Class Notes (836,580)
Canada (509,856)
Psychology (5,217)
PSYCH 2H03 (134)

Ch. 11 - Thinking.pdf

8 Pages
Unlock Document

Judith Shedden

PNB  2XA3    Chapter  11:  Thinking     Judgment   • Deduction:  Start  w/  general  conclusion,  make  a  decision  about  a  specific  case  (top-­‐down)   o Orange  is  a  fruit.  Fruit  grows  on  trees.  Oranges  must  grow  on  trees.   • Induction:  Start  w/  a  specific  case/fact,  draw  general   conclusions  (bottom-­‐up)   o I  see  a  tiger  with  black  stripes  on  orange  fur.  ∴  all  tigers  have  black  stripes  on  orange  fur.   • What  people  actually  do  is  FAR  from  “ideal”  reasoning  (i.e.  attribute  substitution,  heuristics,  availability,   representativeness).     Judgment  Heuristics   • Availability  heuristic  –  Use  whatever  comes  to  mind  (MEMORY  is  crucial)   o Do  more  words  in  English  start  w/  R  or  K,  or  are  there  more  that  letter?  or  K  as  the  3 § They  come  up  w/  a  lot  more  words  that  start  w/  R  or  K  b/c  they’re  more   available.  They  say  there   are  more  that  start  w/  R  or  K.   o Do  more  words  in  English  end  w/  the  pattern  _n_    or  w/    -­‐ing???   § Words  ending  w/  -­‐ing  are  more  available.   • For  judgment,  need  to  be  able  to  judge  FREQUENCIES  of  events.   • People  use  AVAILABILITY  to  judge  FREQUENCY.   • Heuristics  are  GOOD,  except  when  they’re  NOT.   o Overestimate  probability  of  winning  the  lottery ,  homeless  are  mentally  ill,  car  crashes  vs.  plane  crashes  –   more  scared  of  planes  even  though  more  likely  to  get  in  car  accident …We  can  very  easily   lose  track  of  the   actually  frequencies   • The  Von  Restorff  effect  (aka  Isolation  effect  or  Distinctive  Principle) :  Memory  is  better  for  things  that  stand  out;  things   that  are  distinctive,  isolated,  humorous,  bizarre,   etc.  Used  a  lot  by  media.   • When  looking  for  evidence,  we  look  for  event  frequency.   To  estimate  frequency,  we  use  the  availability  heuristic.     • What  is  available  is  highly  influenced  by  memory  &  all  those  things  that  in  nce  memory.   Anchoring  &  adjustment   • Estimates  of  frequency  –  start  from  a  point  (anchor)  &  move  up  or  down.   o Given  only  5sec  to  estimate  the  # :  1*2*3*4*5*6*7ß=?  anchored  to  a  smaller  range   o Again:  8*7*6*5*4*3*2*1*=?   ß much  larger  estimate/  anchored  to  a  higher  range   • Kahneman  &  Tversky  (1974)   o How  long  is  the  Mississippi  River  –  the  river  is  actually  2348  miles  long   § Longer  or  shorter  than  500  miles?  This  group  guessed  1000  miles.   § Longer  or  shorter  than  5000  miles?  This  group  guessed  2000  miles   o People  use  even  a  completely  unreliable  #  as  an  anchor  &  go  from  there   • Problem:  hard  to  put  aside  original  estimates  –  whether  right  or  wrong  &  even  if  subjects  know  it  isn’t  reliable   o How  much  would  you  like  to  donate?   $500?  $250?  $100?  $50?  $10  is  a  much  better  idea  than  $10?  $50?   $100?  $250?  $500?     Representative  Heuristic   • Assume  homogeneity  (that  all  members  of  a  category  are  the  same)   o If  you’ve  seen  only  a  few  examples  of  a  category,  assume  that  all  of  the  category  members  are  like  that   • Note  that  representativeness  isn’t  a  completely  invalid  reasoning  strategy,  just  that  it  will  lead  to  some  mistakes.   • Which  of  the  following  events  is  most  likely?   1) A  man  under  55  has  a  heart  attack   2) A  man  has  a  heart  attack  ß  Most  likely   3) A  man  who  smokes  has  a  heart  attack   4) A  man  over  55  has  a  heart  attack  ß  People  say  this  is  most  likely   • This  is  the  CONJUCTION  FALLACY.     Conjunction  Fallacy   • Linda’s  31yrs  old,  single,  outspoken  &  very  bright.  She  majored  in  philosophy.  As  a  student,  she  was  deeply  concerned   w/  issues  of  discrimination  &  social  justice,  &  also   participated  in  anti-­‐nuclear  demonstrations.  Which  is  more  likely:   o Linda  is  a  band  teller.   o Linda  is  a  bank  teller  &  is  active  in  the  feminist  movement   ß  90%  picked  this  (even  though  there’s  less   probability  of  this  occurring)   o Base  rate  is  not  taken  into  ac count.   • How  likely  is  it  that  in  the  next  year,  a  massive  fire  in  American  will  kill  >1000  people?   • How  likely  is  it  that  next  year,  there  will  be  an  earthquake  in  Cali  causing  a  massive  fire  in  which  >1000poeple  will  be   killed  ß  less  probability  but  people  choose  this     Sunk  Cost  Fallacy  (“Concorde  Fallacy”)   • You  spend  $150  for  a  concert  ticket.  Then  you  spend  $50  better  concert  ticket.  You  realize  they’re  on  the  same  night.   It’s  too  late  to  sell  them;  you  must  use  one   &  not  the  other.  Which  concert  do  you  go  to?  Most  people  pick  more   expensive  concert  –  even  though  the  money’s  already  gone,  no  diff  if  you  go  to  the  cheaper  one.   • A  prior  commitment  (i.e.  already  placed  your  bet)   ↑  confidence  in  a  win   • Lost  a  dollar  in  a  vending  machine  (nothing  came  out).  Do  you  walk  away  or  put  another  dollar  in?  Will  you  pay  twice   to  get  something?  More  likely  to  not  work  2  time.  Wasting  more  $.     Gambler’s  Fallacy   • Toss  a  coin  8  times:  HHHHHHHH   • What’s  the  probability  of  a  tail  on  the  n ext  loss?  Gambler’s  fallacy  is  that  it’s  due  for  a  tail  even   though  it’s  50/50   • Which  is  most  likely  to  be  produced  at  random?  HHHHTTTT  or  HTHHTHTT   o People  pick  2  b/c  it  looks  more  random  but  both  are  equally  likely.   • Which  outcome  is  more  likely  in  lottery  tickets?  7,  45,  22,  19,  27,  37,  4  or  1,  2,  3,  4,  5,  6,  7     Reasoning  from  the  pop  to  an  instance:   • Gambler’s  fallacy:  The  belief  that  prior  outcomes  can  influence  the  outcomes  of  probabilistic  events.   • Law  of  large  #’s  –  things  do  tend  to  even  out  in  the  end,  over  a  LOT  of  trials   • BUT,  no  reason  that  for  small  samples  it  should  work  out  similarly.  Random  events  are  independent!   • “In  a  small  town,  there  are  2  hospitals.  Hospital  A  has  45births/day.  Hospital  B  has  15  births/day .  Overall  the  ratio  of   females/males  born  is  50/50.  Each  hospital  recorded  the  #  of  days  in  which,  on  that  day,  at  least  60%  of  the  babies   born  were  male.  ß  Hospital  A  will  be  more  variable.  People  think  small  samples  are  the  same  as  large  samples.     Reasoning  from  a  single  case  to  an  enti re  pop   • “The  man  who”  argument   • “Smoking  isn’t  that  bad,  I  have  an  uncle  who  smoked  2  packs/ day  &  he  died  in  his  sleep  at  age  85”   • But  the  whole  category  (smokers)  doesn’t  have  to  look  like  this   one  example  –  assumption  of  representativeness   • Hamill,  Wilson,  &  Nisbett  (1980)   o Showed  participants  video  of  a  prison  guard  talking  about  his  job.  Guard  either  compassionate  &  kind  or   contemptuous  &  mean   o Participants  told  that  the  guard  was  very  typical  of  guards  at  that  prison  or  that  the  guard  was  very  atypical   o Participants  interviewed  afterwards  about  their  views  of  the  prison  sy stem   § Saw  compassionate  guard  =  expressed  more  +  feelings  towards  guards   § Saw  mean  guard  =  expressed  more  negative  feelings   o Participants  paid  NO  attention  to  what  they  had  been  told  about   typicality  or  atypicality.   o They  acted  like  the  example  was  representative  of  the  entire  pop.     Hindsight  Bias   • People  think  after  the  fact  that  they  would  have  known  something  before  the  fact  when  they  really  wouldn’t  have.   • It’s  gray  outside,  you  think  it  might  rain.  When  it  DOES  rain,  you  remember  you  were  certain  that  it  would  when  you   say  the  clouds  rolling  in  earlier.   • Medical  diagnosis   –  2  groups  of  doctors  reviewing  medical  cases  (Arkes  et  al.  1981)   o Group  1:  not  told  the  actual  diagnosis   o Group  2:  told  the  actual  diagnosis  in  advance   o Given  the  medical  history  of  a  patient,  asked  to  rate  the  probabilities  of  4  diff  diagnoses .   o The  doctors  who  already  knew  the  dia gnosis  were  told  to  IGNORE  that  &  to  just  do  the  task  on  the  basis  of   the  evidence  at  hand.   o Results  –  Doctors  who  already  knew  the  diagnosis  assigned  probabilities  to  that  diagnosis  3x  HIGHER  than   did  the  doctors  who  didn’t  know  the  diagnosis.     The  Better-­‐than-­‐average  Effect   • People  think  that  they’re  better  than  average.  Of  course,  it  is  possible  fo r  everyone  to  be  better  than  average  (i.e.  Are   you  smarter/kinder/funnier  than  average?   • Clinically  depressed  is  an  exception.       Detecting  Covariation   • Covariation  is  crucial  to  our  reasoning.   o Does  something  have  an  effect  or  not?  Can  I  make  a  decision  about  something  based  on  another  factor?   • Illusions  of  covariation  (i.e.  interpreting  Rorschach  tests)   o People  project  their  own  prior  theories  onto  the  data   –  see  only  the  patterns  you  expect  to  see   • Confirmation  bias:  More  attentive  to  evidence  that  CONFIRMS  our  beliefs  than  to  evidence  that   falsifies  them.   • “Homeless  people  tend  to  be  mentally  ill”  ß  People  don’t  pay  very  much  attention  to  counterexamples   • Memory  schemata  –  helps  to  remember  examples   that  fit  your  prior  belief,  don’t  remember  counterexamples  –  leads   to  availability  effect  for  further  reasoning.     Regression  towards  the  Mean   • A  fundamental  statistical  property  of  the  universe  that  people  routinely  fail  to  account  for  in  their  reasoning.   • Whenever  you  get  an   extreme  example,  the  next  measure  will  likely  be  a  little  less  extreme.     Base  Rates   • Henri’s  a  quiet  man  who  likes  to  read  poetry.  Is  he  more  likely  to  be  an  English  professor  or  a  truck  driver?   o People  normally  say  “English  professor”   –  due  to  REPRESENTATIVENESS.  But  ignore  the  BASE  RATE   –  there   are  far  more  truck  drivers  in  the  world  than  English  professors.   • Need  to  combine  2  pieces  of  info  to  make  this  decision.   o How  DIAGNOSTIC  is  this  (what  proportion  of  English  professors   fit  this  description,  &  what  proportion  of   truck  drivers  fit  this  description)?   o What  is  the  BASE  RATE  for  English  professors  &  truck  drivers?   • Base  rate  neglect  is  extremely  common.   • “Marijuana  is  a  gateway  drug.  98%  of  people  who  are  heroi n  addicts  started  w/  marihuana.”   o Missing  piece  of  info  –  how  many  people  (both  heroin  addicts  &  NOT  heroin  addicts)  have  used  marijuana?   o Example:      200  people.  100  have  tried  marijuana,  100  haven’t.          Out  of  the  100  who  tried  marijuana,  2  are  heroin  addicts.          Out  of  the  100  who  haven’t  tr ied  marijuana,  0  are  heroin  addicts.          BUT  98%  of  the  marihuana  users  are  not  heroin  addicts.   • Kahnemann  &  Tversky,  1973   o 2  groups  of  participants  told  100  people  interviewed  in   a  certain  study.   o Group  1  told:  70%  were  engineers,  30%  were  lawyers   o Group  2  told:  30%  were  engineers,  70%  were  lawyers   o ∴  each  group  is  working  with  a  diff  base  rate.   o Jack  is  a  45yr-­‐old  man.  He’s  married  w/  4  children.  He’s  conservative,  careful  &  ambitious.  He  shows  no   interest  in  politics,  social  issues  &  spends  most  of  his  free  ti me  on  his  many  hobbies,  which  include  home   carpentry,  sailing,  &  mathematical  puzzles.  How  likely  is  it  that  Jack  is  an  engineer  in  the  interview  sample?   o Both  groups  estimated  90%  likelihood   o Based  entirely  on  the  representativeness;  ignoring  the  base  rate   (how  many  engineers  in  the  pop   • If  given  JUST  the  base  rates,  people  use  them  effectively.   • If  given  JUST  the  diagnostic  info,  people  use  it  following  categories,  stereotypes,  heuristics.   • BUT  given  BOTH  base  rates  &  diagnostic  info,  people  tend  to  ignore  the  base  rates.     SATISFICING  rather  than  OPTIMIZING   • It’s  foolish  seek  an  optimal  decision  –  it  may  require  more  time  or  effort  than  the  decision  is  actually  worth.   • Compromise  in  a  sensible  way   –  make  a  choice  that’s  good  enough.     Reasons  why  we  don’t  suck  tha t  much   • Sometimes  we  do  judge  accurately,  sometimes  we  don’t.  Why?  What  causes  us  to  differ  in  how  we  perform?   • Possibilities:   o Dual  System  Hypothesis:  System  1  (heuristics)  &  System  2  (more  effortful,  less  biased)   o Data  format  (it  depends  on  how  you  ask  the  q uestion)   o Triggering  statistical  knowledge.     Data  format   • Your  answer  depends  on  exactly  how  the  question  was  asked.   • People  are  better  at  reasoning  about  frequencies  than  about  probabilities  (although  mathematically  they  usually   come  to  the  same  thing)     Triggering   • People  might  have  KNOWLEDGE  of  statistical  reasoning,  they  just  fail  to  apply  it  sometimes.  Need  to  TRIGGER  it.   • People  are  better  at  remembering  to  use  statistical  knowledge  if  the  problem  involves  CHANCE.     • Also  better  if  the  problem  involves  a  role  of  SAMPLING.     So  what  can  we  do?   • Train  people  to  use  better  statistical  reasoning   • Nisbett  et  al  (1983)  –  phone  survey  of  people  taking  university  statistics  class,  asked  a  question  (baseball  example)   o Looked  at  rates  of  statistical  &  non-­‐statistical  reasoning.   o People  phone  early  in  semester  =  16%  statistical  answers   o Later  in  semester  =  37%  statistical  answers   o Similar  tests  done  on  various  majors  as  f irst  years  vs.  seniors  à  gain  after  1  year     The  Taxicab  Problem   • A  cab  was  involved  in  a  hit   &  run  accident  at  night.  2  cab  companies,  Green  &  Blue,  operate  in  the  city.  You’ re  given   the  following  data:   o 85%  of  the  city  cabs  are  Green   &  15%  are  Blue.     o A  witness  identified  the  cab  as  Blue.  His  vision  was  tested  under  appropriate  visibility.  Presented  w/  a  sample   (½  Blue,  ½  Green  cabs):     o 80%  correct  identifications  20%  errors   o What’s  the  probability  that  the  cab  involved  in  the  accident  was  Blue  rather  than  Green?       Testing  &  Adjusting  Our  Beliefs   • Logic  (ideal):  Logical  reasoning,  Rational  thinking,  Utility  theory   • What  people  do:  Biases,  Heuristics,  Errors,  Irrational  thinking,  Pragmatic  reasoning   • Overconfidence  –  Both
More Less

Related notes for PSYCH 2H03

Log In


Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.