Download Chapter 6 Forgetting. - 서울대 Biointelligence lab

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Chapter 6
Forgetting.
Bodrov Alexey
Outline







Introduction
Consolidation Theory
Interference Theory
Release from proactive interference (PI)
Relative Distinctiveness
Discrimination problem
Conclusions
Biointelligence Lab http://bi.snu.ac.kr
2
Introduction
We are told that the famous Athenian Themistocles
was endowed with wisdoms and genius on a scale quite
surpassing belief; and it is said that the certain learned
and highly accomplished person went to him and
offered to impart to him the science of mnemonics,
which was then being introduced for the first time; and
that when Themistocles asked what precise result this
science was capable to achieving, the professor
asserted that it would enable him to remember
everything; and Themistocles replied that he would be
doing him a greater kindness if he taught him to forget
what he wanted than if he taught him to remember.
Marcus Tullius Cicero
Biointelligence Lab http://bi.snu.ac.kr
3
Consolidation Theory
Third
prediction:
Second
prediction: theory is not really a theory of
First
prediction:
 Consolidation

McEwen,
Azmitia
(1972)
tested is
this
 Quatermain,
An
experiment
by Chorover
Schiller
(1985).
forgetting
but
rather
aand
theory
of
Ebbinghause
(1885)
notedand
that
the
rate
ofwhy
forgetting
prediction
the
same
procedure
(see
Placing
a rat
on a raised
platform
slowed
when ausing
period
of
sleep
occurs
studysecond
and
information
is
not
stored
in
a between
first place.
prediction), but they tested all animals 24hr, 48hr, 72hr
test.
 3
key predictions:
after the ECS. The animal stepped off the platform
1) Memory
should
be inbetter,
then, tested
following
 Results:
evidence
of
amnesia
those
animals
24hr,
The rat received a mild foot shock (if the rat stayed nothing happened)
a
rest
period
than
following
a period
48hr
after
ECS, but
no evidence
of amnesia
afterof72hr.
activity.
2) Memory should be worse, then, following
Chorover and Schiller varied the interval between the
some
event
that
prevent
the
consolidation.
This
directly
contradicts
the third
prediction.
So the weaker
learning
episode
(stepping
down)
and administration
of
3) appeared:
If consolidation
is
prevented,
the
item
form
Some that
biological
changes ismay
indeed
affect
ECS.
They
estimated
consolidation
complete
within
should
benicely
recalled
becauseofthe
an
10sorganism
(this never
coincided
with conception
the duration of
essential
storage
phase
was not complited.
information in
short term
memory)
After some period of time the rats was subjecting to electroconvulsive shock


Biointelligence Lab http://bi.snu.ac.kr
4
Interference Theory
John
McGeoch(1940)
was thedemonstrated
“founder” of that
He divided
Abernathy
thetheory.
alternation
of
 Response
competition
occurs
when
two
or
interference
into
parts: retroactive
proactive.
context. In
his2 experiments
the and
students
who were

more items are potential responses to a
tested in query.
the same classroom they were taught in did
memory
better than students who were tested in a different
 Experiment:
Subjects learned to associate a
classroom.
series of items with cues, the A-B learning,
and
Setthen
can be
thought of
a special
versiongroups
of context
“rested”
foras30
min. Other
effects,
with5,the
subject
using
an
inappropriate
mind
 The
A-D
learning
does
not or
erase
the
memories
of theA-D
A-B
had
either
10,
20,
40
trials
learning
set. TheThere
typical
failing to recognize
a
learning.
is example
no loss ofisinformation,
only a lapse
in
paired
associations.
friend from a college when you happen to run out into
a personit.during
a visit
 retrieving
Conclusions
from
thehome.
experiment: The
 McGeoch
identified
mechanisms
can and
cause
difference
in recall3 between
the that
control
interference:
McGeorge’s 1)
influence
was
so profound
that
no
Response
competition
experimental
groups
could
not
be
attributed
theories of forgetting
from long-term
memory
2)
Altered
stimulus
conditions
solely
to decay
competition
from D items.
propose
as main explanation.
3) Set
Biointelligence Lab http://bi.snu.ac.kr
5
Release from proactive interference
(PI)





Experiment
like
BrownPeterson’s was conducted
(the
difference
is
in
switching from consonants
to numbers for a half of
subjects).
Performance in the switched
group was much better that
in control group.
Release from PI is equal to (x/y)*100
It was showed that change in materials was crucial, rather
than the fact that one material is easier then another.
The decay view cannot predict an increase in performance.
Biointelligence Lab http://bi.snu.ac.kr
6
Relative Distinctiveness (1)


Foundation of a theory: items will be well recalled to the
extant they stand out or differ in some fashion from
surrounding items. (If the item at position 6 of a 12-item
list is red and all other items are black, it will be well
recalled).
Experiment:
Brown-Peterson task. Distractive task – counting
backwards for 10s/15s/20s. The proportion of items
recalled correctly: 0.85, 093, 093 (first trial) and 0.33,
0.30, 0,30 (fourth/last trial).
Results inconsistent with a decay explanation.
Change the time for distractive task for all groups to 15s.
Biointelligence Lab http://bi.snu.ac.kr
7
Relative Distinctiveness (2)
Performance: 0.20, 0.28, 0.38.
Release from PI “doesn’t work” (worth after change for one group)

Baddeley (1976) offered an explanation that attributes this
result to the relative, rather than absolute, duration of the
distractor activity.
Murdock (1960) proposed the model defines distinctiveness
as the extent to which a stimulus stand out from other stimuli.
Measure of how different each item is from all other items:
n
 k   j 1 dk  d j

where d – sum of all the interstimulus intervals + retention interval

Result: As the retention interval increases, memory for the
items at the end of the list gets worse and for the items in the
beginning of the list gets better.
Biointelligence Lab http://bi.snu.ac.kr
8
Discrimination problem


Suggests that forgetting, in the sense of permanent loss,
does not occur; there is only a failure to perform because
of a difference in the stimulus conditions prevailing at
encoding and at test.
Eich and Brinbaum (1982) conducted a test, which showed
that in “right conditions” with right retrieval cue subjects
were able to recall many of the items that they previously
could not.
Biointelligence Lab http://bi.snu.ac.kr
9
Conclusions

Each view, permanent loss and temporary laps, has its
proponents and detractors. The permanent loss view
suggests the information is gone forever, whereas the
temporary laps view suggests that under different
conditions with different ways of testing, the information
might still be revealed.

BUT, current theories of forgetting (supporting with tests)
view memory as discrimination problem in which items will
be recalled well if they are distinctive or stand out from
competing items at the time of retrival.
Biointelligence Lab http://bi.snu.ac.kr
10
Thank you.
Biointelligence Lab http://bi.snu.ac.kr
11