area - from fractals formed by some recursive process that branch and branch and branch and so forth.

Patterns are easy to latch on to like this - they exist in nature almost like physics.

108/n

This is the linkage between these non-functional purely physical properties and then how organisms deploy these patterns in bounded context to achieve some functionality.

109/n

Something like camouflage is afunctional role for something that when outside of its particular situation isn't functional and instead 'just is'.

110/n

DNA is not a blueprint.

It carries information and standard structure of these varying base pairs to do so. Not a blueprint but contains switches that, at the right times, & the right contexts, get switched on so these kinds of processes can unfold.

111/n

For evolution if you only lean on natural selection it seems to imply there's a kind of arbitrariness - any structure or pattern is possible if only you encoded it into DNA & that is not necessarily the case.

112/n

You need that these patterns can form in a distributed & decentralised way somehow for them to be viable as component of a larger system that is the organism.

113/n

Stu Kauffman talks a lot about this in terms of we not only need a kind of variety & selection for evolution but we need these self-organising systems to be selected amongst & to produce this variety.

114/n

So it's not arbitrary just like the size of the elephant isn't arbitrary, you can't get bigger, there's internal constraints by virtue of the kinds of process you need to form these things, that's a very small part of the huge possibility space that you could even think of.
115/n

When we're talking about units we're talking about biological organisms as these discrete units. One of the things we really mean is their behaviours/decisions are at that level - of the unit - there's a global coordination of their states that they're acting as one.

116/n

A chair has a massive rotational constraint as a unit.

There is a similar kind of high order constraint on the unit of an organism - half of me can't go downstairs while the other half stays up.

It's a very strong constraint on the unit that we consider the organism.

117/n

If we think about the complexity profile, the sort of multiscale variety of a developing organism, the amount of mass the organism gains as it goes from a single cell to 80kg adult - that's huge orders & orders of magnitude of mass.

118/n

So as we think about that in terms of complexity profile that's an important thing to take into account:

There's not 'fixed in mass', there's not 'fixed in scope'.

119/n

Failure mode:

-depend on redundant structure, else:

-global failure cascade -> death

120/n

Sometimes there's a fill in - another kidney - pretty much the same structure or for instance the brain after damage - other parts of the brain can help to fulfil that function without being exactly the same as that original piece was.

121/n

Organisms are almost magical in the way that they figure out how to fulfil that function in that kind of final causality to put the right pieces together to achieve that function.

122/n

Or imagine you're out jogging in the wood if you break your ankle real bad and you can't use that leg but you manage to get back home anyway not because you're ever 'walked' that way before but because you're fulfilling this kind of final causal thing.

123/n

So that's one kind of failure mode, failure mode with persistence of being alive and maybe achieving some final that you're after.

124/n

So then, how do organisms die?

They die all at once (essentially).

125/n

There's a global failure cascade that's another consequence of the kind of autopoietic organisation that we're looking at.

Everything's strongly connected together and as one thing goes the next thing does.

126/n

It's just important to keep that in mind because often we think of organisms as maybe how we want certain systems to be kind of 'like an organism' but you also then have to balance with its failure modes.

127/n

Autopoiesis is a really important way of thinking about the organisation of a living system.

Fractals and chaos serve functional properties.

Look at the dynamic development of the system and consider these two disparate archetypes - whole from parts & parts from whole

128/n

That's kind of how we're structured.

We're not meant to last forever.

For the global system it's fine for us to go as individuals & that's sort of the logic of the system.

Sad but true.

Just life.

129/end

https://twitter.com/evolvingcalm/status/1408269888930516996?s=20