How I Know What Is True
Studying the Parts or the Whole
Also called: Reductionism and emergence
- Personal interest
- Philosophical question
- Working interpretation
Some systems can be understood by taking them apart. Others show their most important behavior only in how the parts relate to each other. I want to know when each kind of explanation is the useful one.
What shows up only when the parts interact?
Why it attracts me
Production systems make this tension concrete. You can study each machine carefully and still be surprised by what the whole line does.
The idea
Taking things apart is one of the most powerful habits in science. But some behavior exists only in the relationships between parts. A traffic jam is made of cars, yet the jam can drift backward down the road while every car in it moves forward. Neither view replaces the other. The skill is choosing the level that answers the question in front of you.
An example
The Availability Architecture Lab shows a line of three machines that stops whenever any one of them breaks. Its own summary is that "uptime depends on the layout, not just the machines." Adding a backup beside the middle machine helps. But the lab also lets one shared cause, like a power dip, stop both that machine and its backup at once. Then the backup helps less than the simple arithmetic, which treats each machine as independent, suggests. You cannot see that by studying each machine alone. It lives in how they depend on each other.
Where it connects
Emergence is a defining problem of Systems Whose Parts Learn and Adapt, where parts adapt to one another and reshape the whole. Complex Patterns From Simple Rules shows how simple physical laws can produce surprising complexity. And the limit I keep meeting in Reasoning From the Ground Up appears here too: whether experience itself, in The Mystery of Being Aware, is the kind of thing that parts can explain.
What this does not establish
Saying that behavior emerges from interactions describes a pattern; it does not explain it by itself. It also does not show that a mind, a society or the universe is a single conscious whole, or that any mystery is solved simply by calling it emergent.
Questions I'm still exploring
- How can I tell in advance whether a problem needs the parts view or the whole view?
- Is behavior that emerges from interactions only hard to calculate, or sometimes truly new?
- Which problems in my own life come from how things interact rather than from any single cause?
Sources and further reading
- Timothy O'Connor, "Emergent Properties", Stanford Encyclopedia of Philosophy
- P. W. Anderson, "More Is Different", Science 177 (1972): 393-396
Working interpretation: drafted from my notes and interests for review. It is not a direct quotation, and I may still change it.