Types of memory and models of memory [PDF]

In the lecture today. A review of short-‐term memory, and how much stuff fits in there anyway. 1. Whether or not the n

1 downloads 10 Views 2MB Size

Recommend Stories


Two types of translation memory
Be who you needed when you were younger. Anonymous

Computational models of working memory
If you are irritated by every rub, how will your mirror be polished? Rumi

Memory Consistency Models
At the end of your life, you will never regret not having passed one more test, not winning one more

Memory Consistency Models
Learn to light a candle in the darkest moments of someone’s life. Be the light that helps others see; i

Computation-Centric Memory Models
If you want to go quickly, go alone. If you want to go far, go together. African proverb

Weak memory models
Everything in the universe is within you. Ask all from yourself. Rumi

discrimination of and memory
Don't ruin a good today by thinking about a bad yesterday. Let it go. Anonymous

Serial Binary Memory Models or Parallel Continuous Memory Models?
Ego says, "Once everything falls into place, I'll feel peace." Spirit says "Find your peace, and then

Shared memory - Caches, Cache coherence and Memory consistency models
Seek knowledge from cradle to the grave. Prophet Muhammad (Peace be upon him)

Some targets for memory models
Don't fear change. The surprise is the only way to new discoveries. Be playful! Gordana Biernat

Idea Transcript


Inf1:  Intro  to  Cogni-ve  Science    

Types  of  memory  and  models  of   memory   Alyssa  Alcorn,  Helen  Pain  and  Henry  Thompson  

March 21, 2012

Intro to Cognitive Science

1

1.  In  the  lecture  today   A  review  of  short-­‐term   memory,  and  how  much   stuff  fits  in  there  anyway   1.  Whether  or  not  the   number  7  is  magic   2.  Working  memory   3.  The  Baddeley-­‐Hitch  model   of  memory   hEp://www.cartoonstock.com/directory/s/ short_term_memory.asp    

March 21, 2012

Intro to Cognitive Science

2

2.  Review  of  Short-­‐Term  Memory  (STM)   Short-­‐term  memory  (STM)  is  responsible  for  storing  small   amounts  of  material  over  short  periods  of  Nme   A  short  Nme  really  means  a  SHORT  Nme-­‐-­‐  up  to  several  seconds.     Anything  remembered  for  longer  than  this  Nme  is  classified  as   long-­‐term  memory  and  involves  different  systems  and   processes.     !!!  Note  that  this  is  different  that  what  we  mean  mean  by  short-­‐ term  memory  in  everyday  speech.  If  someone  cannot   remember  what  you  told  them  five  minutes  ago,  this  is   actually  a  problem  with  long-­‐term  memory.     While  much  STM  research  discusses  verbal  or  visuo-­‐spaNal   informaNon,  the  disNncNon  of  short  vs.  long-­‐term  applies  to   other  types  of  sNmuli  as  well.   3/21/12

Intro to Cognitive Science

3

3.  Memory  span  and  magic  numbers     Amount  of  informaNon  varies  with  individual’s  memory  span      =  longest  number  of  items  (e.g.  digits)  that  can  be  immediately   repeated  back  in  correct  order.   Classic  research  by  George  Miller  (1956)  described  the  apparent   limits  of  short-­‐term  memory  span  in  one  of  the  most-­‐cited   papers  in  all  of  psychology.  He  opens  (dramaNcally!)   “My  problem  is  that  I  have  been  persecuted  by  an  integer.  For   seven  years  this  number  has  followed  me  around,  has  intruded   in  my  most  private  data,  and  has  assaulted  me  from  the  pages   of  our  most  public  journals.  This  number  assumes  a  variety  of   disguises,  being  someEmes  a  liFle  larger  and  someEmes  a  liFle   smaller  than  usual,  but  never  changing  so  much  as  to  be   unrecognizable.”   3/21/12

Intro to Cognitive Science

4

Memory  span  and  magic  numbers,  con-nued   The  persecutory  number  in  quesNon  is  the  now-­‐famous  “magical   number  7  plus  or  minus  two”   For  STM,  this  means  that  the  general  span  is  5-­‐9  remembered   items  with  an  average  of  7     “Magical  number  7”  was  derived  from  research  with  absolute   judgement  tasks   •  People  were  asked  to  discriminate  between  sNmuli  varying  on   only  ONE  dimension   •  For  example,  judging  tones  that  vary  only  in  their  pitch   He  notes  performance  on  sNmuli  varying  on  mulNple  dimensions   may  be  quite  different,  giving  examples  illustraNng  that  most  of   the  sNmuli  we  encounter  in  everyday  life  are  of  this  type.     ....but  is  7  the  number  to  quote  when  we  talk  about  remembering   more  complex  sEmuli  like  leFers,  words,  or  digits?   3/21/12

Intro to Cognitive Science

5

4.  Maybe  not  a  magic  number?   Memory  span  results  were  widely  confirmed  and  replicated  for   many  years,  but  (relaNvely)  recent  research  suggests  the   esNmate  of  a  magical  7  is  too  high   •  Some  researchers  suggest  it  should  be  closer  to  3  or  4   •  In  any  case,  definitely  less  than  10!   Alternately,  the  problem  may  be  that  Miller’s  original  discussion   and  much  subsequent  literature  are  not  making  the   straigheorward  limit  they  THINK  they  are  measuring.          -­‐  For  example,  Miller’s  original  retrieval  task  and  similar  ones   tap  into  both  short  AND  long-­‐term  retrieval      -­‐  There  also  appears  to  be  variaNon  in  the  retenNon  properNes   for  different  types  of  informaNon   See  Shiffrin  &  Nosofsky  1994  for  a  review  and  further  references.     3/21/12

Intro to Cognitive Science

6

5.  Stretching  our  short-­‐term  memory  span   (aside  from  whether  the  “magic  number”  is  really  7  or  not....)   Short-­‐term  memory  span  can  be  “stretched”  by  meaningfully   grouping  or  chunking  informaNon    -­‐  easier  to  remember  one  year,  1918,  than  four  numbers  1,  9,  1,  8    -­‐  naturally  “recode”  informaNon  into  chunks  as  aid  to  memory!    -­‐  remember  postcodes,  phone  numbers  and  words  as  units  rather   than  lists  of  individual  pieces     Memory  span  for  words  in  a  sentence  is  significantly  longer  than  for   unrelated,  context-­‐less  words   •  About  15  words  (per  Brener,  1940),  compared  to  a  magical  single-­‐ digit    number     •  Constraints  imposed  by  the  rules  of  grammar,  context,  are  not   sufficient  to  explain  this!     (what  is  longest  number/  text  you  can  recall?)   3/21/12

Intro to Cognitive Science

7

6.  Are  “Short-­‐term  memory”  and  “working   memory”  synonymous?   In  a  psychology  class,  textbook,  or  elsewhere,  you  may  have  also   heard  about  working  memory.   Short-­‐term  memory  and  working  memory  may  sound  iniNally   similar,  but  their  relaNonship  is  more  complex.     These  are  NOT  interchangeable  terms.   •  Working  memory  (abbreviated  to  WM)  is  the  part  of  memory   conceived  as  a  “mental  workspace”  where  informaNon  is   temporarily  maintained  and  manipulated.   •  Short-­‐term  memory    as  discussed  earlier  in  this  lecture  is  more   specifically  concerned  with  storage,  maintaining  small   amounts  of  informaNon  (e.g.  “keeping  them  in  mind”  but  not   manipulaNng  them  or  processing  them  more  deeply)   3/21/12

Intro to Cognitive Science

8

STM  v.  WM  contd.   As  you  will  see  in  this   lecture,  the   disNncNon   between  STM  and   WM  has  evolved   over  Nme.     Short  term  memory  is   currently  conceived   as  one  component   of  the  larger   working  memory   system.  

3/21/12

Intro to Cognitive Science

9

7.  Working  memory  and  models  of  memory   We  will  consider  two  models  of  memory:   •  Modal  model  of  memory  (Atkinson  &  Shiffrin,  1968)   •  Baddeley-­‐Hitch  model  of  memory  (Baddeley  &  Hitch  1974;  and   many  more)   Remember  modelling  memory  to  generate  predic-ons  and  test   theories     •  How  much  of  the  current  data  from  humans  can  the  model   explain?   •  How  well  does  a  model  explain  current  human  behavioural  data?   •  What  hypotheses  does  the  model  suggest?  Are  these  confirmed/ disconfirmed  by  new  behavioural  data?   Plus  addiNonal  benefits  in  that  modelling  a  memory  phenomenon  or   component  means  specifying  and  explaining  underlying   assumpNons  and  resolving  ambiguiNes.   3/21/12

Intro to Cognitive Science

10

8.  The  Modal  Model  of  memory   Early  memory  model  (Atkinson  &  Shiffrin,  1968)  proposes   the  following  general  sequence  of  processing  steps:   1.  InformaNon  enters  from  the  environment   2.  Brief  processing  in  sensory  systems   3.  InformaNon  goes  to  a  short-­‐term  store  (STS)   –  Passes  informaNon  in  and  out  of  LTM   –  Workspace  to  perform  operaNons,  select/  rehearse   informaNon   –  Items  would  be  learned  if  held  in  this  store  

4.  InformaNon  is  output  OR  goes  into  long-­‐term  store  (LTS)   3/21/12

Intro to Cognitive Science

11

8.  The  Modal  Model  of  memory  

From Baddeley textbook 3/21/12

Intro to Cognitive Science

12

9.  Difficul-es  with  the  Modal  Model   While  a  simulaNon  of  the  modal  model  performed  well   on  rote-­‐learning  tasks,  it  ran  into  trouble  when   compared  to  some  aspects  of  human  performance.   Based  on  the  model’s  connecNon  of  STS  and  LTS,  we   would  predict....   •  A  STS  deficit  will  lead  to  impairment  in  the  LTS,   because  the  STS  is  key  for  transferring  informaNon   into  and  out  of  the  LTS  (encoding  and  retrieval)   •  If  STS  is  a  “workspace”  for  manipulaNng  informaNon,   a  STS  deficit  would  impair  a  person’s  ability  to  do   complex  cogniNve  tasks  (e.g.  reasoning)   3/21/12

Intro to Cognitive Science

13

9.  Difficul-es  with  the  Modal  Model,  contd.   However,  empirical  evidence  contradicted  both   predicNons  (these  are  just  two  examples)   •  A  case  in  Shallice  &  Warrington  1970  reported  a   paNent  with  severe  deficits  characterisNcs  of  the  STS,   but  who  also  had  an  unimpaired  LTS.     •  Other  paNents  had  impaired  STSs,  but  managed   various  complex  tasks  like  taxi-­‐driving  or  running  a   business  (Vallar  &  Shallice,  1970).   Clearly,  the  Modal  Model  was  too  simplisEc  and  the   proposed  components  needed  to  be  altered  in  some   way.   3/21/12

Intro to Cognitive Science

14

10.  The  Baddeley-­‐Hitch  model:  Pu[ng  the   “work”    in  working  memory.....  

The  Baddeley-­‐Hitch  (abbreviated  B-­‐H)  model  of  memory   is  in  part  a  “response”  to  the  shortcomings  of  the   earlier  Modal  Model.   Baddeley  and  Hitch  conducted  further  work  to  try  to  find   out  more  about  the  underlying  nature  and  funcNons   of  STM  (see  the  Memory  textbook  chapter  3  for  the   details)   The  new  and  more  complex  model  was  of  working   memory,  and  assumes  that  intermediate  maintenance   and  manipulaNon  of  informaNon  is  necessary  for  many   complex  tasks,  such  as  adding  digits  in  one’s  head.   3/21/12

Intro to Cognitive Science

15

10.  The  Baddeley-­‐Hitch  model:  Pu[ng  the   “work”    in  working  memory.....  

In  Alan  Baddeley’s  own  words,   “The  emphasis  on  ‘working’  aimed  to  dissociate  it  from   earlier  models  of  STM,  which  were  primarily   concerned  with  storage,  and  to  emphasize  its   funcEonal  role  as  a  system  that  underpins  complex   cogniEve  acEviEes,  a  system  that  supports  our   capacity  for  mental  work  and  coherent   thought”  (Memory,  Ch  3,  p43).   This  more  complex  and  acNve  view  of  memory  as   enabling  mental  work  is  one  contradicts  the  way  that   we  ooen  treat  memory  in  everyday  speech-­‐-­‐  as  a   staNc  photo  album  or  filing  system.   3/21/12

Intro to Cognitive Science

16

11.  The  Baddeley-­‐Hitch  model  of  memory   The  B-­‐H  model  (Baddeley,  2003):   -­‐    modified  and  updated  several   Nmes,  remains  a  widespread   and  useful  explanatory  tool.     -­‐  different  components  for   different  types  of  informaNon,   unlike  earlier  Modal  model.     More  typically  called  the   mulEcomponent  theory   The  components  connect  to  one   another  as  shown.        -­‐  Arrows  indicate  in  which   direcNons  informaNon  can   move.   3/21/12

Intro to Cognitive Science

17

12.  The  B-­‐H  model  components   The  components  are  :   1.  A  phonological  loop  for  processing  and  encoding  verbal  and   auditory  informaNon   2.  The  visuo-­‐spaEal  sketchpad  for  processing/encoding  visual   and  spaNal  informaNon   3.  A  central  execuEve  as  an  “aEenNonal  controller”  to  direct  the   “work”  performed  by  working  memory.  As  we  discussed  in   the  aEenNon  unit  more  generally,  this  controller  selects,   manipulates,  and  switches  between  pieces  of  informaNon.   4.  A  mulN-­‐dimensional  episodic  buffer  which  enables  the  WM   subcomponents  to  communicate  with  LTM   5.  LTM  as  a  “crystallised”  system  of  long-­‐term  knowledge.  We   will  not  discuss  this  part  of  the  model  further.     3/21/12

Intro to Cognitive Science

18

13.  The  phonological  loop   Component  of  the  B-­‐H  model  which  temporarily  stores  speech  or   verbally  encodable  informaNon   The  phonological  loop  is  composed  to  two  subcomponents:        -­‐  A  short-­‐term  store  of  limited  capacity        -­‐  An  ar-culatory  rehearsal  process  which  repeats  (rehearses)   informaNon  to  maintain  it  in  memory   InformaNon  goes  into  this  store  if…   •  It  is  from  speech  that  we  physically  hear   •  It  is  “read  out”  from  a  short-­‐term  memory  trace     -­‐  For  instance,  sub-­‐vocally  rehearsing  an  item.  This  is  essenNally  re-­‐ entering  it  in  the  loop  again!  

•  You  see  visual  informaNon  but  remember  by  “saying  it  to   yourself”  (digits,  leEers,  nameable  objects)   3/21/12

Intro to Cognitive Science

19

14.    For  which   phenomena  can  the   phonological  loop   account?   1.  The  phonological   similarity  effect     •  People  show  poorer   recall  for  a  list  of   similar  sounding   items   •  Has  been  suggested   that  these  individual   items  may  have  few   disNnguishing   features  and  are   easier  to  confuse   3/21/12

Intro to Cognitive Science

20

2.  The  word  length  effect  

- Verbal memory span decreases as participants are asked to recall longer words (shorter words are easier) - Suggested that this is because each word takes more time to rehearse in the phonological loop and to recall - A robust and wellreplicated finding! 3/21/12

Intro to Cognitive Science

21

15.  The  visuo-­‐spa-al  sketchpad   This  store  is  responsible  for  image-­‐based  informaNon,  and  is  also   subject  to  memory  span  limits.   Appears  to  consist  of  two  subsystems  which  work  closely   together  to  hold  and  manipulate  informaNon   •  Memory  for  objects  and  their  features  (what)   •  Memory  for  spaNal  locaNons  (where)   Note  the  similarity  of  this  disEncEon  to  the  organisaEon  of  the   visual  system!   Special  tasks  have  been  devised  to  selecNvely  study  one  aspect   or  the  other,  see  textbook  

3/21/12

Intro to Cognitive Science

22

15.  The  visuo-­‐spa-al  sketchpad  -­‐  task   One  illustraNon  of  a  span  limit  on  the  sketchpad  is  visual  pa^ern   span  (Della  Sala  et  al.  1999)          -­‐  ParNcipants  are  shown  matrices  with  some  cells  shaded  in,   required  to  recall  them   -­‐  Matrices  start  small  (2x2)   and  are  increased  in  size   unNl  parNcipant’s  recall   span  is  reached  

3/21/12

Intro to Cognitive Science

23

16.  A  new  addi-on:  the  episodic  buffer   The  original  B-­‐H  model  struggled  to  explain  interacNons  between   working  and  long-­‐term  memory.   The  episodic  buffer  suggested  as  addiNonal  component  to  link   memory  subsystems  with  percepNon  and  LTM  input/output   •  A  mulN-­‐dimensional  (mulN-­‐modal)  space     •  Streams  of  informaNon  from  various  modaliNes  such  as   sound,  imagery  are  bound  into  objects,  scenes,  and  other   meaningful  units   Binding  is  the  process  of  linking  individual  features  into  objects     •  Also  can  apply  to  creaNng  a  meaningful  sentence  out  of  words   •  This  module  could  explain  the  memory  span  finding  that  we   can  remember  twice  as  many  words  in  a  sentence  as  we  can   individual,  unrelated  words.   3/21/12

Intro to Cognitive Science

24

17.  The  central  execu-ve:  `Control  Tower’  for  memory   Central  ExecuEve  is  an  a^en-onal  controller:  aEenNon  crucial  for   deeper  processing  of  informaNon,  and  for  binding  different  types   of  informaNon  together.     •  The  execuNve  selects  which  possible  pieces  of  informaNon  to   manipulate  (the  “work”  in  working  memory!)   •  Switches  between  tasks  (not  necessarily  a  funcNon  specific  to  the   memory  system)   •  Is  finite  -­‐  its  funcNons  and  the  informaNon  it  manipulates  are   limited  by  a^en-on.  Resources  are  finite,  and  processing  has  an   associated  cost.   Many  of  these  funcEons  discussed  more  generally  in  the  aFenEon   unit  and  reappear  in  other  cogniEve  topics.     The  central  execu2ve  as  a  director  of  mental  work  is  an  important   concept  with  far-­‐reaching  implica2ons  outside  of  memory.     3/21/12

Intro to Cognitive Science

25

18.  Back  to  the  big  picture  of  the  Baddeley-­‐ Hitch  model    

1.  Phonological  loop  for  speech  or  any   informaNon  that  can  be  verbally   encoded  and  rehearsed   2.  Visuo-­‐spaEal  sketchpad  for  image   informaNon,  objects  and  features,   spaNal  locaNons/relaNonships   3.  Central  execuEve  for  selecNng  and   controlling  informaNon  flow   4.  Episodic  buffer  for  addiNonal  mulN-­‐ dimensional  storage  and  binding   informaNon   5.  Long-­‐term  memory  for  prolonged   storage  of  declaraNve  and  non-­‐ declaraNve  informaNon   3/21/12

Intro to Cognitive Science

26

References   Course  materials    Memory  (Baddeley,  Eysenck,  &  Anderson,  2009)   Other  references    The  magical  number  seven,  plus  or  minus  two  (Miller,  1956)    Working  memory:  looking  backward  and  forward  (Baddeley,   2003)    An  experimental  invesNgaNon  of  memory  span  (Brener,  1940)    Seven  plus  of  minus  two:  A  commentary  on  capacity   limitaNons  (Shiffrin  &  Nosofsky,  1994)    Human  memory:  A  proposed  system  and  its  control  processes   (Atkinson  &  Shiffrin,  1968)    Working  memory  (Baddeley  &  Hitch,  1974)   3/21/12

Intro to Cognitive Science

27

Smile Life

When life gives you a hundred reasons to cry, show life that you have a thousand reasons to smile

Get in touch

© Copyright 2015 - 2024 PDFFOX.COM - All rights reserved.