emergence

Some things are greater than the sum of their parts.

Comprehending emergent phenomena can be difficult ­because we have been trained to analyze most systems, including complex ones, by taking them apart (methodological reductionism).

The prob­lem is that most emergent phenomena dis­appear when they are taken apart. How then can we understand them? We must learn to shift our focus away from the parts themselves and onto the interactions between the parts. Sometimes these interactions do not even depend on the types of parts that are pre­sent. In describing emergent phenomena, we will have to address some issues that are essentially philosophical in nature.

Emergence is not something magical that jumps into being. Programs that we run on computers are clearly emergent too (just transistors, just atoms and molecules, …, high level languages), we just don’t tend to think about it like that, since we are able to wrap our minds around it.

Transclude of The-Pretense-of-Knowledge---On-the-insidious-presumptions-of-Artificial-Intelligence#^96eda5

Discretization & Separating the parts from the whole (also called methodological reductionism - quite successful in neuroscience (and ML) thus far, but approaching its limits - though advances in technology keep its “death” at bay).

Without a clear goal or definition of intelligence, the field of AI got split into discrete sub-fields (CV, NLP, …), with a 0.5% test set improvement as the goal, not bringing the fields together whatsoever.
Similarily with neuroscience, where there are vast amounts of observational data, but parts of the brain are studied in full isolation etc. and there have been little attempts to try to put it all together and craft theories for the observations (Jeff Hawkins explained this at his first lex appearnce).

Link to original

A wave in a football stadium is an emergent phenomenon and a case against reductionism.

We cannot really explain it by looking at individual people or a small group of them. It takes a bunch of people to do a wave in the first place. The wave moves faster than any individual could, is larger than its constituent parts and has other properties like changing direction, reflecting on boundaries, …
We could replace the football fans to be english or spanish or possibly even well-trained dogs. We could change the shape of the stadium to be triangular, …
→ The lower scales do not always drive the higher scales.
If a person stands up due to the wave, spilling a beer, a higher scale interacts with a lower one.
→ Immaterial things can affect material things.
(Might the likelihood of impact from one scale to another decrease acc. to some power law?)

Properties of emergent phenomena.

→ ​Arise at a scale larger than their smaller components
→ ​Depend on the interactions of their smaller components
→ ​Have properties that differ from t­hose of their smaller components
→ ​Have properties that are not easily deduced from knowledge of their smaller components
→ ​Are in­de­pen­dent of many features of their smaller components
→ ​May causally influence their smaller components (downward causation)
→ ​Are without mass, yet through their arrangement can influence t­hings with mass

Emergent phenomena in the brain.

• All kinds of different waves, some only dring development e.g. through the retina, etc. (influenced by E/I balance)
• pulse like patterns (more compact and concentrated in their width, annihilate when colliding)
• synchrony
• repeating (sometimes precisely timed) spike patterns or local field potential activity

How to study emergent phenomena.

Observe as many interacting parts as possible. … because they are often produced by the interactions of many constituent parts
Use appropriate spatiotemporal sampling. … i.e. at the natural scales of their interaction. Methods that subsample in space may fail to rec­ord from synaptically connected neurons, and methods that subsample in time may artificially lump together multiple inde­pen­dent events that are only approximately coincident
Identify connectivity. Because emergent phenomena depend on the interactions of
constituent parts, it is impor­tant to identify the pattern of connectivity among the parts.Which neurons are nearest neighbors from the perspective of propagating activity? Which neurons are interacting?
Manipulate interactions. ­Because t­hese phenomena emerge as a result of how their
constituent parts interact, it is impor­tant to manipulate the interactions, if pos­si­ble. With a behaving animal, dif­fer­ent states like attention and sleep can alter interactions between neurons.
Make a phase diagram. The results of such manipulations will typically produce changes in characteristic be­hav­iors (e.g., oscillations versus no oscillations, ordered activity versus no activity, amplified versus damped activity). Special attention should be paid to bound­aries between phases, as the transition region often harbors the most subjectively complex and in­ter­est­ing forms of activity.
Observe dynamics in the phase diagram. Does it usually reside near the phase transition, or is it consistently deep in one of the phases? It is useful to identify t­hings that move the system around in the phase diagram. For example, does it move from one phase to another as the organism transitions from sleep to wakefulness? When the organism is in focused attention, does it move more ­toward an ordered phase? Does the system show homeostasis? When it is perturbed from a location in phase space, does it return ­there ­after some time? If so, is such a return merely the result of drift, or are ­there mechanisms actively moving it t­here? Is the system operating to consistently return to some set point?
Look for ­causes at multiple scales. Conclusions drawn from manipulations to constituent parts should be interpreted carefully and not overstated. For example, an experiment may show that REM sleep is eliminated a­ fter a gene is knocked out. This tells us that the gene is somehow involved in REM sleep, or supports it, but it does not tell us that this gene, by itself, c­ auses REM sleep and is the sole explanation for it. This is b­ ecause REM sleep is an emergent phenomenon, produced by the way in which many neurons interact. A full understanding of REM sleep would be expected to involve many dif­fer­ent levels. Any explanation that stops at the smallest scale is likely to be incomplete.

scaling intellgence / cognitive lightcone.

You need to have policies and mechanisms for enabling competent individuals to merge together into a larger emergent individual, a new self with memories, goals, preferences, a bigger cognitive light cone.
→ The higher level distorts the option space for the lower levels. They are still performing their lower level actions, but with coherence with respect to higher level goals.
bio-electricity is an enabler for scaling the cognitive light cone and intelligence by enabling cells to be part of a network.

Link to original

Levin says: emergence might just mean surprised observer.

It’s a relative measure of how much extra the system is doing w.r.t. the beliefs based on your current understanding of the system, according to michael levin.

Life / intelligence is an emergent property of the complex organisation of matter.

Link to original