Friday, 30 November 2018

Emergence and literature

I wrote this after reading "Mind and Emergence" by Philip Clayton (OUP, 2004). Though emergence may seem to have little to do with literature it ties in with some other articles I've written ("Ingarden and the Sense of Resolution", "Literary Depth", etc).

What is Emergence?

I read little philosophy. Partly this is because even if it's clearly written (and Clayton's book is) I struggle to understand it, but there are problems of motivation too. In places Clayton contrasts the scientist's and philosopher's approaches. My instincts are clearly scientific - to get more data, to produce testable hypotheses, to wonder what the point of the theorising is given that whatever the conclusion, the next step is to collect more data. Definitions are ok though. Early in the book it's suggested that there are at least 3 main features of emergence -

  • Physicalism - it begins with observables
  • Novelty - "when aggregates of material particles attain an appropriate level of organisational complexity, genuinely novel properties emerge in these complex systems"
  • Irreducibility - Emergent properties are irreducible to, and unpredictable from, the lower-level phenomena from which they emerge.

If the novel properties are too novel, there's a risk of Dualism being argued for - the two layers are so different that communication between them becomes problematic, a ghost in the machine. Evolution is important in the discussions - it explains how the emergent features we witness could have come about gradually. Examples are in fossil records and existing life-forms.

There are different types of emergence. In some disciplines over 20 distinct layers have been identified. A general theory might be difficult to produce. Complexity theory may help to explain some of the anti-entropy features. Simple rules may generate complex, emergent outcomes - "game of life" and gliders; ant behaviour, etc.


  • quantum world -> classical world
  • particles -> viscosity
  • quarks -> cells -> brains - >thought

I'd add Letters -> Words -> Characters/Narrative -> Plot -> Moral

Weak/Strong emergence

How "real" are these upper levels? Are they just a convenience, a shorthand to make explanations less tedious? Strong emergence is when "downward causation" appears, when something in a higher level seems to acquire an independent existence with control over its surroundings - including its constituent parts. "Person" is a candidate for being the result of strong emergence. If a person sees a tiger (or is injected with adrenaline) their heart-rate will rise and they'll experience fear. That sequence of events goes through the layers in a bottom-up direction - a person is a bundle of cells, so a change in the cells is likely to percolate up. However, if a person dreams about a tiger their adrenaline level might also rise as the result of downward control.

Defining the layers

When does more become different? It's not always clear. Each layer may be defined by the appropriate way of it being studied - physics is a basis for chemistry which in turn leads to biology; each defining a layer. Phenomena in a layer are described with reference to that layer, and explained by reference to the layer below. John Holland suggested that different sciences occur at jumps of roughly three orders of magnitude. The layers may not be clearly defined. For example, there's no clear living/non-living divide.

The dependence of a layer on the previous one can be of several types, part-whole being perhaps the simplest. To describe these dependencies network theory is replacing hierarchies. Within a layer there may be a Game-theory association between components. In a predator/prey relationship there may be an oscillating balance - even though the predator eats the prey, with fewer prey there'll be fewer predators. Take this dynamic element away and you fail to describe the system. A Game-theory association is irreducible.


But the way that higher levels are generated from lower ones might not be simple. Suppose you're looking at a flat hologram of a person's head. There's an illusion of depth. If you focus on the nose, you're looking though two little regions of the flat hologram (one for each of your eyes). That's where the information about the nose is. If you move your head sideways and look at the nose again, you're looking at it through two different regions, so information about the nose must be there too. Unlike a painting, the hologram's representation isn't in a 1-to-1 relationship with the represented object. Where is the nose? It's everywhere on the surface. If you remove part of the surface, the nose will still be there, degraded.

Fanciful though this concept sounds, it's a theory of the way the universe is (the Holographic principle), solving otherwise intractable problems. The 3D world we see might be a projection of a holographic surface.


From a bundle of cells and behaviours, how do you get a person? Is a person any more than a mix of habits, tendencies and memories bound in a particular body? Suppose they have a leg amputated? Are they any less a person? Suppose they have advanced Alzheimers? Suppose they become "a different person" after a stroke?

Apparently consciousness didn't exist until relatively recently. Is consciousness such a special feature though? Some computers pass the Turing test with some people. People can become very attached to a pet or even a temperamental car - memories become associated with it. Sometimes self-awareness is used as a test of advanced consciousness. That's less clear cut than it used to be - the mirror test has been called into question recently, and some people seem more aware of themselves and others than others are. Crows pass the test but not mountain gorillas - perhaps it's a feature that's more useful for community animals; perhaps it's less of a crowning glory than we thought - "In Edelman's treatment, the increasing complexity of dynamic feedback and feed-forward loops just is awareness" (p.119)

Can any system sufficiently complex to model itself be considered conscious? Not necessarily, though it's interesting to note that a simulation may well have structural similarities to the real thing. This world appears to be hierarchically structured - complex structures become units of higher structures.

Whether or not people exist, they're a useful concept. If asked why X killed Y, an answer involving quantum effects is less likely to be successful than one involving concepts like betrayal.

Belief in God could be a side-effect (an overshoot) of belief in self, God being as real or unreal as its substrate, people.


Texts have several layers - letters, words, objects/scenes/characters, plot, moral. Each layer needs the substrate (without letters, no words) but has properties that the substrate lacks (letters don't have meaning, words do; words don't have guilt, characters do). Thinking in terms of layers helps elucidate some issues -

  • Especially in avant-garde work, it can be difficult to go up to the next level - in Finnegan's Wake for example, the step from letters to words can be problematic.
  • These layers can become mixed especially in avant-garde works though also in comedy - in cartoons, Jerry rubs out Tom as if he's the cartoonist.
  • Sometimes layers are bypassed - "pure poetry" bypasses middle layers.
  • Sometimes lower layers show through - in acrostics for example, the letters matter even after they've been composed into words.

This looks like a typical setting for emergence, though I don't think the concept of layers does it full justice - there's too much two-way interaction between the layers. A network model is more appropriate.

Readers easily create people from words - its what they've spent their lives doing without realising it - which is why it's useful to break the process down. In literature though, the lowest level isn't that of letters. Beneath is the author, who can also show through, producing metafiction. People who dislike metafiction often say that they don't like how it destroys the illusion, how it prevents immersion, but how real are the characters anyway? Would they (the putative author included) pass the Turing test?

Holograms have been used as an analogy for how meaning can emerge from a surface. Removing a sentence from a text might reduce the amount of detail about a character, but the character survives. And a sentence may contain information regarding more than one character.

No comments:

Post a Comment