Background

1
Lecture 4 Implicit Learning and Memory Overview

• Background
• Implicit Learning
• Implicit Memory
• Systems vs Processes
• By the end of the lecture you should have
  learned
• The difference between implicit learning and
  implicit memory
• The types of paradigms used to assess memory
  explicitly and implicitly
• Evidence for and against systems and process
  accounts of implicit memory effects

2
Background

• Until mid 1980s - memory research concentrated on
  memory system responsible for rapid, single-trial
  learning, which allows a conscious recollection
  of the learning experience.
• This memory system appears to be particularly
  impaired in amnesia.
• However, amnesic patients can perform normally a
  variety of tasks which do not require the
  capacity to recollect the learning experience
• Increasing evidence that memory for recently
  presented information could be demonstrated in
  healthy control subjects in the absence of
  conscious recollection
• This dissociation led to the development of the
  concepts of implicit learning and implicit memory

3
Background

• Various dichotomies have been proposed to account
  for the variety of LTM phenomena
• Knowing How vs Knowing That
• Declarative vs Non-Declarative
• The Implicit / Explicit distinction (Schacter,
  1987) represents the latest of these approaches
  to understanding the nature of the LTS.
• Explicit memory revealed when performance on a
  task requires conscious recollection of previous
  experiences
• Implicit memory revealed when performance on a
  task is facilitated in the absence of conscious
  recollection.
• Implicit memory remains a very active area of
  research

4
Definitions

• Implicit Learning The acquisition of complex
  information (artificial grammars, production
  rules motor skills etc).
• Explicit Memory - requires conscious
  recollection of previous experience
• Tests of Memory Direct Free recall, cued
  recall, recognition
• Implicit Memory - assessed without reference
  to specific learning event
• Tests of memory Indirect Fragment completion,
  repetition priming, lexical decision.
• It is often not clear whether researchers are
  using the terms implicit and explicit to refer to
  memory tests, the state of awareness of the
  subject during task performance, or theoretical
  constructs (e.g. memory systems).
• These terms are essentially descriptive rather
  than explanatory

5
HISTORY

• Early phenomena
• Descartes, Claparade, Freud...
• (Important not to confuse current use of
  implicit with Freudian concepts of
  unconscious.)
• Subliminal Perception
• Hysteria in 60s over subliminal advertising.
• In reality no evidence this ever took place, and
  highly unlikely to have been effective even if it
  had.
• Subliminal Learning
• Self-Help tapes supposed to contain masked
  information.
• No evidence of any beneficial effects.

6
Implicit Learning

• learning complex information without complete
  verbalisable knowledge of what is learned
  (Seger, 1994)
• Any skill which is acquired through
  practice, and is not easily introspected and
  expressed verbally could be termed implicit
  knowledge -
• e.g. language, social skills, perceptual skills,
  motor skills
• Whilst Implicit Memory usually involves
  learning and testing on the same stimuli,
  Implicit Learning usually demonstrates transfer.
• Implicit learning is not limited to perceptual
  / motor skills
• Most research based on 2 paradigms
• 1. Artificial Grammars
• 2. Control of complex systems

7
Implicit Learning
Artificial Grammars
Subjects trained on grammatical sequences, then
presented with grammatical vs non-grammatical
sequences.
8
Implicit Learning
Artificial Grammars Reber (1967) Group A Learn
grammatical letter sequences Group B Learn
random letter sequences (using same
letters) Both groups then shown 44 letter
sequences, 22 of which were gramatical, 22 of
which were random. Group A successfully
categorised 79 of the sequences Group B were at
chance Effect lasts for years (Allen Reber,
1980) When questioned about the nature of the
grammar, subjects generally claim to be guessing
and are unable to report any knowledge of the
rules (at least early on). Many replications
(but effects tend to be smaller)
9
How Implicit is Implicit Learning?

• Are people really implicitly learning the
  grammar?
• Some evidence that subjects may simply be
  learning initial bigrams last letter.
• BUT learning transfers to same grammar with
  different letters (Matthews et al, 1989)
• Towards the end of implicit learning expts, Ss
  can be capable of articulating some useful
  information concerning the rules learned.
• However, performance has often improved before
  this point.

10
Implicit memory tests
Many different methods, but most common
are Priming paradigms Subjects presented with
target words. Subsequent recognition phase
Targets and distractors. Fragment
Completion A--a--in Word Stem Completion Bri---
Implicit memory is evidenced when Ss complete or
identify more studied than non-studied
words. Perceptual Identification Lexical
Decision Degraded Word Naming Implicit
memory evidenced by faster RTs for studied words
Assassin
grsfersd
Ardenisk
Assassin
11
Not all implicit memory tests are verbal Closure
Pictures
12
Evidence for Implicit / Explicit Distinction
1. Neuropsychological evidence Amnesic patients
have been shown to have normal performance on
several implicit memory tasks Performance
reflects the operation of two separable memory
systems Explicit memory system is impaired in
amnesia
2. Dissociations in normals Extensive evidence
that certain manipulations effect performance
when memory is tested one way, but not the other.
Typically, most manipulations influence
explicit, but not implicit memory
13
Evidence for Implicit / Explicit Distinction
Neuropsychological evidence Graf, Squire
Mandler, (1984) Presented amnesic patients and
controls with word lists - Ss made pleasantness
ratings.
Test stimuli for Cued Recall and Stem Completion
identical - E.g. BRI-- only instructions
differed.
14
Dissociations in healthy controls
LOP Effects Jacoby Dallas (1981) Subjects
receive 60 words, 3 types of question Answer Y
or N Contains the letter L ? - WORLD Rhymes
with Drain? - TRAIN Is part of the nervous
system? - BRAIN
Test Standard Recognition (explicit) or
Perceptual Recognition (Implicit)
Standard LOP effect for recognition memory, but
not for perceptual recognition
15
Dissociations in healthy controls
Generation Effect Memory for words generated at
encoding is superior to memory for words which
were read
Electronic calculating device -
mpretcou Electronic calculating device - computer
Generation effect for Recognition, but not
perceptual recognition
16
Dissociations in healthy controls
Forgetting Tulving et al (1982)
Ss learn list of uncommon words (e.g.
Toboggan). Test standard recognition, fragment
completion (_O_O_GA_)
Repetition priming effect equal for recognised
and non recognised words Fragment completion
performance unchanged after 1w
17
Dissociations in healthy controls
Modality shifts Jacoby Dallas (1981) Targets
presented visually at learning, but spoken at
test No effects on recognition memory Significantl
y reduced priming effects in implicit
test. Roediger Blaxton (1987) Changed
typescript between learning and test No effects
on recognition memory Significantly reduced
priming effects in implicit test. DOUBLE
DISSOCIATION Explicit Memory sensitive to LOP
/ Generation effects / Retention interval /
Dividing attention Implicit Memory sensitive to
manipulations of surface features (e.g. modality
shifts).
18
Theoretical Considerations
What is implicit memory? - Is it a system or a
process?
System accounts - favoured by neuropsychologists
observing preserved abilities in amnesics
Process accounts - favoured by experimental
psychologists working healthy subjects (often
using priming paradigms)
System accounts - Tulving Schacter (1994) -
Differences in implicit / explicit memory reflect
the operation of separate memory systems.
Process accounts - Roediger (1990) - Distinction
between Implicit / Explicit memory is confounded
by distinction between the processes required to
perform these tests Explicit tests require
processes that operate on conceptual information,
implicit tests require data-driven processes
that operate on more perceptual level.
19
Theoretical Considerations
Systems accounts If Implicit and Explicit
memory effects represent the function of
separable memory systems, there should be no
correlations between measures of Implicit and
Explicit memory. Stochastic Independence (Sherry
Schacter, 1987) Tulving et al (1982) No
correlation between recognition and fragment
completion. If implicit memory is one system
then there ought to be correlations between
different measures of that systems
performance. BUT No correlations - so lots of
different implicit memory systems? Quasi Memory
(QM) (Schacter) Perceptual Representation
System(s) (PRS)
20
Theoretical Considerations
Systems accounts Neuropsychological Data Most
implicit memory effects observed in amnesia
involve the priming of pre-existing
representations. What about implicit memory for
novel information? Graf Schacter (1985)
Amnesics and Controls learned word pairs BALANCE
- CHAIR Two conditions at test Same context /
Different context BALANCE-CHA___ /
TREE-CHA___ Priming effects in same context
condition e.g. Amnesics demonstrate equal
implicit memory for novel associations BUT - this
is affirming the null hypothesis. Gooding, Mayes
van Eijk, (2000) Meta-analysis. 27 of 36
studies found numerical superiority in controls
over amnesics.
21
Process Accounts Transfer Appropriate
Processing (TAP)
Process accounts An attempt to stop the endless
proliferation of systems. Memory performance
depends on the extent to which the processes used
at the time of learning are the same as those
used at the time of testing c.f. LOP / Morris,
Bransford Franks (Lecture 2 ) Roediger and
colleagues developed the concept of TAP. Argued
that there are two broad types of cognitive
process 1. Data driven (perceptual) processes
The analysis of surface level features 2.
Conceptually driven processes The analysis of
meaning, or semantic information.
22
Process Accounts
According to this interpretation Data driven
(perceptual) processes underlie performance on
most implicit tests Conceptually driven
processes underlie performance on most explicit
tests Avoiding circularity Perceptual vs
Conceptual distinction can be operationally
defined Read vs Generate Manipulation xxxx
- cold vs hot - ? Read gt Generate Data driven
test Generate gt Read Conceptually driven test
23
Evidence for Process Accounts
Not all Implicit memory tests are data driven,
and not all explicit memory tests are
conceptually driven. It is possible to construct
conceptual implicit tests, and perceptual
explicit tests According to TAP, dissociations
should occur between conceptual and perceptual
tests, regardless of their explicit-implicit
status. General Knowledge Test Ss view list of
unrelated words, told to remember them. Then
several intervening distractor tasks. 100 Q
general knowledge test 50 answers were
previously studied words Conceptually driven
Based on knowledge, not perceptual properties,
BUT Implicit Subjects not consciously aware that
their memory is being tested
24
TAP - Support
Blaxton, 1989 Four tests 1. Free Recall
(Explicit Conceptual) 2. General Knowledge
(Implicit Conceptual) 3. Graphemic cued recall
(Explicit Perceptual) 4. Fragment Completion
(Implicit Perceptual)
How do experimental manipulations dissociate
performance on these tasks? Systems 1 3 vs
24 Process 12 vs 34 Both Perceptual and
both conceptual tasks behaved in the same way.
25
Process Accounts - Problems
Neuropsychological data Key prediction of TAP
Amnesia reflects impaired conceptually driven
processes (Blaxton, 1992) Vaidya (1995) 1.
Perceptual Explicit (Word Fragment) 2. Perceptual
Implicit (Word Fragment) 3. Conceptual Explicit
(Word Associates) 4. Conceptual Implicit (Word
Associates) Amnesics should perform badly on
conceptual implicit and explicit tests (34), but
OK on perceptual implicit and explicit tests. In
fact, amnesic patients performed poorly on 13.
26
Process pure tests?
How implicit are implicit memory tests?
It is very difficult to determine whether
performance on implicit memory tests was not
influenced by explicit memory E.g. it is
possible (probable?) that subjects are aware they
are filling in fragments with previously studied
words. Experimenters have attempted to address
this issue through questions at debriefing /
excluding aware participants. Jacoby, Toth
Yonelinas (1993) developed the Process
Dissociation Procedure (PDP) Attempts to measure
the relative contributions of explicit
(conscious) and implicit (unconscious) processes
on recall. This procedure became very
influential, but makes some rather questionable
assumptions.
27
Process pure tests?
PDP procedure Purifying Measures of Memory Learn
2 lists of words LIST A and LIST B - then
recognition test consisting of A words, B words
and NEW words Inclusion condition Say OLD to any
item from list A or list B Exclusion condition
Say OLD ONLY to items from list A. Assumptions
Recognition involves automatic (implicit A) and
conscious (explicit R) components. Prob of an
item being recognised automatically is
independent of the prob of its being recognised
consciously. SO Inclusion p(LIST B OLD)
RA(1-R) Exclusion p(LIST B OLD) A(1-R) By
the magic of maths (simultaneous equations) R
Inclusion - Exclusion A Exclusion/(1-R)
28
PROBLEMS
Problems with implicit / explicit distinction
Terminology are researchers referring to a)
tasks? b) the subjects state of awareness during
tasks? c) theoretical constructs presumed to
underlie task performance?
Amnesic patients cannot do many things which rely
on implicit memory (e.g. learn their way
around, follow a plot etc)
Dichotomy implies two equivalent systems
Implicit memory may reflect the operation of
several distinct learning mechanism (Baddeley,
1997).
29
Summary

• Current memory theory has been described as
  suffering from a taxonomic crisis, if not a
  full-blown case of terminological chaos
  Lockhart, 1989.
• A vast range long term memory experiences /
  processes can be identified
• Various attempts have been made to classify
  these in some way
• Currently a broad consensus agree that a
  distinction can be made between recollective
  explicit memory and non-recollective implicit
  memory
• Opinion is still divided as to whether this
  distinction is best described in terms of
  separate memory systems or different memory
  processes operating within a single memory
  system.
• Neuropsychological data supports an
  interpretation based on memory systems.

30
Neuropsychiatry
DID Dissociative Identity Disorder (DSM-IV)

• Linked to early childhood trauma
• Key symptom is inter-identity amnesia - One
  identity claims amnesia for events experienced by
  other identities.
• Rafaele et al (2002) Tested 31 DID patients on 3
  implicit memory tests. Also tested 25 controls
  and 25 DID simulators
• Material learned as one personality and tested as
  another
• Equal implicit memory effects in all three groups
  for both data driven and conceptually driven
  implicit memory tasks.
• What we did find in both our implicit and
  explicit memory studies was a dissociation
  between objective memory performance and
  patients subjective reports that is, although
  patients indicated no subjective recollection of
  the encoding phase performed by a different
  identity states at all, their test scores
  indicated normal memory functioning
• Thus (according to Rafaele et al) - DID patients
  suffer from a lack of memory meta-awareness.