Methodological Investigations in Agent-based Modelling with Applications for the Social Sciences

241 Pages·2018·1.894 MB·english

by  Eric Silverman

Eric Silverman
Methodological
Investigations in
Agent-Based Modelling
With Applications for the Social Sciences
With contribution by
Daniel Courgeau (cid:129) Robert Franck
Jakub Bijak (cid:129) Jason Hilton
Jason Noble (cid:129) John Bryden
EricSilverman
MRC/CSOSocialandPublicHealth
SciencesUnit
UniversityofGlasgow
Glasgow,UK
MethodosSeries
ISBN978-3-319-72406-5 ISBN978-3-319-72408-9 (eBook)
https://doi.org/10.1007/978-3-319-72408-9
LibraryofCongressControlNumber:2017962298
©TheAuthor(s)2018.Thisbookisanopenaccesspublication.
OpenAccess ThisbookislicensedunderthetermsoftheCreativeCommonsAttribution4.0Inter-
nationalLicense(http://creativecommons.org/licenses/by/4.0/),whichpermitsuse,sharing,adaptation,
distributionandreproductioninanymediumorformat,aslongasyougiveappropriatecredittothe
originalauthor(s)andthesource,providealinktotheCreativeCommonslicenseandindicateifchanges
weremade.
The images or other third party material in this book are included in the book’s Creative Commons
license,unlessindicatedotherwiseinacreditlinetothematerial.Ifmaterialisnotincludedinthebook’s
CreativeCommonslicenseandyourintendeduseisnotpermittedbystatutoryregulationorexceedsthe
permitteduse,youwillneedtoobtainpermissiondirectlyfromthecopyrightholder.
Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook
arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor
theeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforany
errorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictional
claimsinpublishedmapsandinstitutionalaffiliations.
Printedonacid-freepaper
ThisSpringerimprintispublishedbySpringerNature
TheregisteredcompanyisSpringerInternationalPublishingAG
Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland
Foreword
Following the aims of the Methodos Series perfectly, this 13th volume on agent-
basedmodelsprovidesageneralviewoftheproblemsraisedbythisapproachand
showshowtheseproblemsmaybesolved.
These methods are derived from computer simulation studies used by mathe-
maticians and physicists. They are now applied in many social disciplines such as
artificiallife(Alife),politicalsciences,evolutionarypsychology,demography,and
manyothers.Thosewhointroducedthemoftentookcarenottoconsidereachsocial
science separately but to view them as a whole, incorporating a wide spectrum of
socialprocesses–demographic,economic,sociological,political,andsoon.
Ratherthanmodellingspecificdata,thisapproachmodelstheoreticalideasand
is based on computer simulation. Its aim is to understand how the behaviour of
biological, social, or more complex systems arises from the characteristics of the
individuals or agents composing the said system. As Billari and Prskawetz (2003,
p.42)said,
Differenttotheapproachofexperimentaleconomicsandotherfieldsofbehaviouralscience
thataimtounderstandwhyspecificrulesareappliedbyhumans,agent-basedcomputational
models pre-suppose rules of behaviour and verify whether these micro based rules can
explainmacroscopicregularities.
Thisis,therefore,abottom-upapproach,withpopulation-levelbehaviouremerg-
ing from rules of behaviour of autonomous individuals. These rules need to be
clearly discussed; unfortunately, this approach is now used without sufficient
discussions in many social sciences. It eliminates the need for empirical data on
personalorsocialcharacteristicstoexplainaphenomenon,asitisbasedonsimple
decision-makingrulesfollowedbyindividuals,whichcanexplainsomereal-world
phenomena.Buthowcanwefindtheserules?AsBurch(2003,p.251)putsit,
Amodelexplainssomereal-worldphenomenonifa)themodelisappropriatetothereal-
worldsystem[...]andb)ifthemodellogicallyimpliesthephenomenon,inotherwords,if
thephenomenonfollowslogicallyfromthemodelasspecifiedtofitaparticularpartofthe
realworld.
vii
viii Foreword
Also,atheoreticalmodelofthiskindcannotbevalidatedinthesamewayasan
empirical model with the “covering law” approach, which hinders social research
andleadstoapessimisticviewoftheexplanatorypowerofthesocialsciences.In
Franck’swords(Franck2002,p.289),
But, one has ceased to credit deduction with the power of explaining phenomena.
Explainingphenomenameansdiscoveringprincipleswhichareimpliedbythephenomena.
Itdoesnotmeandiscoveringphenomenawhichareimpliedbytheprinciples.
As the agent-based approach focuses on the mechanisms driving the actions
of individuals or agents, it will simulate the evolution of such a population from
simple rules of behaviour. It may thus use game theory, complex systems theory,
emergence, evolutionary programming and – to introduce randomness – Monte
Carlomethods.Itmayalsousesurveydata,nottoexplainthephenomenonstudied,
butonlytoverifyiftheparametersusedinthesimulationleadtoabehavioursimilar
totheoneobservedinthesurvey.
As we have already said, such an approach raises many problems which this
volume will try to answer. We will present here these main problems, letting the
readerseehowSilvermanhastreatedit.
Thefirstproblemisthatthesemodels“areintendedtorepresenttheimportand
impact of individual actions on the macro-level patterns observed in a complex
system” (Courgeau et al. 2017, p. 38). This implies that a phenomenon emerging
at the aggregate level can be entirely explained by individual behaviour. Holland
(2012,p.48),however,statesthatagent-basedmodelsinclude“littleprovisionfor
agent conglomerates that provide building blocks and behaviour at a higher level
of organisation.” For instance, a multilevel study on the effects of an individual
characteristic (being a farmer) and the corresponding aggregate characteristic (the
proportion of farmers living in an area) on the probability of internal migration in
Norway shows that the effects are contradictory (Courgeau 2007): it seems hard
to explain a macro-characteristic acting positively by a micro-characteristic acting
negatively.Infact,micro-levelrulesareoftenhardtolinktoaggregate-levelrules,
and I believe that aggregate-level rules cannot be modelled with a purely micro
approach,fortheytranscendthebehavioursofthecomponentagents.
The second problem is that this approach is basically bottom-up. However, it
seemsimportanttotakeintoconsiderationsimultaneouslyatop-downprocessfrom
higher-level properties to lower-level entities. More specifically, we should speak
of a micro-macro link (Conte et al. 2012, p. 336) that “is the loop process by
which behaviour at the individual level generates higher-level structures (bottom-
up process), which feedback to the lower level (top-down), sometimes reinforcing
the producing behaviour either directly or indirectly”. The bottom-up approach of
a standard agent-based model cannot take such areciprocal micro-macro linkinto
account,giventhatitonlysimulatesonelevelofanalysis.
The third problem concerns the validation of an agent-based model. Such an
approach imitates human behaviour using some well-chosen mechanisms. It may
bejudgedsuccessfulwhenitaccuratelyreproducesthestructureofthisbehaviour.
Foreword ix
Determining success,however, requires amethod verydifferentfromthestandard
tests used to verify the validity of the effects of different characteristics in the
otherapproaches.Suchtestscanbeperformedinthenaturalsciencesbutaremore
difficultinthesocialsciences.AsKüppersandLenhardobserve(Günteretal.2005,
paragraph1.3),
Thereliabilityoftheknowledgeproducedbycomputersimulationistakenforgrantedif
thephysicalmodeliscorrect.Inthesecondcaseofsocialsimulationsingeneralthereis
notheoreticalmodelonwhichonecouldrely.Theknowledgeproducedinthiscaseseems
tobevalidifsomecharacteristicofthesocialdynamicsknownfromexperiencewiththe
socialworldarereproducedbythesimulation.
To determine if such an exploration has been successful, we need to consider
different aspects. First, how do we test that there are no other models offering a
betterexplanationoftheobservedphenomenon?Researchersoftentryoutdifferent
kindsofmodelssotheycanchoosetheonemostconsistentwithempiricaldata.But
thishardlysolvestheproblem,asthereisaninfinityofmodelsthatcanpredictthe
sameempiricalresultaswellorevenbetter.Second,howdowetestthatthechosen
model has a good fit with the observed data? Unfortunately, there is no clearly
defined procedure for testing the fit of a simulation model, such as significance
tests for the approaches described earlier. We can conclude that there are no clear
verificationandvalidationproceduresforagent-basedmodelsinthesocialsciences.
Whiletheagent-basedapproachappearstoresembleevent-historyanalysis,forit
focusesonindividualbehaviour,itneverthelessaimstoexplaincollectivebehaviour.
Atthatpoint,thekeyquestionis:howdowegeneratemacroscopicregularityusing
simpleindividualrules?Conteetal.(2012,p.340)perfectlydescribethedifficulties
encountered:
First, how to find out the simple local rules? How to avoid ad hoc and arbitrary
explanations? As already observed, one criterion has often been used, i.e., choose the
conditionsthataresufficienttogenerateagiveneffect.However,thisleadstoagreatdeal
ofalternativeoptions,allofwhicharetosomeextentarbitrary.
Without factoring in the influence of networks on individual behaviour, we can
hardly obtain a macro behaviour merely by aggregating individual behaviours.
To obtain more satisfactory models, we must introduce decision-making theories.
Unfortunately,thechoiceoftheoryisinfluencedbytheresearcher’sdisciplineand
canproducehighlydivergentresultsforthesamephenomenonstudied.
In order to go further, Chap.9, co-authored by Jakub Bijak, Daniel Courgeau,
Robert Franck and Eric Silverman, proposes for demography the enlargement of
agent-basedmodelstoamodel-basedresearch.Thiswillnotbeanewparadigmin
thetraditionalsense,aswiththecross-sectional,thecohort,theevent-historyandthe
multilevelapproaches,butanewwaytoovercomethelimitationsofdemographic
knowledge. It is a research programme which adds a new avenue of empirical
relevance todemographic research.Theexamples giveninthefollowingchapters,
despite the simplicity of the models used, give us a glimpse of the importance of
model-baseddemography.
x Foreword
IhopeIhavegiventothereaderofthisvolumeaclearideaofitsimportancefor
socialsciences.
Mougins,France DanielCourgeau
August2017
References
Billari,F.,&Prskawetz,A.(2003).Agent-basedcomputationaldemography:Usingsimulationto
improveourunderstandingofdemographicbehaviour.Heidelberg:PhysicaVerlag.
Burch,T.K.(2003).Data,models,theoryandreality:Thestructureofdemographicknowledge.In
F.Billari&A.Prskawetz(Eds.),Agent-basedcomputationaldemography:Usingsimulationto
improveourunderstandingofdemographicbehaviour(pp.19–40).Heidelberg:PhysicaVerlag.
Conte,R.,Gilbert,N.,Bonelli,G.,Cioffi-Revilla,C.,Deffuant,G.,Kertesz,V.,Loreto,V.,Moat,
S.,Nadal,J.-p.,Sanchez,A.,Nowak,A.,Flache,A.,SanMiguel,M.,&Helbing,D.(2012).
Manifesto of computational social science. European Physical Journal Special Topics, 214,
325–346.
Courgeau,D.(2007).Multilevelsynthesis:Fromthegrouptotheindividual.Dordrecht:Springer.
Courgeau,D.,Bijak,J.,Franck,R.,&Silverman,E.(2017).Model-baseddemography:Towards
a research agenda. In A. Grow & J. Van Bavel (Ed.), Agent-based modelling and popula-
tion studies (Springer series on demographic methods and population analysis, pp. 29–51).
Berlin/Heidelberg:Springer.
Franck,R.(Ed.).(2002).Theexplanatorypowerofmodels:Bridgingthegapbetweenempirical
andtheoreticalresearchinthesocialsciences(Methodosseries1).Boston/Dordrecht/London:
KluwerAcademic.
Holland,J.H.(2012).Signalsandboundaries.Cambridge:MITPress.
Küppers, G., & Lenhard, J. (2005). Validation of simulation: Patterns in the social and natural
sciences.JournalofArtificialSocietiesandSocialSimulation,8(4),3.
Contents
PartI Agent-BasedModels
1 Introduction................................................................. 3
1.1 Overview.............................................................. 3
1.2 ArtificialLifeasDigitalBiology.................................... 4
1.2.1 ArtificialLifeasEmpiricalData-Point.................... 4
1.3 SocialSimulationandSociologicalRelevance ..................... 5
1.3.1 MethodologicalConcernsinSocialSimulation........... 5
1.4 CaseStudy:Schelling’sResidentialSegregationModel........... 6
1.4.1 ImplicationsofSchelling’sModel......................... 6
1.5 SocialSimulationinApplication:TheCaseofDemography...... 7
1.5.1 BuildingModel-BasedDemography ...................... 7
1.6 GeneralSummary.................................................... 7
1.6.1 AlifeModelling............................................. 8
1.6.2 SimulationfortheSocialSciences......................... 8
1.6.3 Schelling’sModelasaCaseStudyinModelling ......... 8
1.6.4 DevelopingaModel-BasedDemography ................. 9
1.6.5 GeneralConclusionsoftheText:Messagesforthe
Modeller..................................................... 9
1.6.6 ChapterSummaries......................................... 10
1.6.7 Contributions ............................................... 13
References.................................................................... 14
2 SimulationandArtificialLife............................................. 17
2.1 Overview.............................................................. 17
2.2 IntroductiontoSimulationMethodology ........................... 18
2.2.1 TheGoalsofScientificModelling......................... 18
2.2.2 MathematicalModels ...................................... 18
2.2.3 ComputationalModels ..................................... 19
2.2.4 TheScienceVersusEngineeringDistinction.............. 20
xiii
xiv Contents
2.2.5 Connectionism:ScientificModellinginPsychology ..... 21
2.2.6 Bottom-UpModellingandEmergence.................... 22
2.3 EvolutionarySimulationModelsandArtificialLife ............... 22
2.3.1 GeneticAlgorithmsandGeneticProgramming........... 22
2.3.2 EvolutionarySimulationsandArtificialLife.............. 23
2.3.3 BedauandtheChallengesFacingALife .................. 24
2.4 TruthinSimulation:TheValidationProblem....................... 26
2.4.1 ValidationandVerificationinSimulation ................. 26
2.4.2 TheValidationProcessinEngineeringSimulations...... 26
2.4.3 ValidationinScientificSimulations:ConceptsofTruth.. 27
2.4.4 Validation in Scientific Models: Kuppers and
LenhardCaseStudy ........................................ 28
2.5 TheConnectionBetweenTheoryandSimulation.................. 29
2.5.1 Simulationas‘MiniatureTheories’........................ 29
2.5.2 SimulationsasTheoryandPopperianFalsificationism... 29
2.5.3 TheQuineanViewofScience.............................. 30
2.5.4 SimulationandtheQuineanView ......................... 31
2.6 ALifeandScientificExplanation.................................... 32
2.6.1 ExplanationThroughEmergence.......................... 32
2.6.2 StrongvsWeakEmergence................................ 33
2.6.3 SimulationasThoughtExperiment........................ 34
2.6.4 Explanation Compared: Simulations vs
MathematicalModels ...................................... 35
2.7 SummaryandConclusions .......................................... 36
References.................................................................... 36
3 MakingtheArtificialReal................................................. 39
3.1 Overview.............................................................. 39
3.2 Strongvs.WeakAlifeandAI ....................................... 40
3.2.1 Strongvs.WeakAI:CreatingIntelligence................ 40
3.2.2 Strongvs.WeakAlife:CreatingLife?..................... 40
3.2.3 DefiningLifeandMind .................................... 40
3.3 LevelsofArtificiality ................................................ 41
3.3.1 TheNeedforDefinitionsofArtificiality .................. 41
3.3.2 Artificial1:ExamplesandAnalysis........................ 42
3.3.3 Artificial2:ExamplesandAnalysis........................ 42
3.3.4 Keeley’sRelationshipsBetweenEntities.................. 43
3.4 ‘Real’AI:EmbodimentandReal-WorldFunctionality ............ 43
3.4.1 RodneyBrooksand‘IntelligenceWithoutReason’....... 43
3.4.2 Real-WorldFunctionalityinVisionandCognitive
Research..................................................... 44
3.4.3 TheDifferingGoalsofAIandAlife:Real-World
Constraints.................................................. 44
Contents xv
3.5 ‘Real’Alife:LangtonandtheInformationEcology................ 45
3.5.1 EarlyAlifeWorkandJustificationsforResearch......... 45
3.5.2 RayandLangton:CreatingDigitalLife?.................. 45
3.5.3 Langton’sInformationEcology............................ 46
3.6 TowardaFrameworkforEmpiricalAlife........................... 47
3.6.1 AFrameworkforEmpiricalScienceinAI................ 47
3.6.2 NewellandSimonLeadtheWay.......................... 48
3.6.3 Theory-DependenceinEmpiricalScience ................ 49
3.6.4 ArtificialDatainEmpiricalScience....................... 50
3.6.5 ArtificialDataandthe‘Backstory’ ........................ 53
3.6.6 Silverman and Bullock’s Framework: A PSS
HypothesisforLife......................................... 54
3.6.7 TheImportanceofBackstoryfortheModeller ........... 55
3.6.8 WheretoGofromHere .................................... 56
3.7 SummaryandConclusions .......................................... 57
References.................................................................... 58
4 ModellinginPopulationBiology.......................................... 61
4.1 Overview.............................................................. 61
4.2 Levins’Framework:Precision,Generality,andRealism........... 62
4.2.1 DescriptionofLevins’ThreeDimensions................. 62
4.3 Levins’L1,L2andL3Models:ExamplesandAnalysis........... 63
4.3.1 L1Models:SacrificingGenerality......................... 63
4.3.2 L2Models:SacrificingRealism ........................... 64
4.3.3 L3Models:SacrificingPrecision.......................... 64
4.4 OrzackandSober’sRebuttal......................................... 64
4.4.1 TheFallacyofClearlyDelineatedModelDimensions ... 64
4.4.2 SpecialCases:TheInseparabilityofLevins’Three
Factors....................................................... 65
4.5 ResolvingtheDebate:IntractabilityastheFourthFactor.......... 66
4.5.1 Missing the Point? Levins’ Framework as
PragmaticGuideline........................................ 66
4.5.2 Odenbaugh’sDefenceofLevins........................... 66
4.5.3 IntractabilityastheFourthFactor:ARefinement......... 67
4.6 ALevinsianFrameworkforAlife................................... 69
4.6.1 PopulationBiologyvs.Alife:ALackofData ............ 69
4.6.2 LevinsianAlife:AFrameworkforArtificialData?....... 70
4.6.3 ResemblingRealityandSitesofSociality ................ 70
4.6.4 Theory-DependenceRevisited............................. 72
4.7 TractabilityRevisited ................................................ 72
4.7.1 TractabilityandBraitenberg’sLaw........................ 72
4.7.2 DavidMarr’sClassicalCascade ........................... 74
4.7.3 RecoveringAlgorithmicUnderstanding................... 75
4.7.4 RandallBeerandRecoveringAlgorithmic
Understanding .............................................. 75
4.7.5 TheLureofArtificialWorlds.............................. 76
xvi Contents
4.8 SavingSimulation:FindingaPlaceforArtificialWorlds.......... 77
4.8.1 ShiftingtheTractabilityCeiling ........................... 77
4.8.2 SimulationasHypothesis-Testing ......................... 78
4.9 SummaryandConclusion............................................ 79
References.................................................................... 80
PartII ModellingSocialSystems
5 ModellingfortheSocialSciences......................................... 85
EricSilvermanandJohnBryden
5.1 Overview.............................................................. 85
5.2 Agent-BasedModelsinPoliticalScience........................... 86
5.2.1 SimulationinSocialScience:TheRoleofModels....... 86
5.2.2 Axelrod’sComplexityofCooperation..................... 87
5.3 Lars-ErikCedermanandPoliticalActorsasAgents ............... 87
5.3.1 EmergentActorsinWorldPoliticsaModelling
Manifesto ................................................... 87
5.3.2 CriticismfromthePoliticalScienceCommunity......... 88
5.3.3 AreasofContention:TheLackof‘Real’Data............ 89
5.4 Cederman’sModelTypes:ExamplesandAnalysis ................ 89
5.4.1 Type1:BehaviouralAspectsofSocialSystems.......... 89
5.4.2 Type2:EmergingConfigurations.......................... 90
5.4.3 Type3:InteractionNetworks .............................. 91
5.4.4 OverlapinCederman’sCategories......................... 91
5.5 MethodologicalPeculiaritiesofthePoliticalSciences............. 92
5.5.1 ALackofData:RelatingResultstotheRealWorld...... 92
5.5.2 ALackofHierarchy:InterdependenceofLevels
ofAnalysis.................................................. 93
5.5.3 ALackofClarity:ProblematicTheories.................. 93
5.6 InSearchofaFundamentalTheoryofSociety..................... 93
5.6.1 TheNeedforaFundamentalTheory...................... 93
5.6.2 ModellingtheFundamentals............................... 94
5.7 SystemsSociology:ANewApproachforSocialSimulation?..... 95
5.7.1 NiklasLuhmannandSocialSystems...................... 95
5.7.2 SystemsSociologyvs.SocialSimulation ................. 96
5.8 PromisesandPitfallsoftheSystemsSociologyApproach ........ 97
5.8.1 DigitalSocieties?........................................... 97
5.8.2 RejectingthePSSHypothesisforSociety................. 98
5.9 SocialExplanationandSocialSimulation .......................... 98
5.9.1 Sawyer’sAnalysisofSocialExplanation ................. 99
5.9.2 Non-reductiveIndividualism............................... 99
5.9.3 MacyandMiller’sViewofExplanation................... 101
5.9.4 AlifeandStrongEmergence............................... 102
5.9.5 Synthesis.................................................... 102
5.10 SummaryandConclusion............................................ 103
References.................................................................... 104