Brains, Evolution, Computers

1
Brains, Evolution, Computers Companies

• Correlative mappings of decision making and
  learning systems
• by
• Ed Lee

2
Objectives of talk

• Demonstrate how some decision making and learning
  in companies correlates to similar processes in
  the brain
• Present a useful tool for analyzing our own
  decision making and learning attitudes

3
Introductory remarks

• Researching a book Plantations in the Rain
  Forest the future of civilization a study of
  human systems in the biosphere
• Fundamental questions of book
• How has nature generated robust life for 4
  billion years?
• What can we do to increase civilizations
  robustness?
• Problem author is an engineer, businessman doing
  a cram course on nature

4
Overview of talk
Artifact
Organism
5
Strata of Evolution
6
Evolution as seen by an engineer

• Localized order (structure) emerges from global
  randomness
• Local negative entropy
• Extraction, construction processes
• Strata with bi-directional stochastic coupling
• Catalysts and enzymes
• Dynamic equilibriums
• Overproduction and vigorous pruning (opposites)
  keep each other robust

7
Evolution extracted complex brains
8
Uncertainty is a key tool

• Weak bonds, stochastic links, ensembles of crude
  cues
• Mutations
• Plasticity
• Produces diverse, niche solutions
• Robust response for unforeseen global changes
• Stabilizes dynamic systems
• Friction in mechanical systems
• Trading costs, varied beliefs in stock markets

9
Ambivalent attitudes

• Can be threatening
• Error, noise, turbulence, chaos, illness
• Hierarchies formed to control, eliminate them
• Essential to playfulness
• Games
• Humour
• Critical to communication

10
How old are you?

• An incomplete question
• To nearest year culturally implicit
• Uncertain answer enabled a quick response
• Decoupled most contexts
• Short coding of question and answer
• Probably a maximum
• Value/effort
• Value/time
• Other extreme Descartes

11
Which attitude fits you?

• Ready, fire, aim!
• Ready, aim, fire!
• Ready, aim, dont fire unless certain of
  bulls eye!
• You made me miss!
• Choice determines
• Time to respond to a stimulus (sense of urgency)
• Probable accuracy of response
• Number of stimuli responded to
• Rate and nature of learning/change
• Which choice fits evolution? Brains?

12
Attitudes of decision makers
13
Companies are organisms

• Life Cycles
• Embryos (startups)
• Growth and specialization
• Maturity
• Senility and death
• Metabolic requirements
• Profits measure input/output efficiencies
• Can reproduce
• (sexual, asexual, cloning)
• Living community (flesh and structure)
• People (employees, customers, investors, etc.)
• Methods (maps, procedures, norms, policies)
• Materials (money, equipment, facilities, products)

14
Companies exercise brain-like functions

• Mildly intelligent
• Learn, remember
• Experiences
• Simple to moderately complex algorithms
• Store in locally meaningful maps
• Layers of decision making elements
• Complexity, cycle times, number of similar
  decisions per year
• Differing cue sets

15
Theoretically hierarchical
16
Typical Organization Chart
17
Bottlenecks in hierarchies
In hierarchies the decision making bottleneck is
always at the top
18
Actually Stratified
19
Time scales of decisions for strata
Short term decisions from experience, internal
processing, predictable results Long term
observe competitors choices, uncertain results
20
Strategic CEO Functions

• Select key executives
• Sponsor them
• Lead executive team
• Maintain cohesive/timely strategic vision
• Maintain flexibility in changing market
• Set tone, spirit by example
• Resolve intrinsic strategic conflicts
• Select, lead key stakeholders
• Board
• (Results of efforts affect health 2-3 yrs
  later)

21
Distribution of decision making attitudes

• Young, small companies
• Adventurer at top
• Craftspeople dominate lower strata
• Quickly adapt to market
• Old, large companies
• Craftsperson or Bureaucrat at top
• Bureaucrats dominate middle management
• Hierarchical processes
• Expect market to adapt to them

22
Dorsal view of company
Anterior
Posterior
23
Learning and Memory

• Local maps specialized by function
• Working memory people
• Short term notes, redlines
• Long term formal documents, data bases
• Some long term information received from other
  areas within company converted to local maps

24
Thermal maps
Launch new design project
Launch new product
25
Marketing Funnel extracting customers from
environment
26
Computers artifacts of Reason

• Artifacts of 19th century belief in a clockwork
  universe
• Rigid hierarchical control
• Synchronous
• Centralized decision making
• Deterministic

27
Characteristics

• Finite states
• Deterministic, pig-headed
• make mistakes, but never learn
• Fast, 109 ops/s
• 2 x106 ops during an action potential
• Extremely complex algorithms
• Fragile, tolerates
• lt10-17 bit errors/ s
• No connection errors

28
Computers hierarchical Architecture
29
Architecture

• CPU rich in logic, only element capable of
  reading maps, implementing algorithms
• CPU controls all timing and relationshipsthe
  ultimate micro-manager
• Memory stores patterns that have no intrinsic
  meaning.
• Only meaningful to CPU provided it keeps track of
  storage locations relative to program
• String of patterns, one degree of associative
  freedom
• Bottleneck is in transporting codes in and out of
  the CPU (von Neumann Bottleneck)
• Thermal maps dont change for novel or familiar
  tasks

30
Some conclusions

• Companies have some useful correlations to Brains
• extract and process the familiar
• ignore or adapt to the unfamiliar
• selectively learn
• stratified, stochastic
• bi-directional influences
• time scales and complexities
• Computers dont correlate with brainsbut do
  correlate with some rational beliefs
• process the specified
• ignore or crash from the unspecified
• - never learn
• deterministic hierarchy
• bottom-up data
• top-down control

31
Conclusion

• A major difference between biological systems and
  computers is the role of uncertainty.

32
Thanks

• To Susumu Tonegawa and Bob Silvey for the
  opportunity to be here
• To Matt Wilson and Morgan Sheng for helpful
  feedback
• To Jeffrey Goodman for his repeated help and some
  great laughs.

33
For more information

• To download copies of this presentation and
  related management essays go to www.elew.com