Distributed functional decomposition

A summary of Chapter 1, Section 4 of Andy Clark’s Supersizing the mind

1.4. Distributed Functional Decomposition

The author introduces distributed functional decomposition as a way of explaining (or a perspective or approach to) systems exhibiting nontrivial causal spread and using the principle of ecological assembly, in order to explain and understand them better.

This approach views systems in terms of the flow and transformation of four things:

• Energy
• Information
• Control
• Representation (only where appropriate)

“Functional” here means that, in understanding the overall system, what matters is the role (function) each element performs, rather than how each element is implemented. In programming terms, each element’s responsibility or interface does more explanatory work than its implementation.

For example, pairing the elements of the block-copying task described in Active sensing with their functional roles gives:

• The world in front of you: memory
• Eye movements: retrieving specific information from memory

In another example, the passive walker introduced in A walk on the wild side, the structure of the body performs the computation needed for walking (morphological computation).

Explanations based on distributed functional decomposition may provide a better understanding of the overall system than traditional approaches that assume, for example, that memory occurs only inside the head and the skeleton merely supports the body.