philosophy of science

12
20130905 1 Philosophy of Science Why do we have a lecture on the philosophy of science? To communicate science, you need to think criCcally about the topic To do this: You need to know what science tries to achieve You need to know what can go wrong What does it mean to be “scienCfic”? You’ve been studying it intensively for two years Many of you are now doing research projects Many of you plan to pursue a scienCfic career (doctoral research, medicine…) So what is it?

Upload: musy9999

Post on 04-Feb-2016

223 views

Category:

Documents


0 download

DESCRIPTION

Lecture

TRANSCRIPT

Page 1: Philosophy of Science

2013-­‐09-­‐05  

1  

Philosophy  of  Science  

Why  do  we  have  a  lecture  on  the  philosophy  of  science?  

•  To  communicate  science,  you  need  to  think  criCcally  about  the  topic  

 To  do  this:  •  You  need  to  know  what  science  tries  to  achieve  

•  You  need  to  know  what  can  go  wrong  

What  does  it  mean  to  be  “scienCfic”?  

•  You’ve  been  studying  it  intensively  for  two  years  

•  Many  of  you  are  now  doing  research  projects  

•  Many  of  you  plan  to  pursue  a  scienCfic  career  (doctoral  research,  medicine…)  

 So  what  is  it?  

Page 2: Philosophy of Science

2013-­‐09-­‐05  

2  

Science  is  derived  from  facts  

•  Historically  knowledge  was  derived  from  authority  

•  The  idea  of  systemaCcally  tesCng  “truths”  is  oRen  aSributed  to  Galileo  in  the  17th  century  (though  it  was  probably  more  common)  

•  E.g.  Aristotle,  Galileo  and  falling  masses    

AssumpCon  of  Facts  

1.  Careful  observaCon  will  lead  the  observer  to  

collect  an  accurate  fact  

2.  Facts  are  independent  of  any  theory  (they  lead  

to  theory)  

3.  Facts  are  firm  and  reliable  (REPEATABILITY)  

1.Careful  observaCon  will  lead  the  observer  to  collect  an  accurate  fact  

Page 3: Philosophy of Science

2013-­‐09-­‐05  

3  

1.Careful  observaCon  will  lead  the  observer  to  collect  an  accurate  fact  

Two  observers  don’t  necessarily  see  the  same  thing  

X-­‐ray  InterpretaCon  Fish  Eye  Colour  

Page 4: Philosophy of Science

2013-­‐09-­‐05  

4  

2.  Facts  are  independent  of  any  theory  (they  lead  to  theory)  •  But  wait!  Lets  think  about  the  stair  example  

Not  everyone  sees  a  staircase!  Our  experiences  influence  our  percepCons  

3.  Facts  are  firm  and  reliable  

•  Facts  change    E.g.  Earth’s  axis  changes  its  Clt,  the  length  of  a  day  increases  over  Cme,  the  magneCc  poles  reverse    Our  judgement  about  the  validity  of  an  observaCon  is  guided  by  what  we  know  or  assume.  This  makes  facts  just  as  fallible  as  our  assumpCons.  

DeducCve  Reasoning  

•  Proceeding  from  general  statements  to  a  specific  conclusion.  

•  This  conclusion  must  be  true  given  the  previous  statements  

Page 5: Philosophy of Science

2013-­‐09-­‐05  

5  

Logical  arguments  

1)  Philosophy  is  dull  2)  This  lecture  is  on  

philosophy  

Thus:  3)  This  lecture  is  dull  

3  is  a  logical  conclusion    which  flows  from  1  and  2    If  1  and  2  are  correct,    3  must  be  correct  

1)  A  bird  is  a  mammal  2)  A  crow  is  a  bird  

Thus:  3)  A  crow  is  a  mammal  

This  is  also  logical    It  is  not  correct  because    Premise  1  is  incorrect.      We  have  drawn  a  false    But  logical  conclusion  

1)  Copper  expands  when  heated  

2)  Iron  expands  when  heated  

3)  Gold  expands  when  heated  

Thus:  4)  All  metals  expand  when  heated  

This  is  not  a  logically  valid    argument    It  is  however  correct      It  is  possible  for  1,2  and  3  to  be  true  but  to  also    find  a  metal  that  does  not    expand  when  heated.  

InducCve  Reasoning  

Page 6: Philosophy of Science

2013-­‐09-­‐05  

6  

InducCve  Reasoning  

•  A  lot  of  science  actually  proceed  like  this  we  use  a  finite  amount  of  available  evidence  to  generalize  about  how  all  cases  will  respond.  

•  Why?  

 How  do  we  make  a  good  inducCve  

conclusion?    •  Large  number  of  observaCons  

–  (however  this  can  get  ridiculous  –  should  I  burn  my  hand  100  Cmes  to  demonstrate  fire  burns?)  

•  True  under  a  wide  variety  of  condiCons  –  (this  can  also  get  ridiculous  –  should  I  do  it  with  different  coloured  socks?)  

•  No  contradictory  cases  –  (is  this  fair?  There  are  very  few  rules  that  do  not  have  excepCons)  

Karl  Popper  and  FalisficaConism  •  InducCvist  theories  can  be  structured  so  vaguely  that  any  fact  can  support  them.    

 •  FalisificaConists  advocate  that  all  scienCfic  theories  should  make  testable  predicCons  

•  We  set  them  up  to  be  rigorously  tested,  and  embrace  the  fact  that  they  can  be  found  false  

Sir  Karl  Popper  1902-­‐1994  

 London  School  of  Economics  

Page 7: Philosophy of Science

2013-­‐09-­‐05  

7  

FalsificaCon  

Hypothesis  1:  All  substances  expand  when  heated    Hypothesis  2:  You  may  find  success  today  if  you  pursue  a  risky  venture  

FalsificaConists  argue:  •  The  best  theory  makes  many  specific  predicCons  and  runs  the  highest  risk  of  being  found  false.  

•  Any  new  hypothesis  should  be  more  falsifiable  than  the  one  it  replaces  

•  Hypotheses  cannot  be  “true”  but  can  be  more  widely  supported  than  a  previous  one  

•  Allows  us  to  compare  the  relaCve  strength  of  compeCng  hypotheses  

Lets  see  if  you  get  it….  

•  Hypothesis  1:  mars  moves  around  the  sun  

•  Hypothesis  2:  all  planets  move  around  the  sun  in  an  ellipCcal  orbit  

•  Hypothesis  3:  all  planets  move  around  the  sun  in  closed  loops  

Page 8: Philosophy of Science

2013-­‐09-­‐05  

8  

•  InducCvism:  the  goal  of  science  is  to  increase  the  observaCons  which  support  an  hypothesis  

•  FalsificaCon:  the  goal  of  science  is  to  test  the  falsifiability  of  an  hypothesis  or  to  replace  it  with  a  more  sophisCcated  hypothesis  

The  problem  with  FalsificaCon  

•  A  general  hypothesis  is  usually  based  on  a  series  of  interrelated  hypotheses  

•  TesCng  a  predicCon  normally  relies  on  a  complicated  experiment  

•  If  the  hypothesis  is  falsified,  it  may  be  impossible  to  decide  why…  a  minor  sub  hypothesis?  A  technological  problem?  

The  1960’s  =  Theory  based  methods  

•  The  previous  philosophies  were  seen  as  two  rule  following  

•  The  response  in  the  1960s  was  the  emergence  of  the  idea  that  science  should  be  based  on  an  underlying  theory,  not  the  acquisiCon  of  facts  

Page 9: Philosophy of Science

2013-­‐09-­‐05  

9  

Kuhn’s  Paradigms  

Thomas  Kuhn  1922-­‐1996  

 University  of  California  

Berkeley  

Pre  science  

Normal  science  

Crisis  

Paradigm  RevoluCon  

New-­‐normal  science  

New  Crisis  

Progress  

Paradigm  

Lakatos’  Research  Programmes  

Imre  Lakatos  1922-­‐1974  

 London  School  Of  Economics  

-­‐  Strongly  influenced  by  Popper    

-­‐  ASempted  to  “fix”  falsificaCon    

-­‐  Not  all  theories  are  equal,  if  you  falsify  a  theory,  you  can  modify  “lesser”  components  and  thus  maintain  the  “research  programme”    

-­‐Programmes  can  be  replaced  by    one  that  makes  beSer  predicCons  

Feyerbend’s  anarchisCc  theory  

Paul  Feyerbend  1924-­‐1994  

 University  of  California  

Berkeley  

-­‐  Anything  goes  

-­‐  Argued  that  scienCsts  rarely  follow  a  specific  method,  and    to  do  so  would  constrain  creaCvity    

-­‐  Did  not  think  science  should  be  given  any  special  status    -­‐  Argued  that  scienCsts  should    pick  any  theory  they  want  to  follow    just  as  we  pick  poliCcal  parCes  

Page 10: Philosophy of Science

2013-­‐09-­‐05  

10  

The  Bayesians  

Thomas  Bayes  1701-­‐1761  

 Kent,  UK  

-­‐  Rediscovered  Bayes  Theorem    -­‐  Assign  probabiliCes  to  various  hypotheses  based  on  prior  evidence    -­‐  New  types  of  evidence  provide  more  support  than  addiConal    evidence  of  the  same  kind    -­‐  We  will  return  to  Bayesian  maths  in  our  discussion  of  phylogeneCcs      

Mayo’s  experimentalism  •  Experimentalists  have  pracCcal  methods  of  measuring  effects  without  the  need  for  theories  

•  Progress  in  science  is  the  accumulaCon  of  experimental  data  

•  This  likely  encompasses  some  of  what  we  call  applied  science    and  the  science  of  error  staCsCcs  

Deborah  Mayo  Virginia  Tech  ,  USA  

London  School  of  Economics  

Summary  •  Induc&vism:  gather  lots  of  data  to  generate  theory  •  Falsifica&on:  theory  dependent,  theories  are  those  that  survive  test  

•  Kuhn’s  Paradigms:  consensus  of  the  community  on  what  paradigm  to  use  

•  Lakatos  Research  Programmes:  protect  core  ideas  and  modify  peripheral  hypotheses  

•  Feyerbend:  rejects  “method”  in  favour  of  theory  •  Bayesian's:  use  probability  to  evaluate  compeCng  theories  

•  Experimentalists:  it  is  possible  to  use  method  to  evaluate  data  independent  of  theory  

Page 11: Philosophy of Science

2013-­‐09-­‐05  

11  

What  is  “science”  and  how  do  I  know  if  I  am  being  

“scienCfic”?  

Today’s  take  home  message:  

•  The  answer  to  that  quesCon  is  not  obvious  

•  The  answer  has  changed  through  Cme  

•  The  answer  is  not  universal  between  or  within  disciplines  

Clare’s  Science    (forgive  the  imperCnence)  

•  DisCnguish  between  the  goals  and  method.  

Goals  of  my  science:  1)  To  gather  an  understanding  of  cause  and  

effect  relaConships  2)  To  predict  future  event  based  on  past  

observaCons  3)  To  explain  mechanisms  

Page 12: Philosophy of Science

2013-­‐09-­‐05  

12  

Clare’s  Science    

Methods  of  my  science:  1)  Make  observaCons  2)  Explain  observaCons  in  light  of  theory  3)  Make  predicCons  which  test  theory    4)  Be  willing  to  acknowledge  sources  of  error  5)  Be  willing  to  change  theory  

Finally:  A  note  on  the  philosophy  of  science  

•  Consider  the  evoluCon  of  science  philosophy  

•  Just  as  we  falsify  hypotheses,  change  theory  and  celebrate  the  recogniCon  of  errors,  we  also  revise  the  philosophical  mechanism  of  study  

•  What  could  be  more  scienCfic?  

What  is  this  thing  called    Science?  A.F.  Chalmers  

If  you  find  this  topic  interesCng  I  suggest  you  read  this  book.    Most  of  this  lecture  is  based  on  the  text.  At  my  PhD  insCtuCon    we  were  all  required  to  read  this  before  we  graduated!