Title: Studying the Ecology of Globalized Societies
1Studying the Ecology of Globalized Societies
René te Boekhorst Adaptive Systems Research
Group Dept. of Computer Science University of
Hertfordshire U.K
r.teboekhorst_at_herts.ac.uk
2Theme Modelwork on the effect of the Local
environment on social interactions/ social
structure
3Meta systems entities can be systems
(communities) themselves
4The IT Designers Dream
All People Happily Connected Absorbing
Information Knowledge To All! No More Drudge
Fraternité
Equalité
Liberté
5- Slavish attitude of governments towards
business and
e-capitalism
Destruction of Environment
Global Warming Biotech
Degradation of Biodiversity
- Domination of Multi-Nationals (e-mperialism)
Social Issue
6Opposing Views
Scientific Approach to a Social/Political Issue
7This Talk
Integrated Project for FrameWork VI
Suggestions for applying those ecological
models to
Pattern Formation Connectedness of
Participating Human Communities
Meta systems
8Theoretical Ecology
Mass Interaction Models
Czárán, T. (1998) Spatiotemporal Models of
Population and Community Dynamics.
Chapman Hall (London)
9- What are their merits/flaws ?
-
- Can they be used to study social structure ?
-
-
- in particular networks of human communities
Beyond modelling Experiments with robots
Examples from Primate studies
Proposals
10Mass-Interaction Models
Basic object is the population homogeneous
mass instead of individuals
Analogous to reaction kinetics (mass action)
of a well mixed chemical system
Model formalism set of coupled (non linear )
linear, (partial) ODEs Dynamical Systems Approach
11- (limited) Population Growth
Logistic Equation (continuous) Asymptotic density
Carrying Capacity
Instantaneous rate of increase r m
Net birth rate
Net migration rate
12- Interactions between Populations
Lotka Volterra Equations (Un-)stable equilibria
Competition
isoclines
Exclusion
13Multi-Species (Community) Models Stability and
Complexity
Coexistence of 2 competing species
(communities)
parameters must satisfy 2
inequalities
(r/k S AND R/K s)
Strobeck (1973) n competing species
(communities)
2(n-1) inequalities
Chance of coexistence for large n is small
(when parameters have randomly chosen values)
14If number of species (m) OR connectance C
increases average strength of interactions S
Stability decreases
when interactions are random
May, R.M (1973) Stability and Complexity in
Model Ecosystems. Princeton Univ. Press
15Systematic Connectivity
Maynard Smith, J. (1974) Models in Ecology.
Cambridge Univ. Press
Same conclusions
16From Communities Down To Groups
Dynamics of Group Size
Within-Group Competition (WGC)
Net birth rate
Net migration rate
Instantaneous rate of increase r m
Between-Group Competition (BGC)
Larger group displaces smaller group
Members defect smaller group
Dominants should relax WGC Egalitarian instead
of Despotic societies
van Hooff, J. A. R. A. M C. P. van Schaik
(1992). In Coalitions and Alliances in Humans
and other Animals. Oxford University Press.
17Claim of van Hooff van Schaik DOES NOT hold!
Smaller groups can come into equilibrium and
hence coexist with larger ones
k
te Boekhorst, I.J.A. Hemelrijk, C. K. (2000).
In Dynamics of Human and Primate Societies
Agent-Based Modelling of Social and Spatial
Processes. Santa Fe Institute Studies in the
Sciences of Complexity. Oxford University Press.
18AND MUCH MORE POSSIBILITIES
- Dynamical systems model for sub-group formation
(in prep) - (following a charismatic leader)
- Homeomorphism
- Same equations model dynamics of a general
motivational - system based on allocation of (physiological)
energy -
Can possibly be realized in hardware Brings
together concepts from various ethological
models Rich behaviour dynamical threshold
hysteresis
excitable (viz. neuronal dynamics)
19- Problem Mass Interaction Models
- Not based on what individuals actually do
- Quickly analytically intractable
- Strategic models
- deep(er) theoretic insight hard to back
up by - empirical data
20Object oriented (neighbourhood -) Models
Two individuals cannot be present on the same
place simultaneously and that matters!
- Explicitly spatial and local
- sites Cellular
Automata - entities
Individual Oriented Models - Simulation instead of Analysis
Agents
EXAMPLE Socio-spatial structuring in primates
te Boekhorst Hogeweg CHIMPs
and ORANGs Hemelrijk
Dominance structures in Macaques
21Select Random Partner
DOMINANCE INTERACTION
Move On
Ego Wins
Ego Looses
OTHERS
Flee From Opponent
Move To Other
Goto Opponent
Turn SEARCH ANGLE At Random To Left or Right
Behavioural rules of the entities in Hemelrijks
models. Execution of acts depends on whether
threshold values of three parameters are
surpassed that symbolize critical distances
Personal Space (PERSSPACE)
2212
10
8
DomValue
6
Dominance Hierarchy
4
2
n
0
time
Socio-Spatial Distribution
te Boekhorst, I.J.A. Hemelrijk, C. K. (2000).
In Dynamics of Human and Primate Societies
Agent-Based Modelling of Social and Spatial
Processes. Santa Fe Institute Studies in the
Sciences of Complexity. Oxford University Press.
23Problem Individual-based Models
Its all in silica
How Real are the Rules?
No real-world constraints
24Cognitive Science
- Classical A.I.
- Computer/Programming metaphors
- Cognition as Computation
- Behavior as Problem Solving
-
Algorithmic Disembodied Non-situated
I exist ( I think ..)
Rational Solutions (Knowledge, Reasoning
Planning)
25Only One Problem IT DOESNT WORK
Or just marginally and then only in a very
artificial environment
UNFORTUNATELY, the real world is Dynamic,
Convoluted, Nonlinear, Messy
and Noisy
- Static
- Decomposable
- Linear
- Deterministic
26Artificial Autonomous Agents
Why Robots?
- Confronts us with
-
- important real world conditions and
- physical constraints
- that are hard to program or would go otherwise
unnoticed
27Artificial Autonomous Agents
28Object Avoidance by a Braitenberg Vehicle
(Braitenberg, 1984)
Architecture Operation
MOTORS
- Based on R/C car (Tyco Scorcher)
- Very fast
- Differential 4WD (4 propulsed out of 6)
- Intel 16-bit 196KD microcontrollers (20 MHz)
- IR, and ambient light sensors
- Programmable in C and assembler
Excitatory Connection
Obstacle Detector (IR sensor)
Inhibitory Connection
29Experimental Set Up
initial configuration
30Factor Analysis of Didabot Behaviour
ad.d adjust in presence of didabot ad.o
adjust in presence of object av.d avoid
didabot fo.d follow didabot nb.o nudge
object with backside ta..d turn around (
180o) in presence of didabot ta.o turn
around ( 180o) in presence of object ta.ow
turn around ( 180o) before an object
against the wall to.o touch
object wi wiggle wi.d wiggle in front
of didabot
F3 ObjectAversive DidaOriented (0.23)
.80
ad.d
ta.o
0.72
0.76
0.76
0.70
ta.ow
0.76
0.78
wi
fo.d
F5 nb.o (0.12)
ta.d
0.77
F6 (0.19)
0.79
0.77
ad.o
wi.d
0.78
0.86
0.64
0.84
av.d
F4 av.o (0.10)
0.77
to.o
Behaviour of situated robots can be very
complicated!
F1 ObjectOriented, DidaAversive (0.27)
31(No Transcript)
324
3
2
1
5
How Didabots bring objects together
33HOW ARE LARGE HEAPS FORMED ?
Why not just PAIRS of blocks ?
PAIRS Still many configurations that escape
detection
TRIOS/QUARTETS Prob(Contact) much lower!
CLUSTERS (n 4) GROW BY ADDITION OF SINGLE BLOCKS
34Against Wall
Object right ahead
Push object
Against other object(s) Pair (Trio)
Other Dida approaches
Leave Object
Single object
Heap
Destroy Pair/Trio
Avoid Dida
35SOME SUGGESTIONS
36Indiv.Based Models
What type of network (topology)
develops depending on
encounter frequency group size
time budget of
individuals
?
Forgetting-rate!
What happens when established topologies become
connected?
Can we study stability/ diversity of these
(linked) webs as in ecology?
Motivational dynamics (to signal willingness)
DynSys Models
In Robots?
37WHAT IS LACKING
ROBOT CULTURES
Proposed IP by J.L. Deneubourg in FET
Current Project