Transcript Component process model of memory

Cognition of purposeful action
• Overview
– the focus of this lecture is on the performance of
skilled action in a variety of situations
– how does knowledge/memory get converted into
action?
– How does this knowledge get encoded into
memory?
Cognition of purposeful action
• Effects of practice on performance
– it is well established that skilled performance
improves with practice
– but how should we conceptualize this change?
Cognition of purposeful action
• Singley & Anderson (1989)
– studied development of text-editing skill
– participants were secretarial students who were
skilled typists, but had not yet used word
processors
– given practice over 6 days, 3 hours/day editing a
written manuscript containing 6 changes per
page
Cognition of purposeful action
• Singley & Anderson (1989)
– Dependent variables
measured total time to make edits
thinking time: whenever more than 2 seconds elapsed
between successive keystrokes
keystroke time: whenever less than 2 seconds elapsed
between keystrokes
Cognition of purposeful action
• Singley & Anderson (1989)
– Results
most of the improvement is a result of a reduction in
thinking time
keystroke rate did not change substantially
in other words skill acquisition starts out with a large
cognitive component that decreases with practice
Cognition of purposeful action
Minutes per page
Improvement in text editing
10
8
Total time
Thinking time
keystroking time
6
4
2
0
1
2
3
4
Days
5
6
Cognition of purposeful action
– Generalization
this and many other studies suggest that skill
acquisition starts out with a large cognitive component
that decreases with practice
eventually with enough practice the cognitive
component of practice is squeezed out entirely and
there is only an automated motor routine
Cognition of purposeful action
• Power law of learning
– many studies have investigated how memory
improves with massive amounts of practice
– e.g., Pirolli & Anderson (1985)
subjects practiced memory for 15 sentences for 25
days, 2 hours per day
Memory for actions
• Pirolli & Anderson (1985)
e.g., targets. The doctor hated the lawyer; the radical
touched the debutante; the sailor shot the barber
e.g., lures. The doctor touched the barber; the radical
shot the lawyer
– Results
time to make a recognition judgement; improvement is
rapid initially, but rate of improvement slows down with
amount of practice
Memory for actions
– Results
When the data are plotted on a log-log scale the data
are fitted very well by a straight line
in other words you take the log of recognition time and
the log of days of practice
What this means is that the relation between practice
and performance is fitted by a power function
Memory for actions
– Results
Power law of learning because the function is a power
function
Time to Respond = Practiceb
T = 1.40 P -.24
Note:
ln T = ln 1.40 -.24 ln P
 ln T = .34 -.24 ln P
Memory for actions
– The power function of learning appears to fit to a
wide variety of complex skills
– e.g., Asimov’s writing skill appears to be fit by a
power function
Asimov wrote 500 books over a 40 year period
he sat at his keyboard every day from 7:30 am to
10:00 pm
What this means is that his speed of writing books
increased throughout his career
Cognition of purposeful action
• Stages of skill acquisition
– Fitts (1964) proposed that skills go through three
stages as they develop
cognitive: work from instruction or an example; use of
verbal cueing prominent
associative: more direct representation of what to do;
probably still verbal
autonomous/automatic: cognitive involvement drops
Cognition of purposeful action
• Stages of skill acquisition
– cognitive stage
conceptualized as a type of problem solving
how is it that people can go from some initial limited
factual information to their first solutions of problems in
that domain?
E.g., learning to apply salve to both hands if you are
hemiparetic (Goldenberg)
e.g., learning to use a novel tool (Goldenberg)
Cognition of purposeful action
• Stages of skill acquisition
– overview of problem solving
Newell & Simon conceptualize problem solving as
consisting of using operators to achieve goals
operators are procedures for changing a current
situation into something that is closer to goal
Cognition of purposeful action
• Stages of skill acquisition
– types of operators
difference reduction: apply an operator that moves you
closer to goal state
e.g., cockroaches flee light, people cook meals by
selecting steps that bring them closer to cooking a
meal
Cognition of purposeful action
• Stages of skill acquisition
– types of operators
operator subgoaling: when trying to achieve a goal,
there are times when a precondition of applying an
operator is not satisfied; then need to set a subgoal of
achieving the precondition
e.g. bicycle tire soft; goal: inflate tire; failed
precondition (no bicycle pump); subgoal, find bicycle
pump
Cognition of purposeful action
• Stages of skill acquisition
– Associative stage
as people become practiced in a skill, they recognize
directly what they had to think through
Logan (1988) proposed that as people become
practiced, they learn a solutions to problems, which
are stored, and then can be retrieved directly
no longer solve problem, but retrieve needed
information
Cognition of purposeful action
• Stages of skill acquisition
– Associative stage
Raichle, Fiez, Vidden, MacLeod, Pardo, Fox, Peterson
(1994)
generate associates to a word like apple
this task is repeated several times; with practice
people give the same response (e.g., peel)
brain activation: anterior cingulate and other frontal
areas initially; with practice, more posterior regions of
brain are activated
Cognition of purposeful action
• Stages of skill acquisition
– Autonomous stage
characteristics: requires less attention; difficult to
interrupt; may not be accessible to consciousness
– motor program: a prepackaged sequence of
actions
signing your name
making a hammering action
Memory for actions
• Motor programs
– open-loop versus closed-loop performance
closed-loop system: wait for feedback from one action
before performing next action
open-loop system: execute a sequence of actions
before checking to see whether earlier actions
achieved their intended effects
3 lines of evidence that motor programs are open-loop
at the cortical level
Memory for actions
• Motor programs
– evidence that motor programs are open-loop at
the cortical level
slowness of closed-loop behaviour (200 ms simple
RT)
e.g., skilled typists, pianists, violinists etc. execute
action sequences much too quickly
movements appear to be planned in advance
e.g., time to initiate typing of a word increases with
word length; movements cannot be easily stopped
once initiated
Memory for actions
• Motor programs
e.g., movements cannot be easily stopped once
initiated
Slater-Hammel (1960) had subjects view a sweep
timer that made one revolution per second; task, stop
timer when it reached a certain position by raising
finger from a key
evidence suggests that you need to send a signal at
least 250 ms before it reached target; if sweeper
stopped less than 250 ms on its own subjects could
not stop their finger from moving
Memory for actions
• Motor programs
– deafferentation studies of monkeys
these studies eliminate sensory input by cutting
through dorsal roots of the spinal cord
this eliminates all sensory feedback from the limbs
although movement is unimpaired
results: animals can learn new movements and can
perform them successfully after they are learned in the
dark
Memory for actions
• Motor programs
– motor programs are general, not specific
sequences of behaviour
e.g., people can write with different limbs (hands, feet,
mouth)
e.g., hammering can take place on different planes
Memory for actions
• Motor programs
– learning of motor programs
Keele has proposed that new motor programs are
learned by stringing together individual actions that are
under individual control so that with practice
sequences of actions become bundled together into a
motor program that can be executed without outside
control
Memory for actions
• Motor programs
– Jenkins, Brooks, Nixon, Frackowiak, &
Passingham (1994) performed a study in which
subjects learned to push a sequence of 8 buttons
using PET
– Results showed that early on the lateral
prefrontal area and the posterior parietal cortex
were active, whereas with practice the
supplementary motor area and hippocampus
were active
Memory for actions
• Motor programs--Schema theory
– Schmidt hypothesizes that the learner develops
to representations of a skill
recall memory: the motor program itself
recognition memory: representation of the desired
outcome of the action -- response-produced feedback
and external sensory consequences
Memory for actions
• Motor programs--Schema theory
– Schmidt hypothesizes that recall memory (the
motor program) is improved by comparing the
action produced to the internal standard of
recognition memory
– this implies that recall memory can improve even
when there is no external feedback
Introduction
 Apraxia
– inability to perform skilled or learned movements
that cannot be attributed to a language
comprehension disorder, an elementary motor
deficit, or a sensory deficit
Brief, selective historical review
 term coined in 1871 by Steinthal; agnosia
 Liepmann in early 1900s formulated the
original description of apraxia
 skilled movement was primarily mediated
through left hemisphere
 ideational apraxia, a disruption of the idea or the
representation of the plan of the movement and
 ideomotor apraxia, plan is intact, but it cannot be
implemented or produced
Symptoms of Apraxia
 Conceptual apraxia
 errors in tool selection or use
 impaired recognition of gestures
 impaired in producing sequence of actions
Symptoms of Apraxia
 Ideomotor apraxia (cont’d)
 errors in spatial or temporal aspects of actions
 impaired imitation and performance to command
 Kinematic analyses show abnormal joint angles
and abnormal kinematics in tool use
 Deficits in declarative knowledge of manipulation
actions, mechanical problem solving, and
difficulty learning new actions
Learning from viewing
 How can new actions be learned when they
are viewed?
 To answer this question, I’ll begin with a review
of the dorsal and ventral visual streams
 Then afterward will postulate the existence of two
separable streams, a grasp-to-use system, which
is impaired in apraxics and a grasp-to-move
system, which is impaired in optic ataxics
Dorsal and Ventral visual streams
Ventral and Dorsal Streams
 Ventral stream (what stream)
 Associated with object recognition and form
representation
 Connected to medial temporal lobes and dorsal stream
 Dorsal stream (where/how stream)
 Function of dorsal stream less certain\
 Originally thought to be associated with spatial
awareness and spatial grasping
 Goodale and Milner proposed it is associated with howto-information
Limitations of ventral-dorsal model
 It has been argued by some investigators (e.g.,
Buxbaum) that the dorsal stream is overly simplistic
and that there are two different dorsal streams
 One stream is specialized for grasp-to-move; the
other for grasp-to-use an object
 Grasp-to-move- moving an object (e.g., hammer)
with your hand from one location to another
 Grasp-to-use – picking up and object with the
intention of using it (skilled use or manipulation of
the object)
Two dorsal routes
• Grasp-to-use
• Imagining, planning, or
making judgments about
skilled use
• Hand postures are skilled
• Postures based on internal
models
• Intrinsic frame of reference
• Ideomotor apraxics have
damage to this system
• Grasp-to-move
• On-line system that
enables one to reach and
grasp objects
• Adjustments based on
visual system
• Hand postures based on
structural characteristics of
object (external)
• Extrinsically driven
• Optic ataxics have
impairment to this system
Support for two dorsal route model
 It has been shown that ideomotor apraxics are
impaired on pantomime of skilled gesture,
recognition of skilled gesture, and skilled gesture
use
 However, there is evidence that px with skill gesture
use impairment may have intact grasp-to-move
 Conversely px with optic ataxia (an impairment in
visually-guided reaching) may be intact in skilled
grasp-to-use (see Buxbaum, 2006 for a review)
Representational model of apraxia
 The next slide shows the representational model of
apraxia
 An important feature of this model is that it
postulates skilled actions may be mediated
(semantically)
 In addition, note that skilled actions can result
directly from a visual input to a motor program
 Thus, according to this model it may be possible to
produce skilled actions without semantic knowledge
of the action
Visual/Object
Input
Visual/Gestural
input
Visual analysis
(motion)
Visual analysis
(static)
Structural
Description
System
Action input
lexicon
Auditory/
Verbal
Input
Auditory
analysis
Semantic
knowledge
Action output
lexicon
Motor
programmes
Representational model of praxis (from Chainay
& Humphreys, 2002)
Tactile
Analysis
Tactile/
Proprioceptive
Input
Support for conceptual vs ideomotor
distinction
 As described in the next slides there is evidence
that conceptual action impairment and ideomotor
apraxia dissociate
 As will be reviewed some px with ideomotor apraxia
may not have conceptual action knowledge
impairment (Ochipa, 1992)
 Also, it has been reported that px with conceptual
action impairment may not have gestural
impairment (Riddoch & Humphreys, 1987)
Case study Ochipa et al. 1989
• History
– 67 year old left-handed male
– high school education
– sudden onset of left hemiparesis & speech
difficulty
– CT scan 1 week post onset showed infarct in
right middle cerebral artery involving frontal,
superior temporal, and inferior parietal lobes
Case study Ochipa et al. 1989
 Reason for referral
 investigate the reason for why patient showed
inappropriate use of actual tools in hospital
environment (e.g., eat food with a toothbrush)
 Tests administered
 Western Aphasia Battery
 Praxis Test Battery
Case study Ochipa et al. 1989
 Results
 unable to identify tool when function described
(7/20)
 unable to describe verbally function of tool given a
visual picture of tool (3/20)
 pantomime to command (0/20) (errors: no
response or irrelevant movement; not production)
 pantomime to imitation extremely poor (4/20)
 performance improved only marginally when
allowed to hold tools prior to gesturing tool fx
Case study Ochipa et al. 1989
 Conclusions
 inability to perform a skilled movement-- yes
 not attributable to language comprehension
problem (WAB; can point to tool given name)
 not attributable to a sensory problem (can name
tool given picture)
 results suggest there was a loss of knowledge
about the function of a tool, and the actions
associated with a tool
Semantic or Conceptual Memory
 Semantic memory
 composed of knowledge about the world
including facts, concepts, and beliefs
 knowledge is shared by a culture rather than
episodic or autobiographical memories that are
unique to an individual and tied to a specific time
or place
 How should we conceptualize semantic memory
and what relation does it have to praxis?
Action semantics
 knowledge of the function of tools and objects
 tool, used to provide mechanical advantage in an
action;
 object, recipient of an action
 knowledge of actions independent of tools,
the association between tools and actions,
and the association between tools and
objects
 mechanical advantage of tools
Action semantics
 Action semantics (cont’d)
 knowledge about the organization of single
actions into sequences
 knowledge of symbolic meaning of actions
Conceptual apraxia empirical
 Ochipa et al. (1992)
 Purpose: a) to determine whether AD px
have conceptual apraxia; and b) whether
conceptual apraxia can be dissociated from
linguistic semantics and praxis production
 Method
 32 px with prob AD (NINCDS-ADRDA criteria);
and equal number of intact controls
Conceptual apraxia empirical
 Tests administered
 Descriptive or Screening: MMSE, Problem solving,
apperceptive agnosia (match plaster mold to tool)
 Grouping: ideomotor praxis test (imitate gestures);
semantic language test (auditory comprehension:
given a single word select target picture from an
array of 4 pictures
Conceptual apraxia empirical
 Experimental procedures
 tool-object relationship: perform action: with tool
present, object present, tool and object present
 tool selection: select correct tool from array of 5
for a partially completed task
 mechanical knowledge (a) use an alternate tool;
and (b) solve mechanical puzzles
 Primary Result
 even the group with good praxis and no semantic
language impairment were impaired on 3 of the 6
conceptual praxis tasks
Is semantic knowledge needed to
produce skilled action?
• Riddoch & Humphreys (1987) investigated a
patient with optic aphasia
– Patients with optic aphasia are unable to access
semantic knowledge about visually presented
objects, but they are often able to gesture their
use
– Case study of patient JB
Is semantic knowledge needed to
produce skilled action?
•
•
•
•
•
•
Results (JB)
Visual modality, naming
Tactile modality, naming
Visual modality, gesturing
Spoken name, gesturing
Naming an auditory def’n
45%
75%
75%
93%
100%
Is semantic knowledge needed to
produce skilled action?
• Discussion of Results
– JB is not anomic because he can name objects in
different modalities
– Recall anomia refers to an impairment in naming that
afffects all modalities
– JB does not have an agnosia because he can gesture
appropriately to visually presented objects
– JB has optic aphasia—poor ability to access semantic
knowledge from visually presented objects; impairment is
modality specific; gesture to visual objects spared
Is semantic knowledge needed to
produce skilled action?
• Further results showed that JB has no major deficit
in low-level visual processing
– JB can copy visual drawings
– JB can classify drawings as meaningless or meaningful
• JB is impaired in his ability to match pictures from
audition
– (for semantically similar material: hand, arm, foot, leg)
– Exp’ter names one stimulus; JB had to point to
appropriate picture (70% performance)
Is semantic knowledge needed to
produce skilled action?
• Theoretical interpretation
– JB is able to gesture accurately because
accurate description can be made on the basis of
structural nonsemantic information
– See representational model of apraxia