Downloaded June 09, 2021 from Katherine Brading’s Website:

[Link]
 
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
In  her  paper  “Metaphysics  as  modelling:  the  handmaiden’s  tale”,  L.  A.  Paul  (2012)  seeks  to  
defend  contemporary  metaphysics  against  the  charge  that  “exploring  and  understanding  
the  world  through  metaphysical  reflection  is  obsolete”.  I  argue  that  her  defense  fails,  but  
that  the  reasons  why  enable  us  to  make  progress  in  providing  an  epistemological  
justification  for  metaphysical  enquiry.    
Paul’s  defense  of  contemporary  metaphysics  comes  in  two  steps.  First,  she  seeks  to  
establish  a  domain  of  metaphysical  enquiry  distinct  from  that  of  scientific  enquiry.  Second,  
with  that  in  place,  she  seeks  to  show  that  the  methods  of  metaphysics  are  similar  to  those  
of  science  in  just  those  ways  that  would  make  it  appropriate  for  a  scientific  realist  to  also  
be  a  metaphysical  realist.  This  second  step  involves  three  distinct  claims,  such  that  her  
overall  defense  involves  the  following  four  theses:  
1.  The  domain  of  metaphysical  enquiry  does  not  coincide  with  that  of  science.  
2.  The  semantic  view  of  theories  offers  a  characterization  of  metaphysical  and  
scientific  theories  appropriate  for  the  scientific  realist  and  metaphysical  realist  
alike.  
3.  The  evaluation  of  metaphysical  and  scientific  theories  is  similar  in  epistemically  
relevant  ways  (through  the  role  of  epistemic  virtues).  
4.  Ordinary  experience  provides  a  defeasible  yet  broadly  reliable  guide  to  the  
metaphysics  of  the  actual  world.  
I  argue  against  each  of  these  in  turn.  Nevertheless,  I  take  myself  to  have  a  shared  agenda  
with  Paul,  in  maintaining  that  there  are  legitimate  and  distinctively  metaphysical  questions  
and  areas  of  inquiry.  Where  we  differ  is  in  our  understanding  of  the  relationship  between  
metaphysical  and  scientific  theorizing,  and  in  the  methodologies  that  we  advocate.  In  this  
paper  I  attempt  to  identify  and  explicate  those  differences,  as  a  contribution  towards  
deepening  our  understanding  of  how  we  might  best  pursue  our  metaphysical  enquiries.    
 
_____________________________________________________   1  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
1.  The  domain  of  metaphysical  enquiry  does  not  coincide  with  that  of  science.  
 
Paul’s  first  thesis  is  that  “the  questions  and  problems  addressed  by  metaphysicians  are  
often  distinct  from  those  addressed  by  scientists”  (p.  4).  There  is  a  weak  reading  of  this  
claim  on  which  it  seems  to  me  unobjectionable.  I  agree  with  Paul  that  there  is  no  reason  to  
suppose  that  all  the  legitimate  questions  that  one  may  ask  about  the  natural  world  are  
exhausted  by  those  that  scientists  ask,  and  especially  no  reason  to  suppose  that  they  are  
exhausted  by  the  questions  scientists  currently  ask  (let  alone  those  that  they  are  also  able  
to  address  by  their  current  methods).    
However,  Paul  intends  a  much  stronger  reading  of  her  first  thesis  than  this.  She  is  
making  the  claim  that  scientific  theories  and  the  scientists  who  develop  and  use  them  
presuppose  metaphysical  concepts  in  a  manner  that  is  naïve  and  uncritical,  these  concepts  
lying  outside  the  domain  of  science  and  within  the  domain  of  metaphysics.  According  to  
Paul,  these  metaphysical  concepts  relate  to  the  distinctive  subject-­‐matter  of  metaphysics,  
and  she  claims  both  ontological  and  conceptual  priority  for  this  subject-­‐matter.  It  is  this  
stronger  thesis  with  which  I  wish  to  take  issue.  
The  first  question  we  might  ask  is  what  distinguishes  this  metaphysical  subject-­‐matter  
from  the  subject-­‐matter  of  science.  As  Paul  herself  makes  clear,  there  is  no  clear  boundary  
between  the  domains  of  metaphysics  and  science,  and  the  boundary  itself  moves  as  our  
sciences  develop.  I  would  add,  moreover,  that  “scientist”  is  not  a  timeless  natural  kind,  and  
that  the  methods  of  the  sciences  evolve  over  time.  So,  she  and  I  agree  that  a  crisp  
demarcation  criterion  between  metaphysics  and  science  that  picks  out  a  timeless  domain  
for  metaphysics  is  not  what  we  are  looking  for.  What,  then,  constitutes  or  demarcates  the  
domain  of  metaphysics?  
Paul  suggests  (p.  4)  that  paradigmatically  metaphysical  projects  include  “systematic,  
general  truths  concerning  fundamental  facts”  about  natures  (the  most  basic  ontological  
categories),  types  of  composition,  and  primitive  distinctions.  For  example,  a  metaphysical  
project  concerning  composition  would  try  to  “determine  whether  and  how  less  
fundamental  constituents  of  the  world  are  built  from  their  metaphysically  prior  
constituents”.  She  then  describes  how  she  understands  the  relationship  of  this  project  to  
_____________________________________________________   2  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
the  sciences,  claiming  that  “the  different  approaches  are  not  in  tension,  for  the  ontological  
account  involves  features  of  the  world  that  are  metaphysically  prior  to  those  of  the  
scientific  account”.  When  it  comes  to  composition,  for  example,  she  claims  that  while  
physics  and  chemistry  may  provide  a  causal  story  about  how  the  parts  cohere  to  form  a  
composite  physical  object,  it  is  metaphysics  that  provides  the  account  of  unity  by  which  the  
composition  of  the  parts  results  (or  fails  to  result)  in  a  genuine  whole.  It  is  in  this  sense  that  
the  subject-­‐matter  of  metaphysics  is,  according  to  Paul,  ontologically  prior  to  the  subject-­‐
matter  of  science  (p.  6):  “[m]etaphysics  tries  to  tell  us  what  laws,  naturalness,  properties,  
objects,  persistence,  and  causal  relations  fundamentally  are,  in  terms  of  natures,  and  
science  tries  to  discover  which  entities  there  are  or  how  these  natures  are  exemplified.”  
Elaborating  on  this  claim  of  ontological  priority,  Paul  says  (p.  6):  “The  fact  that  the  subject  
matter  of  metaphysics  can  be  ontologically  prior  to  the  subject  matter  of  science  is  
reflected  in  the  fact  that  many  concepts  of  metaphysics  are  conceptually  prior  to  the  
concepts  of  science”  and  goes  on  to  claim  that  “scientific  theorizing  usually  uncritically  
assumes  the  very  organizing  principles  and  deep  general  truths  that  metaphysics  is  
concerned  to  prescriptively  develop  and  understand.”  Thus,  we  have  both  the  ontological  
and  conceptual  priority  of  metaphysics  over  science,  and  it  is  in  this  sense  that,  according  
to  Paul,  the  domain  of  metaphysical  enquiry  does  not  coincide  with  that  of  science.  
I  think  this  account  of  the  relationship  between  metaphysical  theorizing  and  scientific  
theorizing  is  misleading.  Take,  for  example,  causation.  I  agree  with  Paul  that  we  have  pre-­‐
scientific  notions  of  causation  that  we  may  make  use  of  in  developing  a  given  scientific  
theory.  However,  her  conceptual  priority  claim  requires  that  the  pre-­‐scientific  notions  of  
causation  that  we  use  in  this  way  have  an  epistemic  status  that  is  importantly  independent  
of  the  development  of  that  scientific  theory.  I  agree  with  Ismael  (2013,  pp.  231-­‐233)  that  
this  is  a  mistake:  if  a  pre-­‐scientific  notion  of  causation  is  central  to  the  theorizing  in  
question,  then  the  very  development  of  that  scientific  theory  itself  involves  a  critique  and  
development  of  that  concept,  so  that  there  is  no  independent  concept  of  causation  to  be  
studied  by  metaphysicians  in  isolation  from  the  details  of  scientific  theories.  On  this  view,  
while  scientists  themselves  may  not  explicitly  engage  in  metaphysical  discussions  

_____________________________________________________   3  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
concerning  causation,  any  philosopher  who  wishes  to  do  so  must  pay  attention  to  the  
details  of  the  relevant  scientific  theorizing.    
In  order  to  further  develop  this  discussion,  we  need  to  make  two  distinctions,  one  
between  synchronic  theories  versus  diachronic  theorizing,  and  a  second  between  the  
questions  scientists  ask  versus  the  questions  philosophers  may  address  using  science.  A  
scientific  theory,  taken  off  the  shelf  and  studied  as  a  timeless  object,  may  indeed  
presuppose  concepts  that  the  theory  itself  fails  to  properly  explicate.  This  may  seem  to  
support  Paul’s  position.  However,  when  theories  are  viewed  diachronically,  such  that  the  
process  of  theorizing  is  the  subject  of  our  philosophical  scrutiny,  we  are  then  able  to  see  
which  of  those  concepts  are  themselves  being  subjected  to  investigation  through  the  
process  of  scientific  theorizing.  A  classic  example  here  is  in  our  theorizing  about  space  and  
time.  In  his  book  Understanding  Space-­‐Time,  DiSalle  (2006)  provides  an  extended  
treatment  of  the  development  of  our  concepts  of  space  and  time  under  the  pressures  and  
demands  of  physical  theorizing  about  the  motions  of  bodies,  arguing  that  “[d]espite  the  
delusions  of  philosophers  and  scientists  of  having  purely  epistemological  or  metaphysical  
insights  into  the  nature  of  space-­‐time,  philosophy  is  not  an  independent  source  of  
knowledge  of  space-­‐time”  (p.  157).  Rather,  he  argues,  the  developments  in  physical  
theorizing  are  part  of  an  ongoing  critique  of  concepts  that,  though  having  their  origins  in  
everyday  experience,  have  been  refined,  revised  and  elaborated  as  physics  has  developed.  
In  sum,  the  process  of  physical  theorizing  is  an  engagement  with  those  very  concepts  with  
which  the  metaphysician  is  concerned,  and  moreover  there  is  no  alternative  source  of  
spatiotemporal  concepts  to  which  she  can  appeal.  In  order  to  respond  to  this  line  of  
argument,  the  metaphysician  would  need  to  show  otherwise.  To  do  so  would  require  
engagement  with  the  details  of  scientific  theorizing,  in  order  to  show  that  what  the  
metaphysicians  claim  has  been  left  out,  has  indeed  been  left  out.  Merely  pointing  to  a  
scientific  theory  (rather  than  to  the  process  of  theorizing  by  which  that  theory  came  about)  
and  saying  that  it  does  not  include  some  aspect  of  our  pre-­‐theoretic  concept  does  not  
suffice,  since  that  aspect  may  have  been  subject  to  revision  in  the  process  of  theorizing  that  
led  to  that  theory.  

_____________________________________________________   4  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
 The  second  distinction  mentioned  above  is  between  the  questions  scientists  ask  versus  
the  questions  philosophers  may  address  using  science,  and  my  point  is  simply  this:  just  
because  scientists  themselves  may  not  ask  questions  about  the  metaphysics  of  properties,  
objects,  persistence,  and  causation,  it  does  not  follow  that  their  theorizing  does  not  probe  
these  concepts.  Indeed,  where  these  concepts  are  relied  upon  in  scientific  theorizing,  there  
is  every  chance  that  they  will  be  subject  to  development  and  revision  in  the  process  of  that  
theorizing,  and  it  behooves  philosophers  engaged  in  metaphysical  questions  to  pay  
attention  to  the  details  of  that  theorizing.    
Paul  argues  for  a  type  of  metaphysical  work  that  need  not  pay  attention  to  the  details  of  
scientific  theorizing  in  this  way.  Recall  the  quotation  given  above,  in  which  Paul  says  that  
“scientific  theorizing  usually  uncritically  assumes  the  very  organizing  principles  and  deep  
general  truths  that  metaphysics  is  concerned  to  prescriptively  develop  and  understand.”  
The  word  “prescriptive”  here  is  important  for  understanding  her  view:  metaphysicians  will  
tell  us  how  to  think  about  the  concepts  that  are  “uncritically  assumed”  (p.  6)  in  scientific  
theorizing.  Scientific  theories,  she  says  (pp.  6-­‐7),  are  not  to  “preemptively  define  the  role  or  
concepts  of  metaphysics”.  Rather,  the  relationship  between  the  work  of  the  metaphysician  
and  scientific  theories  is  simply  that  the  metaphysicians’  accounts  of  “their”  concepts  
should  be  in  some  weak  sense  “consistent  with  accepted  scientific  theories  of  the  world”  (p.  
6).  In  my  opinion,  this  fails  to  do  justice  to  the  conceptual  work  that  is  done  in  the  process  
of  scientific  theorizing,  and  fails  to  hold  the  metaphysician  appropriately  accountable  to  the  
details  of  that  theorizing.  Where  the  a  priori  metaphysician  described  by  Paul  assumes  that  
there  is  a  concept  (or  family  of  concepts)  of  causation  to  be  explicated,  the  empiricist  
metaphysician  accepts  that  there  may  be  no  such  concept  applicable  to  the  actual  world,  
and  that  our  empirical  theorizing  within  the  actual  world  is  an  ineliminable  resource  in  
determining  what,  if  any,  concept  or  concepts  of  causation  are  available  and  appropriate  
for  those  wishing  to  say  something  about  the  actual  world  in  which  we  find  ourselves.  
Similarly,  where  Paul’s  a  priori  metaphysician  assumes  that  we  can  develop  a  general  
account  of  parts  and  wholes  applicable  to  our  actual  world  independent  of  any  particular  
physical  theory,  the  empiricist  metaphysician  makes  no  such  assumption,  and  seeks  to  

_____________________________________________________   5  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
mobilize  our  best  empirical  theories  as  tools  for  investigating  metaphysical  questions  of  
mereology.1  
As  should  be  clear  by  now,  I  find  Paul’s  account  of  the  relationship  between  
metaphysics  and  science  problematic  not  because  I  think  that  scientists  ask  and  answer  all  
the  legitimate  questions  there  are  about  the  natural  world,  but  because  Paul’s  account  fails  
to  avail  the  metaphysician  of  epistemic  resources  crucial  to  her  own  projects  and  
questions.  This  is  a  difference  of  opinion  about  the  appropriate  way  to  understand  the  
relationship  between  metaphysical  and  scientific  theorizing,  and  it  is  one  which  has  
implications  for  the  appropriate  methodologies  for  pursuing  our  metaphysical  questions.  
By  framing  the  debate  not  as  one  of  metaphysics  versus  science,  but  of  the  appropriate  
relationship  between  metaphysical  and  scientific  theorizing,  we  can  make  progress  on  
methodology  with  respect  to  questions  of  metaphysics.  Thus,  I  share  Paul’s  rejection  of  the  
view  that  “exploring  and  understanding  the  world  through  metaphysical  reflection  is  
obsolete”,  but  I  disagree  with  her  about  how  best  to  proceed  with  our  metaphysical  
reflections.  
 
2.  The  semantic  view  of  theories  offers  a  characterization  of  metaphysical  and  
scientific  theories  appropriate  for  the  scientific  realist  and  metaphysical  realist  
alike.  
 
Paul  suggests  that  we  think  about  metaphysical  theorizing  as  a  process  of  modelling  
analogous  in  important  ways  to  modelling  in  science.  Here,  we  are  restricting  our  attention  
to  a  type  of  metaphysics  that  seeks  to  make  true  claims  about  the  actual  world  in  which  we  
find  ourselves,  and  I  think  Paul’s  proposal  is  interesting  for  metaphysicians  and  
philosophers  of  science  alike.  For  example,  her  proposal  includes  discussions  of  thought  
experiments  and  of  the  importance  of  modelling  for  the  investigation  of  counterfactual  

                                                                                                               
1  Quantum  mechahnical  non-­‐separability  is  the  most  famous  example  of  science  

challenging  our  “intuitions”  about  parts  and  wholes,  but  we  need  not  turn  to  quantum  
theory  to  see  mereological  work  being  done  within  scientific  theorizing.  For  problems  of  
part  and  whole  in  Newtonian  physics,  see  Brading  (2011,  2013).  
_____________________________________________________   6  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
dependencies,  about  both  of  which  there  is  a  large  literature  in  philosophy  of  science.  The  
extent  to  which  metaphysical  modelling  can  be  thought  of  as  similar  to  scientific  modelling  
will  depend  on  the  details  of  issues  in  that  literature.  I  agree  with  Paul  that  attempting  to  
make  concrete  a  methodology  or  methodologies  for  metaphysics  is  an  extremely  healthy  
move,  not  least  because  it  enables  us  to  get  more  precise  about  the  strengths  and  
weaknesses  of  metaphysics  as  an  approach  to  obtaining  knowledge  about  the  actual  world.  
In  what  follows,  I  outline  some  problems  for  a  methodology  for  metaphysics  that  appeals  to  
modelling  and  to  the  semantic  view  of  theories.  
For  Paul,  “the  most  important  differences  between  the  scientific  method  and  the  
metaphysical  method  derive  merely  from  the  difference  in  subject  matter  and  the  resultant  
difference  in  the  role  they  give  to  ordinary  experience.”  I  will  argue  that  this  “merely”  
makes  all  the  difference  in  the  world.  The  central  claim  Paul  makes  is  this  (p.  10):  “The  
theory  is  true  just  in  case  it  has  a  model  that  is  isomorphic  to  the  relevant  features  of  the  
world,  including  (but  not  limited  to)  the  structures  that  can  be  described  in  experimental  
and  measurement  reports.”2  For  the  metaphysician  to  be  justified  in  her  claim  that  she  has  
anything  to  say  about  the  metaphysics  of  the  actual  world,  she  must  be  able  to  justify  her  
claims  of  isomorphism  between  her  models  and  the  world.  Whether  she  can  do  so  in  a  
manner  analogous  to  the  scientific  realist  will  be  the  subject  of  this  and  the  following  
sections.  
For  the  scientific  realist,  the  empirical  success  of  science  plays  an  important  role  in  
justifying  the  claim  of  isomorphism  between  theory  and  world,  via  some  form  of  “no  
miracles”  argument:  if  our  models  were  not  (at  least  approximately)  isomorphic  to  the  
structures  in  the  world  then  the  success  of  science  would  be  a  miracle.3  And  by  success  of  
science  here  we  mean  detailed  empirical  success  including  successful  novel  predictions  and  
so  forth.  The  semantic  view  of  theories  offers  no  royal  road  to  realism  because  it  leaves  
open  the  question  of  the  relationship  between  the  models  of  the  theory  and  the  world.  
                                                                                                               
2  Within  philosophy  of  science,  there  is  abundant  literature  on  the  semantic  view  of  

theories,  both  in  its  original  formulation  and  subsequent  developments,  and  on  its  
problems  (see  Winther,  2016,  and  references  therein).  However,  for  our  purposes  I  set  
these  to  one  side  and  pursue  the  spirit  of  Paul’s  appeal  to  the  semantic  view.  
3  Brading  and  Landry,  2006,  section  4.  

_____________________________________________________   7  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
Some  version  of  “no  miracles”,  involving  feedback  between  theory  and  empirical  evidence,  
is  central  to  the  move  from  instrumentalism  to  a  scientific  realist  position,  and  if  the  
metaphysical  realist  is  to  adopt  the  methods  of  the  scientific  realist  then  she  will  need  an  
analogous  argument  in  order  to  justify  her  claim  of  isomorphism  between  her  models  and  
the  world.  Contrary  to  Paul,  it  is  not  merely  the  different  subject-­‐matter  and  the  resulting  
role  of  ordinary  experience  that  differentiates  the  methods  of  the  scientific  realist  from  
those  of  the  metaphysical  realist,  but  the  role  of  detailed  empirical  evidence  in  justifying  
the  theory-­‐world  connection.  So  the  metaphysician  who  claims  a  methodology  analogous  to  
that  of  a  scientific  realist  must  offer  an  alternative  justification  where  success  does  not  
mean  detailed  empirical  success.  In  her  paper,  Paul  offers  no  such  alternative  justification,  
and  this  presents  a  direct  challenge  to  her  attempt  to  minimize  the  differences  between  the  
methods  of  the  metaphysician  and  those  of  the  scientific  realist.  
 
3.  The  evaluation  of  metaphysical  and  scientific  theories  is  similar  in  epistemically  
relevant  ways  (through  the  role  of  epistemic  virtues).  
 
Paul  claims  (p.  19)  that  both  metaphysical  and  scientific  theorizing  rely  “on  a  priori  
reasoning  based  on  the  evaluation  of  theoretical  virtues  such  as  simplicity,  strength,  
elegance  and  the  like”,  and  “in  both  sorts  of  theorizing,  one  thing  that  can  justify  the  use  of  a  
priori  reasoning  is  that  they  employ  inference  to  the  best  explanation  based  on  the  idea  
that  theories  that  maximize  simplicity,  strength,  elegance,  and  other  theoretical  virtues  are  
more  likely  to  be  true.”  She  concludes:  “The  a  priori  elements  of  the  method  used  by  
metaphysicians  are  often  just  part  of  the  standard  arsenal  of  tools  employed  by  any  
theorist  of  the  unobservable,  the  indirectly  confirmable,  and  the  abstract.”  Later  on,  she  
says  (p.  21):  “We  use  theoretical  desiderata  as  guides  to  truth  in  metaphysics  just  as  we  use  
such  desiderata  as  guides  to  truth  in  science,  since  the  method  is  fundamentally  the  same  
even  when  the  subject  matter  is  different.”  I  believe  that  this  is  false.  
At  any  given  time,  our  empirical  evidence  is  typically  insufficient  to  uniquely  constrain  
scientific  theorizing  and  theory  choice.  We  therefore  make  use  of  additional  virtues,  and  
Paul’s  suggestion  is  that,  in  metaphysical  theorizing,  since  empirical  evidence  will  have  
_____________________________________________________   8  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
little  if  any  relevance,  it  is  these  virtues  on  which  we  must  rely  to  make  theory  selection:  
inference  to  the  best  explanation,  she  suggests,  is  here  used  to  infer  the  best  theory  on  the  
basis  of  virtues  other  than  empirical  adequacy,  in  just  the  same  way  as  in  science.  It  is  
worth  pausing  to  understand  the  steps  in  this  argument.  First,  we  note  that  in  scientific  
theorizing  we  make  use  of  theoretical  virtues.  Second,  we  claim  that  these  theoretical  
virtues  are  truth-­‐conducive  in  scientific  theorizing.  Third,  we  claim  that  since  they  are  
truth-­‐conducive  for  scientific  theorizing,  similarly  they  are  truth-­‐conducive  for  
metaphysical  theorizing.  The  problems  for  the  argument  arise  in  the  second  claim.  The  
second  claim  is  ambiguous  between  two  distinct  claims:  first,  that  at  least  in  some  
particular  circumstances,  certain  theoretical  virtues  may  be  taken  to  be  truth-­‐conducive  in  
scientific  theorizing,  albeit  fallibly;  second,  that  certain  theoretical  virtues  are  truth-­‐
conducive  and  this  truth-­‐conduciveness  is  context-­‐independent.  The  latter,  stronger,  claim  
is  not  something  that  the  scientific  realist  needs,  but  it  is  necessary  for  establishing  Paul’s  
third  claim,  that  since  theoretical  virtues  are  truth-­‐conducive  for  scientific  theorizing,  
similarly  they  are  truth-­‐conducive  for  metaphysical  theorizing.  Paul  puts  this  stronger  
claim  herself  in  the  following  way  (p.  21):    
“The  theoretical  desiderata  we  use  to  choose  a  theory  include  simplicity,  
explanatory  power,  fertility,  elegance,  etc.,  and  are  guides  to  overall  explanatory  
power  and  support  inference  to  the  truth  of  theory.  A  scientific  realist  should  take  
such  desiderata  to  be  truth-­‐conducive,  since  it  is  hard  to  see  how  such  desiderata  
can  lead  us  to  truth  if  they  are  merely  or  even  mainly  pragmatic  virtues.  If  such  
theoretical  desiderata  are  truth  conducive  in  science,  they  are  also  truth  conducive  
in  metaphysics  (and  in  mathematics,  and  in  other  areas).  The  main  point  I  want  to  
make  here  is  that  if  the  method  can  lead  us  to  closer  to  the  truth  in  science,  it  can  
lead  us  closer  to  the  truth  in  metaphysics.”    
And  she  is  even  more  explicit  later  on  (p.  22):  
“if  such  features  are  truth  conducive  in  the  case  of  science,  they  should  be  truth  
conducive  more  generally.  That  is,  if  simplicity  and  other  theoretical  desiderata  are  
truth  conducive  in  scientific  theorizing,  they  are  truth  conducive  in  metaphysical  
theorizing.  This  is  a  central  part  of  my  thesis:  if  we  accept  inference  to  the  best  
_____________________________________________________   9  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
 explanation  in  ordinary  reasoning  and  in  scientific  theorizing,  we  should  accept  it  in  
metaphysical  theorizing.”  
I  think  Paul  is  right  that  her  conclusion  goes  through  if  and  only  if  the  scientific  realist  in  
fact  justifies  her  use  of  theoretical  virtues  by  claiming  that  they  are  inherently  (i.e.  
independently  of  context)  truth-­‐conducive  such  that  a  given  theory  is  the  best  theory  –  the  
most  likely  to  be  true,  the  closest  fit  to  the  truth  –  because  it  best  satisfies  truth-­‐conducive  
virtues.  However,  I  also  believe  that  this  is  not  how  the  scientific  realist  argues,  or  is  
justified  in  arguing.  
I  agree  that  scientific  theorizing  requires  appeal  to  theoretical  virtues,  and  that  the  
scientific  realist  may  appeal  to  such  virtues  in  justifying  her  claims  about  the  likely  truth  
(or  approximate  truth,  or  whatever)  of  a  given  theory  in  order  to  address  
underdetermination  challenges  to  scientific  realism.  However,  I  do  not  believe  that  there  
are  theoretical  virtues  that  have  been  shown  to  be  context-­‐independently  truth-­‐conducive.  
Once  again,  this  becomes  visible  if  we  turn  our  attention  away  from  scientific  theories  
considered  sychronically  to  consider  instead  the  diachronic  process  of  scientific  theorizing.  
Here,  we  see  a  history  of  learning  when,  and  how,  and  in  which  contexts,  different  
theoretical  virtues  are  helpful  in  developing  empirically  successful  (including  empirically  
predictively  novel)  theories.  Consider,  for  example,  that  classic  case  of  empirical  
underdetermination,  geocentric  astronomy  versus  heliocentric  astronomy  in  the  sixteenth  
and  seventeenth  centuries.  In  the  process  of  resolving  this  problem  of  underdetermination,  
multiple  theoretical  virtues  were  invoked  by  Kepler  (for  example)  in  his  attempts  to  argue  
for  heliocentrism  as  the  real  structure  of  our  planetary  system.  These  included  harmony,  
simplicity,  explanatory  power,  and  so  forth.  However,  it  was  only  through  the  process  of  
two  hundred  years  of  theorizing,  including  the  interplay  between  theory  and  empirical  
evidence,  that  we  learned  which  of  these  virtues  were  more  helpful  than  others  for  the  
particular  problem  of  solving  the  system  of  the  world.  Harmony  and  simplicity  turned  out  
to  be  misleading  and  equivocal  (respectively),  and  explanatory  power  turned  out  to  be  
highly  effective  if  and  only  if  it  was  tied  to  particular  causal  commitments  (every  
acceleration  of  a  body  must  have  a  material  source)  and  freed  from  others  (vortex  theory).  
We  learned  about  which  virtues  are  effective  in  gravitational  theory,  and  these  proved  
_____________________________________________________   10  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
themselves  within  that  context  in  the  ensuing  centuries.  However,  during  that  time  we  also  
learned  that  they  are  not  context-­‐independent;  where  the  phenomena  seem  to  demand  
non-­‐linear  theories,  for  example,  a  different  set  of  virtues  is  needed.  
The  onus  is  on  Paul,  I  believe,  to  show  that  the  scientific  realist  not  only  requires,  but  
also  successfully  deploys,  a  context-­‐independent  set  of  truth-­‐conducive  theoretical  virtues,  
such  that  the  metaphysician  can  help  herself  to  these  same  virtues  with  a  reasonable  
presumption  that  they  will  remain  truth-­‐conducive  for  her,  too.  No  such  case  has  yet  been  
made.4  In  sum,  Paul  may  be  right  that  metaphysical  theorizing  involves  maximizing  
theoretical  virtues,  but  she  has  not  provided  any  grounds  for  believing  that  such  a  method  
is  truth-­‐conducive.  
 
4.  Ordinary  experience  provides  a  defeasible  yet  broadly  reliable  guide  to  the  
metaphysics  of  the  actual  world.  
 
Paul  argues  that  it  is  appropriate  for  the  metaphysician  to  “privilege  ordinary  experience  in  
the  sense  of  relying  on  it  as  an  initial,  but  defeasible,  guide  to  the  nature  of  the  world.  Such  a  
metaphysician  starts  with  the  defeasible  assumption  that  the  relevant  feature  of  the  world  
is  as  it  seems  to  us,  given  ordinary  experience.”  (p.  16)  Thusfar,  few  scientists  would  
disagree,  so  the  metaphysician  and  the  scientist  share  a  common  starting  point.  The  
difference,  Paul  claims,  is  that  the  metaphysician  seeks  to  explore  general  truths  or  features  
of  the  world  that  “hold  across  all  levels  of  experience,  from  the  macroscopic  level  to  the  
microscopic  level”,  and  therefore  hold  for  macroscopic  objects  and  events.  But  these  

                                                                                                               
4  The  claims  are  made,  but  the  argument  is  not  given  that  this  is  how,  in  fact,  the  scientific  

realist  can  and  must  argue.  Here  is  another  example  of  the  claim  being  made  (p.  25):  “To  
the  extent  that  one  can  endorse  the  realist  view  that  scientific  theories  are  true  (and  that  
we  can  infer  truth  from  successful  explanation),  one  endorses  the  thesis  that  maximizing  
theoretical  desiderata  brings  one  closer  to  the  truth.  To  the  extent  that  the  naturalist  
endorses  the  thesis  that  maximizing  theoretical  desiderata  brings  one  closer  to  the  truth,  
the  naturalist  can  endorse  the  view  that  doing  metaphysics,  and  philosophy  more  
generally,  is  a  rational  and  reasonable  way  to  try  to  discover  fundamental  and  general  
truths  about  the  world.”  
 
_____________________________________________________   11  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
general  features,  which  seem  to  hold  at  the  macroscopic  level,  are  also  present  in  the  
scientists’  starting  point,  and  scientists  have  found  empirically  powerful  methods  for  
discovering  just  where  and  when  that  defeasible  guide  is  misleading.  There  is  no  
independent  domain  of  “general  features”  of  our  empirical  experience  that  is  somehow  
invisible  to  scientists  and  not  found  within  the  empirical  evidence  on  which  they  draw.  
Contrary  to  Paul,  the  metaphysician  is  not  perfectly  justified  in  focusing  on  ordinary  objects  
and  properties  of  experience:  she  must  first  enquire  whether  those  features  of  our  
experience  have  already  proven  themselves  to  be  misleading,  and  for  this  she  must  
acquaint  herself  with  the  relevant  details  of  the  relevant  scientific  theories.  I  argued  above  
that  the  process  of  scientific  theorizing  has  itself  often  been  a  process  of  enquiring  into  the  
very  concepts  that  the  metaphysician  takes  herself  to  be  concerned  with  (such  as  
properties,  causation,  composition,  and  so  forth),  and  if  this  is  right  then  the  claim  that  
ordinary  experience  is  just  as  informative  for  the  metaphysician  in  such  cases  is  mistaken.  
To  repeat  a  point  made  earlier:  scientific  theories,  taken  synchronically,  do  not  wear  their  
metaphysical  investigations  on  their  sleeves,  but  taken  diachronically,  the  conceptual  
investigation  that  such  theorizing  involves  can  be  uncovered  and  made  visible  by  the  
philosopher,  and  the  metaphysical  import  of  such  work  developed  and  made  plain.    
Paul  claims  that  “many  interesting  and  important  metaphysical  theories  are  concerned  
with  the  actual  world  and  its  near  relatives”,  but  it  is  unclear  what  justifies  the  claim  that  
these  theories  are  indeed  about  the  actual  world.  If  they  fail  to  engage  with  scientific  
theorizing  in  the  way  I  have  described  above  then,  on  the  contrary,  we  have  good  reason  to  
suspect  that  they  may  not  concern  the  actual  world,  or  even  any  of  its  near  relatives,  at  all.  
This  is  because,  as  noted  above  (section  3),  the  scientific  realists’  “no  miracles”  argument  is  
unavailable  to  the  metaphysician  when  she  attempts  to  justify  a  relationship  (such  as  
isomorphism)  between  her  models  and  the  actual  world;  ordinary  experience  can  be  
profoundly  misleading  about  the  actual  world,  and  is  not  to  be  relied  upon  as  a  guide;  and  
superficial  consistency  with  some  scientific  theory  is  unreliable  as  a  means  of  engaging  
with  the  metaphysical  critique  that  is  involved  in  the  scientific  theorizing  of  which  that  
theory  is  a  part,  and  therefore  unreliable  as  a  route  to  the  actual  world.  For  example,  when  
Paul  says  (p.  24)  that  “we  perceive  certain  basic  properties  such  as  cohesiveness  and  
_____________________________________________________   12  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
continuity,  and  we  may  use  that  information  to  construct  a  theory  of  how  objects  persist”,  
we  must  recognize  that  (a)  this  theory  applies  to  the  actual  world  if  and  only  if  
cohesiveness  and  continuity  are  basic  properties  of  the  actual  world,  and  (b)  ordinary  
experience  has  already  proven  itself  to  be  an  unreliable  guide  with  respect  to  both.5  Thus,  a  
theory  of  how  objects  persist  that  is  built  on  these  “basic  properties”  is  unlikely  to  turn  out  
to  apply  to  the  actual  world,  or  to  any  world  close  to  our  own.  
Turning  her  attention  to  the  example  of  space,  Paul  writes  (p.  27)  that  “metaphysicians  
to  this  day  respect  the  fact  that  facts  about  spacetime  and  motion,  among  other  things,  have  
been  shown  to  be  empirically  determinable.”  This  may  be  true,  but  it  is  insufficient.  The  
philosophical  significance  for  space  and  time  of  Newton’s  physics  does  not  lie  in  the  fact  
that  absolute  accelerations  are  (allegedly)  empirically  determinable  in  Newtonian  theory  
whereas  absolute  velocities  are  not;  nor  of  Einstein’s  special  theory  of  relativity  in  the  fact  
that  we  have  no  empirical  way  to  determine  whether  spatially  separated  events  are  
simultaneous;  nor  of  Einstein’s  general  theory  of  relativity  in  the  fact  that  what  is  
empirically  determinable  turns  out  to  be  the  inertiogravitational  structure  of  spacetime  
and  not  the  inertial  structure  by  itself.  These  “facts”  could  be  incorporated  into  
metaphysical  theorizing  about  space  and  time  in  such  a  way  that  maintains  consistency  
with  these  “facts”,  but  which  utterly  fails  to  take  on  board  the  associated  critique  of  the  
very  concepts  of  space  and  time  with  which  the  metaphysician  is  concerned.  To  take  the  
philosophical  significance  for  time  of  Newton’s  physics  as  an  example,  Newton  inherited  a  
range  of  philosophical  options,  positions  and  distinctions  concerning  the  nature  and  
structure  of  time  which,  through  the  process  of  developing  his  project  to  solve  the  system  
of  the  world  (in  the  Principia)  he  was  forced  to  refine  and  revise  (see  Brading,  2016).  
Developments  of  this  project,  in  turn,  led  to  further  conceptual  clarifications  and  revisions,  
including  those  in  Einstein’s  special  and  general  theories  of  relativity  (see  DiSalle,  2006).  If  
we  are  to  understand  how  the  upshot  of  this  process  connects  to  our  shared  starting  point  

                                                                                                               
5  Continuity,  for  example,  was  a  powerful  and  successful  principle  of  physical  theorizing  

from  the  18th  century  into  the  early  20th  century,  before  empirical  evidence  forced  a  
reconsideration.  
_____________________________________________________   13  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
–  to  our  pre-­‐theoretical  concepts  of  time  –  then  we  must  understand  the  conceptual  
transformations  hard-­‐won  through  the  scientific  theorizing  itself.  
 
5.  Conclusions  
 
Paul  argues  for  a  degree  of  autonomy  of  metaphysics  from  science  that  seems  to  me  
epistemically  unjustifiable.  I  have  argued  against  her  assertion  of  the  ontological  and  
conceptual  priority  of  metaphysics,  and  against  her  view  that  there  is  a  methodology  for  
studing  the  metaphysics  of  this,  the  actual  world,  that  need  not  pay  attention  to  the  details  
of  scientific  theorizing.  It  is  important  to  stress  that  I  am  not  claiming  that  all  our  legitimate  
metaphysical  questions  are  already  in  deep  ways  engaged  with  by  the  details  of  scientific  
theorizing,  or  that  there  is  nothing  for  the  metaphysician  to  do  that  does  not  involve  
detailed  knowledge  of  various  areas  of  science.  I  agree  with  Paul  that  developing  toy  
models,  and  playing  around  with  “possibilities”  that  currently  lie  outside  our  empirical  
reach,  can  be  a  worthwhile  activity  for  theoretical  scientists  and  philosophers  alike.  
However,  the  areas  of  overlap  between  the  interests  of  the  metaphysicians  and  the  
activities  of  the  scientists  are  far  greater  than  her  position  allows.  This  is  easier  to  see  once  
we  set  aside  the  dichotomy  between  the  a  priori  metaphysician  and  the  a  posteriori  
scientist,  as  Paul  urges  us  to  do.  Lying  in  between  is  the  empiricist  metaphysician,  who  
takes  empirical  details  and  the  detailed,  local,  processes  of  scientific  theorizing  to  be  
epistemically  relevant  to  our  shared  metaphysical  questions.  We  do  not  need  to  suppose  
that  the  scientists  are  those  who  ask  all  the  legitimate  questions  about  the  natural  world  in  
order  to  believe  that  the  details  of  science  are  important  for  philosophy.  Once  we  pay  
attention  to  the  the  processes  of  scientific  theorizing,  we  see  that  the  theorizing  carried  out  
by  scientists  engages  in  detail  with  a  wider  range  of  questions  than  the  ones  that  they  
themselves  happen  to  ask.  These  questions  include  many  that  belong  to  us,  the  
metaphysicians.    
The  danger  associated  with  the  methodology  for  metaphysics  advocated  by  Paul  is  that  
it  will  fail  in  its  central  aim:  it  will  fail  to  say  anything  about  the  actual  world.  The  appeal  to  
the  semantic  view  of  theories  and  to  theoretical  virtues  cannot  bridge  the  gap  between  
_____________________________________________________   14  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
theory  and  world.  The  empiricist  metaphysics  methodology  that  I  advocate,  and  of  which  I  
have  given  some  brief  indications  in  this  paper,  seeks  to  significantly  reduce  that  danger  
and  thereby  to  offer  far  greater  epistemological  justification  for  our  metaphysical  
theorizing  than  that  which  is  to  be  had  by  mere  metaphysics  as  modelling.  
As  I  said  at  the  outset,  I  take  myself  to  have  a  shared  agenda  with  Paul,  in  maintaining  
that  there  are  legitimate  and  distinctively  metaphysical  questions  and  areas  of  inquiry.  My  
goal  here  has  been  to  try  to  identify  precisely  where  and  why  we  diverge,  in  the  hope  that  
this  will  further  our  conversation  about  methodologies  for  pursuing  the  metaphysical  
questions  that  we  share.  
   

_____________________________________________________   15  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)  
 
Brading,  K.  2011.  ‘On  composite  systems:  Descartes,  Newton,  and  the  law-­‐constitutive  
approach’,  in  Vanishing  Matter  and  the  Laws  of  Nature:  Descartes  and  Beyond,  ed.  Dana  
Jalobeanu  and  Peter  Anstey,  Studies  in  Seventeenth-­‐Century  Philosophy,  Routledge,  pp.  
130-­‐152.  
-­‐-­‐  2013.  ‘Three  principles  of  unity  in  Newton’,  Studies  in  History  and  Philosophy  of  Science  
44,  pp.  408-­‐415.  
-­‐-­‐  2016.  ‘Time  for  empiricist  metaphysics’.  Forthcoming  in  Metaphysics  and  the  Philosophy  
of  Science,  ed.  M.  Slater  and  Z.  Yudell.  Oxford  University  Press.  
Brading,  K.  and  Landry,  E.  2006.  ‘Scientific  Structuralism:  Presentation  and  
Representation’,  with  E.  Landry,  Philosophy  of  Science  73,  pp.  571-­‐81.  
DiSalle,  R.  2006.  Understanding  Space-­‐Time,  Cambridge  University  Press.  
Ismael,  J.,  2013.  ‘Causation,  Free  Will,  and  Naturalism’,  in  Scientific  Metaphysics,  ed.  D.  Ross,  
J.  Ladyman,  H.  Kincaid.  Oxford  University  Press.  
Paul,  L.  A.  2012.  ‘Metaphysics  as  modelling:  the  handmaiden’s  tale’.  Philosophical  Studies  
160:  1-­‐29.    
Winther,  Rasmus  Grønfeldt,  2016.  “The  Structure  of  Scientific  Theories”,  The  Stanford  
Encyclopedia  of  Philosophy  (Spring  2016  Edition),  Edward  N.  Zalta  (ed.),  URL  =  
<[Link]  
 
 

_____________________________________________________   16  
Metaphysics  as  modelling:  a  reply  to  L.  A.  Paul  
Katherine  Brading  (Draft:  September  5th,  2016)