Transcript lecture 05

Implicit memory
• Schacter and many other researchers distinguish
between implicit and explicit memory
• This lecture will begin by surveying some historical
observations regarding implicit memory
• It will then focus on a particular type of implicit
memory, habits or skills, that are mediated by the
basal ganglia
• Particular attention will be given to lesion studies
and studies of patients with Parkinson’s disease,
which is associated with impaired basal ganglia
functioning
Implicit memory
• implicit memory
– occurs when previous experiences facilitate
performance on a task that does not require
conscious recollection of those experiences
• explicit memory
– occurs when the task requires conscious
recollection of past experiences
Implicit memory
• Historical survey
– Maine de Biran proposed that after sufficient
repetition a habit can be executed automatically
and unconsciously without awareness of the act
itself or of the previous episodes in which the
habit was learned
– Maine de Biran also developed a form of multiple
memory system
Implicit memory
• Historical survey
– Carpenter (1874) noted the importance of
autobiographical recognition or awareness in
normal memory
– Neurology
Korsakoff noted that amnesic patients were affected
by previously experienced events even though they
were consciously unaware of these events
Claparede refusal of amnesic to shake hands after he
had pricked her hand
Implicit memory
• Historical survey
– Neurology
Schneider 1912 showed that amnesics required less
information across learning trials to identify fragmented
pictures
Implicit memory
• Historical survey
– Psychiatry
Freud and Janet investigated patients suffering
hysterical amnesia as a result of emotional trauma
these patients could not explicitly remember the
traumatic event, but their memories of these events
were expressed indirectly (implicitly)
Janet --hysterical amnesia consists of: 1. the inability
of a subject to evoke memories consciously and
voluntarily; and 2. Automatic, compelling, and untimely
activation of these memories
Implicit memory
• Modern research on implicit memory
– effects of subliminally encoded stimuli
several studies have shown that stimuli that are not
represented in subjective awareness (consciously) are
nevertheless processed to high levels by the
perceptual system
Implicit memory
– e.g., Kunst-Wilson & Zajonc (1980)
presented geometric shapes for 1 ms; claimed it was
too brief a period of time to permit its perception;
subsequently tested recognition (forced choice) and
preference (forced choice)
Results: recognition (RN) was at chance; Subjects
preferred the previously presented geometric shape
Implicit memory
– e.g., Bargh & Pietromonaco (1982)
presented ‘hostile words’ and then later had
participants rate a target person
results showed that explicit recognition memory of
‘hostile words’ was at chance, although ratings of the
target person were more negative than those who did
not receive prior exposure
e.g., divided attention study of Eich (1984)
Implicit memory
– e.g., divided attention study of Eich (1984)
auditory divided attention task
unattended channel -- presented homophones (e.g.,
taxi fare)
subsequently participants showed no explicit RN
memory in yes/no task, but tended to spell
homophones in biased direction compared to baseline
performance
Implicit memory
• Learning and conditioning without awareness
– participants learn rules or contingencies without
explicit memory for them
– this phenomenon was studied in multitrial
learning experiments and in classical
conditioning experiments
Implicit memory
– Implicit learning studies of Reber
subjects were presented letter strings that were
organized according to rules of an artificial grammar
Reber reported that subjects could categorize these
strings correctly even though they were unable to
consciously aware of the rules
Implicit memory
– Repetition priming effects
facilitation in processing of a stimulus as a function of
recent prior exposure to the same stimulus
note: repetition priming has been observed under a
wide variety of test conditions, none of which require
explicit reference to a prior study episode
lexical decision (word/nonword) -- DV = latency
word identification
word stem or fragment completion (e.g., __ ss__ss__)
Implicit memory
– Repetition priming effects
– 1. used to study nature of lexical representation
e.g., studied the effects of auditory presentation on
subsequent word identification and lexical decision
tasks
results showed little or no priming
e.g., morphologically similar words (e.g., seen)
facilitate priming of (sees), but visually similar words
do not prime each other (e.g., seen) versus (seed)
Implicit memory
– Repetition priming effects
– 2. used to study relation of implicit to explicit
memory
– stimulated by studies of Warrington and
Weiskrantz on amnesics
– this study showed that amnesics showed
excellent retention when they were asked to
complete three-letter stems of recently presented
words even though their yes/no recognition
memory was impaired
Implicit memory
– Repetition priming effects
– 2. used to study relation of implicit to explicit
memory
several studies have shown that variations in level or
type of study processing have differential effects on
priming versus remembering
e.g., Jacoby & Dallas (1981) showed that answering
questions about the meaning of a target word
improved yes/no recognition memory relative to
answering questions about presence of a letter, but
that word identification was unaffected
Implicit memory
– Repetition priming effects
– 2. used to study relation of implicit to explicit
memory
modality shift
Jacoby & Dallas (1981) showed that changing from
auditory (at study) to visual (at test) severely
attenuated priming effects as assessed by word
identification, but had little effect on yes/no recognition
performance
Implicit memory
– Repetition priming effects
– 2. used to study relation of implicit to explicit
memory
other factors that have been manipulated
study-test delay--variable effects
manipulations that affect retroactive and
proactive interference (and hence explicit
memory) have little effect on word-stem or
word fragment completion
Implicit memory
– Repetition priming effects
– 2. used to study relation of implicit to explicit
memory
other factors that have been manipulated
study-test delay--variable effects
manipulations that affect retroactive and
proactive interference (and hence explicit
memory) have little effect on word-stem or
word fragment completion
Implicit memory
– Habit or skilled learning
– Several lines of evidence suggest that the
caudate nucleus mediates a particular type of
implicit memory, habit memory or memory for
skills, that are incrementally acquired
– This type of memory can be dissociated from
explicit memory
– Caudate nucleus is part of the basal ganglia
Implicit
Memory
Explicit
Memory
Priming
Neocortex
Nonassociative
Learning
Reflex Pathways
Facts
(Semantic
Memory)
Events
(Episodic
Memory)
Skills and
Habits
Striatum
Simple
Classical
Conditioning
Medial Temporal Lobe
Emotional
Response
Amygdala
Squire and Knowlton (1994)
[Squire (1987): Declarative vs. procedural memory]
Skeletal
Musculature
Cerebellum
Habits and skills
• Characteristics of habits and skills
– Habits are learned (acquired via experience dependent
plasticity)
– Habits are performed repeatedly over time
– Habits are performed almost automatically and
nonconsciously
– Habits tend to involve an ordered, structured, action
sequence
– Habits are elicited by a particular context or sequence
(either internal or external)
– Habits can be habits of thought and habits of motor
expression
Habits and skills
• Background
– Anatomy
– Basal ganglia (BG) plays an important role in
normal voluntary movement
– Input to BG received from cerebral cortex and
output sent to brain stem and via the thalamus
back to the prefrontal, premotor, and motor
cortices
BG in relation to brain structures
• Basal ganglia receives
i/p from cerebral cortex
• o/p from BG goes from
thalamus back to
cerebral cortex
• o/p also goes to spinal
cord
Habits and skills
• Background
–
–
–
–
–
–
–
Basal ganglia
Consists of 4 major nuclei
Caudate nucleus
Putamen
Globus pallidus
Subthalamic nucleus
Striatum consists of caudate nucleus, putamen, and
ventral striatum
BG
• Coronal view of BG in
relation to surrounding
structures
• Medial view of BG: note
left side is anterior;
right side is posterior
Habits and skills
• Background (circuits in BG)
– There are several circuits connected to prefrontal regions
of the cortex; these include
• Dorsolateral prefrontal circuit
– Involved in executive function
• Lateral orbitofrontal circuit
– Plays a major role in mediating empathic, socially
appropriate responses
• Anterior cingulate circuit
– Appears to play an important role in motivated behavior
Habits and skills
• Background (diseases associated with BG deficits)
– Movement disorders associated with basal ganglia
• Parkinson’s disease (described by Parkinson 1817)
–
–
–
–
–
–
–
Hypokinetic disorder
Impaired initiation of movement (akinesia)
Reduced amplitude and velocity of movement (bradykinesia)
Involuntary tremulous movement
Lessened muscular power
Increased muscle tone (rigidity)
Shuffling gait
Habits and skills
• Parkinson’s disease (described by Parkinson 1817)
– Results from an insufficiency of a neurotransmitter (dopamine) in the
substantia nigra
– It has been shown that px with PD have a deficiency in dopamine in the
striatum, most particularly in the putamen
– in 1960s it was shown that injection of L-DOPA, the precursor of dopamine
led to a reversal of symptoms temporarily (for about 5 years)
• Huntington’s disease
– Hyperkinetic disorder
– Excessive movements
– Heritable and results from a gene defect
Habits and skills
• Role of BG in learning and memory
– BG is involved in S-R association formation or
the formation of habits or skills, and this form of
memory is dissociable from declarative memory
• Animal lesion study (see Eichenbaum, 2002
for further information)
– This has been investigated in an 8-arm radial
maze with different task requirements (e.g.,
McDonald & White, 1993)
Habits and skills
• Animal lesion study
– Win-shift maze
– maze is put in a room and the rat is permitted to
encode stimuli around the room to provide spatial cues
– on every trial a food reward is placed at the end of the
maze; rat is released from centre of maze and is
allowed to retrieve food
– optimal performance occurs if the rat enters each arm
only once. This means the rat must have memory for
the mazes that were previously entered
Habits and skills
• Role of BG in learning and memory
– Win-stay maze –
– 8-arm radial maze used, but maze surrounded by a curtain,
and lamps were used to illuminate 4 randomly selected arms
of the maze;
– food was baited in the illuminated mazes only; when an arm
was entered for first time, the maze was rebaited; after second
time the light was turned off and no more food was provided;
– task was to associate a particular stimulus (light) with food
across all trials and it was expected that declarative memory
was not required, but habit memory was required
–
radial arm mazes
• Radial arm maze
configurations to
illustrate win-shift and
win-stay procedures;
• + = rewarded arm
Habits and skills
• Results
– Win-shift
– Rats with hippocampal damage made more errors than controls by
visiting previously visited maze arms
– Rats with striatal damage were unimpaired compared to normal
animals (controls)
– Win-stay
– Normal control rats learned strategy over several sessions
– Rats with hippocampal damage learned somewhat faster than
controls
– Rats with striatal damage were badly impaired and were at about
chance levels of performance even after extended practice
Habits and skills
• Conclusions
– Selective damage to the hippocampus impairs
performance on a task requiring declarative
memory
– Selective damage to the striatum interferes with
habit learning
Sequence learning
• Human research
– Further support for the dissociation between habit
memory and declarative memory
– Sequence learning
Sequence learning
• Human research
– Evidence suggests that amnesics are unimpaired in
manual sequence learning (Reber & Squire (1998).
Journal of Cognitive Neuroscience 10, 248-265
– This perceptual-motor habit was investigated in a serial
response time (SRT) task paradigm
– SRT task. Participants are shown a visual cue in 1 of 4
locations by pressing a key corresponding to the location
of the cue
– Cues follow a repeating sequence of locations (n = 12;
124313214234)
Sequence learning
• Amnesics and controls and controls were given
SRT training. Other unimpaired control groups
(Memory) memorized the sequence of responses
(see Table for details). All groups had their implicit
and explicit memory tested
• The Memory control groups memorized the
sequence of responses by watching the computer
screen and attempting to memorize the sequence
of responses
Sequence learning
Sequence learning
• Results
• Amnesics and controls had decreasing SRT
RTs to practiced SRT across trial blocks
• Next figure presents implicit memory
performance by group
• Note amnesics and controls were the only
groups to show implicit memory
Sequence learning
• Top panel. Open bars
present RT for sequence
S1 (practiced by amn and
con) open bars; shaded bar
presents RT for N1
(unpracticed)
• Bottom panel—difference
between S1 and N1
(measure of implicit
memory
Sequence learning
• This Figure presents for
each group explicit
memory performance,
assessed by
recognition memory
performance of 5
sequences (1 target
and 5 lures) on a scale
from 0 to 100
Sequence learning
• Results
• Note: amnesics showed little evidence of
explicit memory, in comparison to other
groups
• Memory controls (esp. young) showed
explicit memory performance
Sequence Learning
• Filled squares – amnesic
• Open square –combined
performance of young
subjects
• Open triangle – controls
• Open circle – Mem old
participants
• Note: cross-over interaction
Sequence learning
• Conclusions
• Proposed that cross-over interaction shown in
Figure 5 reflects encapsulation of implicit and
explicit memory in different memory systems
• Explicit memory mediated by
hippocampal/diencephalic brain structures
• Implicit memory mediated by motor cortex,
neostriatal, supplementary motor area
Sequence learning
• Human research (PD)
– Wilkinson 2009, Neuropsychologia, 47, 25642573
– Investigated implicit & explicit memory in SRT
task in PDs and controls
– Hypothesized that PDs would be impaired in their
implicit memory for SRTs and would also be
impaired in their explicit memory for explicitly
presented SRTs
Sequence learning
• Human research (PD)
– Wilkinson 2009, Neuropsychologia, 47, 2564-2573
– Exp. 1 – implicit SRT
– Sequence was presented probabilistically to introduce
noise and decrease opportunities for explicit learning
– E.g., one group of Ps were shown 3-1-4-3-2-4-2-1-3-4-12, 85% of the time and 4-3-1-2-4-1-3-2-1-4-2-3, 15 % of
the time
– 10 blocks of trials each block with 120 trials of SRT task
Sequence learning
• Human research (PD)
– Implicit sequence learning assessed by
comparing difference between probable and
improbable sequence Reaction time
Sequence learning
• PDs shown in back;
controls shown in white
• ** designates blocks in
which difference between
probable and improbable
sequence is significant (p <
.05)
• Note: evidence of implicit
learning for controls
stronger than for PDs
Sequence learning
• Human research (PD)
– Explicit sequence learning
– After final implicit SRT block presented, explicit sequence
learning was tested
– Inclusion condition – participants were presented a
sequence of 5 cues from probable sequence and were
instructed to recall the next correct response (Inclusion)
– Exclusion condition – participants were presented
sequence of 5 cues from probable sequence and were
instructed to respond with the next cue that was
inconsistent with the training sequence
Sequence learning
• Results show that for
controls and PDs there was
no significant difference
between inclusion probable
and baseline and between
exclusion probable and
baseline
• Hence no evidence of
explicit memory for
sequence
Sequence learning
• Human research (PD)
– Conclusion – Wilkinson data suggest that PDs are
impaired in probabilistic sequence learning relative to
controls
– There is no evidence for explicit memory for either group
– Findings suggest that BG plays a role in implicit sequence
learning;
– Findings by Reber and Squire, which showed that
amnesics were not impaired in sequence learning,
suggest that a form of habit memory was not mediated by
MTL
Implicit memory- cognitive habits
• Human research: Cognitive habits
– Traditionally the impairment in the basal ganglia was
known to result in motor deficits based on observations of
patients with PD and Huntington’s disease
– Aside PD affects dopaminergic input to the striatum;
Huntington’s disease affects cells in the striatum
– It now appears that BG is involved in acquisition of
cognitive habits that involve the gradual formation of S-R
associations
Implicit memory- cognitive habits
• Human research
– Probabilistic classification learning
– Knowlton and her colleagues have studied px and their
performance on the probabilistic classification task
– weather prediction task
– Participants are presented 1 to 3 of 4 cards and task was
to predict “sunshine” or “rain” based on cues presented;
cards had complex geometric shapes
– Each card and card combination was probabilistically
associated with 2 outcomes
– after making prediction participant was given the actual
outcome (see next figure)
Probabilistic classification learning
• Panel A. Probability of
sunshine for individual
cards
• Panel B. conjoint
probabilities associated
with various card
combinations and
example of trial series
Implicit memory- cognitive habits
• Human research
– Probabilistic classification learning
– Note: task has been designed so that specific memory for
outcomes associated with card combinations is not as
useful, first because there are several card combinations,
and second, because any given card combination may
have an outcome of ‘sunshine’ or ‘rain’ associated with it
– More useful is a general sense of the relationship
between cues and outcomes across trials
Implicit memory- cognitive habits
• Human research
• Probabilistic classification learning
– Results
– Controls
– Over 50 trials normal controls improved from pure guessing (50%) correct to
about 70 % correct (highest performance possible)
– PD px did not show significant learning and performance was worse in px
with more severe Parkinsonian symptoms (Knowlton, 1994, Science, 273,
1399-1402)
– After training PD px and controls were given questions about the task and
nature of the stimuli (declarative tasks); performance of the two groups did
not differ
– Results from amnesics show an opposite dissociation – unimpaired
performance on the weather prediction task and impaired performance on
the declarative memory task (Knowlton, 1994, Learning and Memory 1, 106120; Reber, 1996, Behavioral Neuroscience, 110, 859-869)
Implicit memory- cognitive habits
• Human research
– Interactions between implicit and declarative
memory
– When task conditions favor implicit learning
medial temporal activity decreases as striatal
activity increases (Foerde, 2006; Foerde, 2007)
Implicit memory- cognitive habits
• Human research
– Foerde et al. (2006) PNAS, 103, 11778-11783
– This experiment investigated the hypothesis that the
declarative and habit memory systems compete with
each other in tasks in which both systems are involved
– This raises the possibility that it might be possible to
modulate the operation of these systems by introducing a
secondary task
– Previous research has shown that performing a
secondary task that requires working memory should
decrease declarative memory and thereby bias (increase)
the operation of the habit memory system
Implicit memory- cognitive habits
• Human research
– Prediction then is that dividing attention should affect the
way in which a task is performed
– Overview of procedure and design (see also next figure).
Participants were given weather prediction training (PCT
task) for two different cities based on presentation of two
sets of cues; half the cues for a given city were trained
under single task conditions (ST) and half the cues for
that city were trained under dual task conditions (DT)
– Secondary task was tone counting task in which
participants counted the number of high tones
Implicit memory- cognitive habits
• Human research
– After training participants were given a probe run in which
participants were shown cues that had been trained
under ST or DT conditions and were required to predict
sun or rain (done in fMRI); no tone task administered here
– After scanner, participants were given test of declarative
memory which required flexible knowledge about task
(e.g., participants shown all 4 single cues and asked to
select cue most likely to be associated with sun; task
repeated for two-cue combinations, and 3-cue
combinations)
Foerde procedure
• Tone counting
• Train on weather
prediction (single, dual)
• In DT condition report
number of high tones
• Baseline button push
condition
• Probe run – predict
weather
Implicit memory- cognitive habits
• Results
– Behavioral results shown in next figure showed that PCT
performance during training was higher after run 2 than
run 1 and marginally higher in ST than DT (See a)
– Probe task (see b) no difference between ST and DT in
weather prediction performance in probe task
– Cue selection (test of declarative memory, see c)
participants were significantly better in the ST than DT
condition suggesting that a more flexible memory
representation was obtained
• Fig a. PCT
performance during
training
• Fig b. PCT
performance during
probe task
• Fig c. declarative
memory performance
Implicit memory- cognitive habits
• Results
– fMRI results in next figure (see b) showed that probe
accuracy was correlated with right hippocampus activity
in ST but not DT condition
– Figure d showed that probe accuracy was correlated with
left putamen activity in DT but not in ST condition
– Recall hippocampus associated with declarative memory
– Putamen is part of BG, which is implicated in habit
memory
fMRI results
Implicit memory
• Conclusions
– Memory for PCT can be either MTL or striatal
– Declarative memory depends on executive
functions, but habit learning does not appear to
depend upon it (recall secondary task effects)
– Although PCT performance can be supported by
MTL or striatal systems, the type of memory
system mediating performance affects flexibility
of memory (recall declarative memory test)
Implicit memory
• Conclusions
– In disease states involving dysfunction of the BG
medial temporal lobe may be present when
normally striatum activity would dominate
(Moody, 2004)