Transcript Chap4
BHS 499-07
Memory and Amnesia
Models of Memory
Plato’s Model
Plato extended the wax tablet metaphor
to refer to birds in an aviary.
• Birds are located in specific places.
• Search processes are needed to hunt for
them.
William James referred to primary and
secondary memory.
• New experiences linger briefly and need not
be stored forever (secondary memory).
The Modal (Multi-Store) Model of
Memory
Modal refers to sensory modality (way of
receiving info from outside world).
Heuristic means “rule of thumb” – this
theory of stages is a way of thinking
about memory not to be taken literally.
This multistore (modal) model was the
guiding framework for decades.
Multiple Memory Systems
Memory is not unitary but consists of
several subcomponents (parts).
Atkinson & Shifrin:
Tulving’s Triarchic Theory:
• Sensory store, short term, long term stores
• Episodic
• Semantic
• Procedural
Autonoetic (self)
Noetic (formal knowledge)
Anoetic (automatic skills)
Other Classifications
Declarative vs Nondeclarative
• Declarative includes episodic and semantic
•
memory
Nondeclarative includes procedural memory,
classical conditioning and priming
Explicit vs implicit
• Explicit memory involves consciousness,
implicit does not.
Beyond Multi-store Models
Levels of processing theory (Craik &
Tulving) – it isn’t where memories are
processed that matters, but how.
• Shallow vs deep processing
• Elaborative rehearsal vs repetitive rehearsal
Short term memory was replaced by
Baddeley’s model of the central
executive (where rehearsal takes place).
Current Issues
Neurological bases for memory
Impact and importance of emotion on
memory
Use of multiple memory sources (fuzzy
trace theories)
Embodied cognition – how our grounding
in the world influences memory
Sensory Memory
Three sensory registers discussed by
Radvansky and Parkin texts:
• Visual sensory register (iconic memory)
• Auditory sensory register (echoic memory)
• Touch sensory register (haptic memory)
Briefest duration -- < 5 sec
Retains characteristics of the stimulus so
that meaning can be interpreted
Iconic Memory
How many items, how long does it last?
It is difficult to study the capacity of
iconic memory because items fade
before people can report them aloud.
• Averbach found that increased presentation
•
time did not improve memory for dots.
Sperling’s partial report procedure showed
that an entire array was remembered well, but
only for ~.250 seconds, ¼ sec.
Increasing the duration of
the stimulus increases the
dots remembered up to 6,
but not much after that.
Sperling’s partial report procedure showed greater recall
when a tone cued people to look at a specific line immediately
after viewing the stimulus, but not after a delay.
Anorthoscopic Perception
Images in the iconic register are
combined to form a single mental
representation.
Anorthoscopic perception (seeing more
than is there) – an image is passed
through a slit at 250-300 ms (quickly).
• The original icon is compressed to build up
the entire representation – so not an after
image but a constructive process.
The longer a stimulus
is viewed, the closer
the response is to the
actual stimulus (see
Actual and 2110 ms)
Trans-Saccadic Memory
Saccade – an eye movement (~ 30 ms).
• Our eyes are constantly moving over the
objects in the world.
Fixation – when eyes stop on a point
(typically ~ 300 ms).
A trans-saccadic memory is needed to
build-up a mental representation from all
of the eye movements
Subjects were
unable to make the
necessary
comparisons in
either of these two
tasks, so there
must exist a transsaccadic memory
store.
Change Blindness
Visual memory is not always accurate.
• Movies frequently contain errors of detail that
•
go unnoticed – continuity errors
Only 33% of subjects noticed the change of
an actor of the same ethnicity & gender.
Top-down expectations affect what is
noticed.
• Students notice students not construction
workers, or changes that belong in a scene.
Echoic Memory
Echoic memory lasts briefly to permit a
mental representation to be formed.
• More exists in memory than can be reported.
• Retention of info is longer (~ 4 sec).
Because sounds can only be heard
once, info is kept available longer.
Flow of speech is constructed from a
series of passing sounds.
Haptic Memory
Less studied.
Air jets were used in a whole/partial
report procedure.
• Duration was ~ 1.3 seconds.
Short Term Memory
Dispute – is short term memory
qualitatively different from long-term
memory?
• Or is it just the part of long term memory that
is currently active?
Severely limited in capacity (unlike
sensory registers).
• Miller’s magic number of 7 +/- 2 (or 4 +/- 1).
Chunking
We are capable of thinking about more
by forming units out of smaller pieces of
information.
• The number of chunks is the same as the
number of units that can be remembered.
Prior knowledge guides the chunking
process.
• Race lengths, chess games are large chunks
S.F.’s digit span improved with practice as he learned to
chunk digits in terms of race results.
Duration of STM
Without active attention, info is forgotten
in ~ 30 sec (some sources say 15)
• How do you tell people not to think about
something so you can test it?
Is forgetting due to decay or
interference?
• Decay -- passage of time erases trace.
• Interference – new info displaces the old
This task tests for
interference.
This task tests for delay.
STM Retrieval
Does retrieval involve a serial or a
parallel search?
• Sternberg’s paradigm of digits followed by a
•
•
probe showed serial exhaustive search.
Most people find this result surprising,
showing why self-report may be flawed.
An alternative explanation is parallel search
with limited resources (requiring more time).
Most likely both processes are involved.
Serial Position Curves
Position in a list of items affects likelihood of
being recalled.
Primacy effect – items at the beginning of a list
are remembered better.
•
Due to encoding in long term memory via more
rehearsal.
Recency effect – items at the end of the list are
remembered well because still active in short
term memory.
Modifying Serial Position Effects
Placing an irrelevant item at the end of a
list eliminates the recency effect.
Memory for actions is different – no
primacy effect because the focus is on
individual actions so less rehearsal.
Suffix effect – recency effect is
diminished when new info is added at
the end (the more the greater the effect).
Memory for Order of Items
The position of the item must be
remembered, not the item itself.
Slot-based models – info is dropped into
a series of slots, read off in order.
Chaining models – associative links form
a chain (but items can be skipped).
Perturbation model – hierarchy of
chunks is disturbed.
More Serial Order Models
Inhibition models – retrieval starts with the
most active, then inhibits it and goes to the
second most active, and so on…
•
•
Inhibition prevents things just recalled from being
recalled again immediately.
Repetition blindness – people fail to see the same
word presented soon after it was first seen on a
screen.
Context-based models – order tied to context.