Psyc 313

1
Psyc 313
2
III. Retrieval Processes in Long-Term Memory

• A. Availability without Accessibility
• failure to remember information is not
  necessarily due to failure of encoding or a loss
  of information from storage (unavailability)
• retrieval failure can result from the fact that
  the information is available in memory, but not
  accessible for retrieval
• retrieval failure now recognized as a primary
  cause of forgetting
• Tulving and Pearlstone (1966)
• participants encoded categorized lists that
  contained two target words from each of 24
  categories (e.g., Type of Spice garlic,
  parsley)
• participants were instructed to remember the
  target words for a memory test
• some participants were tested with free recall
• these participants recalled substantially less
  than half of the items
• other participants took a cued recall test in
  which the category names (i.e., type of spice)
  were given as retrieval cues
• these participants recalled nearly three-quarters
  of the words
• the info was available, and the right retrieval
  cues made it accessible

3
B. Principle of Encoding Specificity

• What makes a good retrieval cue?
• memory retrieval is aided by a cue to the extent
  that the cue helps reconstruct the encoding
  situation
• i.e., memory depends on the amount of overlap
  between whats happening at retrieval and what
  happened at encoding
• Thomson and Tulving (1970)
• participants encoded weakly-related word pairs
• like plant-BUG the word bug is the word that
  has to be remembered (i.e., the target)
• after encoding a list of such pairs, a
  cued-recall test was given in which a cue was
  presented for each of the targets
• a strongly related word was presented to cue the
  target. (e.g., participant would see insect as
  a cue for bug)
• the word from the original word pair, was
  presented as a cue for bug (i.e., plant)
• the strong associate was a relatively ineffective
  cue for recall relative to the word presented
  with the target at encoding

4
B. Encoding Specificity, cont.

• the encoding specificity principle is similar to
  the transfer-appropriate processing principle
• both emphasize the overlap between encoding and
  retrieval as the key determinant of remembering
• transfer-appropriate processing view focuses on
  encoding, and guides elaboration
• emphasizes that the encoding processes one uses
  should be appropriate for how memory will be
  tested
• about getting information into memory
• encoding specificity principle focuses on
  retrieval, and improves it
• the best retrieval cues are those that tap into
  how something was encoded
• about getting information out of memory

5
Extensions of Encoding Specificity (ES)

• studies manipulating physical context, presence
  or absence of music, odor, drug or alcohol
  intake, and mood at encoding and retrieval have
  all revealed context-dependency effects
• General structure of a context dependency
  experiment
• for a given encoding condition, memory will be
  better if the same context is reinstated at
  retrieval
• condition AA should be better than condition AB,
  and condition BB should be better than condition
  BA
• Note these are the only comparisons relevant to
  ES
• It may seem reasonable to infer that AA should
  also be better than BA, but this wont
  necessarily be the case
• encoding specificity principle states that for a
  given encoding condition, memory is best when the
  retrieval condition matches the encoding
  condition
• a comparison between AA and BA involves different
  encoding conditions, so any differences in memory
  performance cant necessarily be attributed to
  the encoding specificity principle

6
More Extensions of Encoding Specificity

• a. External Context Godden and Baddeley
  (1975)
• deep-sea divers encoded words in one of two
  conditions on a beach or under several feet of
  water
• recall was tested in the same environment in
  which they had encoded the information or in the
  other environment
• results revealed a context-dependency effect
  --- see Table 6.4, p. 229
• b. Internal Context Miles and
  Hardman (1998)
• participants encoded words while pedaling an
  exercise bike vigorously enough to double their
  heart rate or encoded the words while at rest
  (sitting on the bike, but not pedaling)
• physiological context was manipulated in the same
  way at retrieval
• results revealed a context-dependency effect ---
  see Table 6.5, p. 229
• c. Effects of Test Type
• context dependency effects vary with type of
  memory test given
• more likely to occur in free recall than in cued
  recall or recognition
• outshining hypothesis the use of context as a
  cue is a last resort that youll turn to only
  when better cues are unavailable
• the more direct the contact between the retrieval
  cue and the memory trace, the less likely it is
  that context will be needed as a cue

7
Grant, Bredahl, Clay, Ferrie, Groves, McDorman,
and Dark (1998)

• used the standard research design, manipulating
  the presence or absence of general background
  noise at both encoding and retrieval
• participants encoded a two page article on
  psychoimmunology, and were then tested with a
  fill-in-the-blank test, followed by multiple
  choice questions
• results
• --- see Table 6.6, p. 231
• no main effect of distraction at encoding,
  supporting many a students claim that studying
  with background music does not hurt memory
• however, there was a significant interaction
• memory performance was context-dependent whether
  or not participants were distracted at encoding,
  they were better off in the same environment at
  retrieval

8
The Testing Effect

• testing effect while studying information that
  you need to remember, it is useful periodically
  to attempt to retrieve the information you are
  trying to encode
• found in free recall, cued recall, and
  recognition
• Carrier and Pashler (1992, Experiment 2)
  -- in 3 phases
• first phase
• pairs of words were presented to participants for
  20 seconds
• each pair consisted of a stimulus term (i.e., the
  first word--a St. Lawrence Island/Yupik Eskimo
  language word) and a response term (i.e. the
  second word--the English equivalent of the
  stimulus term)
• second phase
• on the pure study (PS) trials, word pairs were
  re-presented for 10 secs
• on the pure-study trials, participants simply had
  to read it
• on the test trial/study trial (TTST) trials,
  participants were presented with the stimulus
  word for 5 seconds, to which was then added the
  response term for an additional 5 seconds
• on TTST trials, participants had to retrieve the
  response term from memory
• on both types of trials, participants were to
  give the response term associated with each
  stimulus word as quickly as possible

9
Testing Effect, cont.

• third phase (delayed test)
• All participants were given the stimulus word
  from each pair presented in phase 2 and asked to
  recall the corresponding response word
• the next day, the participants returned and were
  again given the same cued recall test
• Results --from phase 3
• more words were recalled from TTST trials in both
  an immediate test (6.4 vs. 5.7) and a delayed
  test (4.6 vs. 3.9)
• This was true despite the fact that in the pure
  study condition, the response items were
  physically presented for twice as long (10
  seconds) as in the TTST condition (5 seconds)
• Retrieving helps retrieving practicing retrieval
  improves it

10
Testing Effect, cont. How to improve learning

• Implications
• after some level of initial learning has been
  achieved, you should test yourself on the
  material by attempting to retrieve the
  information (e.g., flashcards, review questions)
• testing yourself proves to be a much more
  effective study strategy than simply reading a
  definition or a text section over and over and
  over.... (basically maintenance rehearsal, which
  is relatively ineffective method for storing
  information in LTM)
• testing effect demonstrates how to increase the
  benefit of maintenance rehearsal
• read terms and their definitions (or review
  questions and their answers) a few times to reach
  some level of initial competency
• follow these maintenance rehearsal trials by
  attempts at retrieving the definitions and
  answers using the terms and questions as cues

11
Encoding, Retrieval, and Brain Hemispheres

• HERA (hemispheric encoding/retrieval asymmetry)
  model
• based on various neuroimaging studies
• verb-generation task
• participants are presented with nouns, one at a
  time, and are required to generate an appropriate
  verb
• involves retrieval from semantic memory (e.g.,
  knowing the verb that goes along with joke is a
  fact that you just know)
• involves encoding into episodic memory (seeing
  words like joke and thinking of their associates
  is an event that youll remember later)
• results indicate greater activation in the left
  pre-frontal area, relative to the same area in
  the right hemisphere
• indicates the left hemisphere is instrumental in
  retrieval from semantic memory and encoding of
  episodic memories
• left-hemisphere activation is influenced by
  exactly how incoming information is processed
• processing items deeply at encoding leads to
  greater left hemisphere activation than
  processing them shallowly

12
neuro-imaging studies of episodic retrieval

• HERA model continued
• retrieving information from episodic memory is
  associated with more intense activation in the
  pre-frontal area of the right hemisphere than in
  the left
• studies have used a wide range of retrieval
  situations (e.g., free recall, cued recall,
  recognition) and a wide range of material to be
  remembered (words, sentences, objects, locations,
  odors)

13
IV. Implicit Memory

• Implicit memory does not require conscious
  recollection of a previous episode for successful
  performance
• memory is reflected implicitly, as an improvement
  or change in some task that occurs even if the
  participant remembers nothing about the original
  event
• IM may be responsible for unconscious plagiarism
• you come up with an idea and when youre
  describing it someone informs you that they had
  the same idea and discussed it with you three
  months ago
• your failure to recall that it was your friends
  idea is a failure of explicit memory yet you do
  remember the encoding episode implicitly, as
  reflected by your generation of the idea that had
  originated in that conversation
• IM may be responsible for deja vu experiences
• deja vu as example of encoding specificity
  principle in implicit memory
• deja vu occurs when we have the distinct
  impression that weve been in some place or had
  some experience before, when in reality we have
  not
• in keeping with the encoding specificity
  principle, some piece of a memory gets activated
  by a cue, but the entire memory is not retrieved
  one is left with a feeling of familiarity that
  cannot be readily explained

14
Implicit Memory Study Design

• general experimental paradigm for implicit memory
  studies
• participants study some type of information, most
  commonly a word list
• later, memory is tested but there is no mention
  of these previously encoded words but presenting
  the word earlier primes the person to come out
  with it later
• implicit memory tests
• word-fragment completion
• youre faced with a word where some letters are
  there and some arent and your task is to try and
  come up with the appropriate word (Wh_ _l)
• word-stem completion
• the initial letters of a word are presented, and
  your task is to complete the stem with the first
  word that comes to mind (Frag_ _ _ _)
• implicit memory is measured in terms of priming,
  or the benefit in performance from having
  previously seen a word
• you encoded list of 5 words and are tested a week
  later with 10 word fragments that included those
  5 words, plus 5 other words
• you successfully complete 4 of the 5 fragments
  from words you saw earlier (80)
• for the other 5 word fragments, you only complete
  one, or 20 Priming is 60. there is a 60
  benefit in performance from having seen the words
  earlier

15
Warrington and Weiskrantz (1970)

• compared memory functioning in amnesics and
  non-amnesics (people without amnesia)
• a list of words was presented
• recognition, an explicit memory test
• amnesics recognized far fewer words than the
  non-amnesic participants
• implicit test (word stems or word fragments)
• amnesics demonstrated priming, just like
  non-amnesic participants
• amnesics were better able to complete stems and
  fragments that corresponded to words they had
  seen earlier even though they could not
  consciously recollect many of those same words

16
Retrieval Dissociations

• a dissociation occurs when some variable
  influences performance in different ways,
  depending on how performance is measured
• the ability to remember explicitly is impaired in
  amnesics, while remembering implicit is
  unaffected
• but you dont have to be amnesic to reveal
  dissociations
• Dissociations also found with levels of
  processing, visual encoding, organization and
  retention interval -- How are these to be
  explained?
• Smith and Branscombe (1988) -- for a
  phase 3, see later!
• in Phase 1 of the experiment, participants
  encoded trait words
• some of these words were simply read, while
  others were generated from clues and an initial
  letter
• in Phase 2, participants were tested in one of
  several ways
• some participants had a free recall test, in
  which they tried to remember as many of the trait
  words as possible
• other participants were tested implicitly with
  word fragment completion
• Results - found a dissociation between implicit
  and explicit memory
• when tested explicitly, generated words were more
  likely to be remembered than read words (the
  generation effect)
• when tested implicitly, read words were more
  likely to be remembered than generated words

17
Accounts of Explicit-Implicit Dissociations

• a. Memory Systems Account
• LTM is not a unitary entity it is made of
  several sub-systems
• major distinction is between conscious forms of
  memory (like retrieving memory for facts and
  events - declarative memory) and nonconscious
  forms of memory (like priming, and the learning
  of skills and habits -procedural memory)
• these two types of LTM are mediated by different
  brain systems with different neurological
  underpinnings
• because performance on explicit and implicit
  tests is based in different systems, they are
  affected by different variables
• the dissociation between conscious and
  nonconscious forms of memory in amnesics suggests
  that the brain systems underlying these types of
  memory have been differentially affected by the
  associated brain damage
• the structures associated with declarative memory
  have been damaged, but the structures underlying
  procedural memory have been spared

18
Memory Systems Account or not?

• declarative memory -- knowing that...
  something is so
• responsible for retention of factual information
  about the world and ones personal past serves
  as the basis for performance of explicit memory
  tasks like recalling and recognizing that some
  event occurred earlier
• episodic and semantic memory are subsystems of
  declarative memory
• procedural memory -- knowing how... to do
  something
• examples include skills (tying your shoe, typing,
  swinging a golf club) the priming involved in
  word fragment completion, and the formation of
  simple associations (like a classically
  conditioned taste aversion)
• the fact that similar effects can be revealed on
  explicit and implicit tests indicates that the
  same memory system may underlie both
• However, comparison of non-amnesics and amnesics
  on implicit tests of memory does not always yield
  equivalent performance
• Ostergaard (1999) argues that priming effects are
  based on episodic memories for previous exposures
  to the item, not on distinct system
• episodic memories involve conscious remembering,
  complete with contextual elements the same
  source that forms the basis for explicit
  remembering
• this conscious remembering might be used in an
  implicit memory task

19
Ostergaard (1999) Against Multiple Systems

• participants are instructed to name presented
  words as quickly as possible in order to do
  this, they will use whatever info is available
  (prior experience, info from the stimulus itself,
  or prior exposure to the word during the
  experiment)
• priming is seen when naming a word becomes faster
  after previous exposure
• words were presented in one of two conditions
• condition 1--the words gradually appeared over a
  period of 5 seconds
• upon the first presentation, the only information
  available to name the word will be information
  inherent in the stimulus (i.e., bottom-up
  processing) naming speed will be slow because of
  the slow accumulation of data
• upon a second presentation, non-amnesics will be
  able to add top-down processing, relying on
  previous exposure to the word (i.e., an episodic
  memory for a previous presentation) in order to
  help identify it
• Prediction the improvement in naming speed from
  presentation 1 to presentation 2 (i.e., priming)
  will be large for nonamnesics
• amnesics will not be able to fully use the
  information from top-down processing because of
  their impaired ability (not complete inability)
  to retrieve prior episodes therefore, priming
  will be small
• the comparison of amnesics to non-amnesics
  revealed the same pattern found on explicit
  memory tests--non-amnesics better than amnesics

20
Ostergaard (1999) cont.

• condition 2--intact words were presented
  instantly
• at the first presentation, naming speed will be
  fast because information from the stimulus is
  completely available (i.e., bottom-up processing)
• upon the second presentation, top-down processing
  from having previously seen the word will be less
  important because of the ready availability of
  the data for bottom-up processing
• Prediction non-amnesics will still get some
  small benefit from top-down processing but given
  that initial naming speed was so quick, the
  priming effect will be small
• given the reduced role of top-down processing in
  this condition, amnesics did not show as much of
  a deficit differences in priming were small or
  nil
• this is the traditional pattern found on implicit
  memory tests
• indeed, much of the research that demonstrates
  this pattern utilizes retrieval tasks that are
  more akin to condition 2
• in this model both types of tests are based in
  the declarative system, more specifically, on
  memory for previous episodes
• dissociations between implicit and explicit
  performance arise simply because of how the
  retrieval task allows (or doesnt allow) the use
  of these past episodes
• So perhaps there are not multiple systems

21
Transfer-Appropriate Processing Account

• Roediger (1990)
• whether a test is implicit or explicit is not
  critical the critical factor is whether there is
  a match or mismatch between encoding and
  retrieval processes
• dissociations between implicit and explicit
  memory tests occur because these tests typically
  depend on different sorts of processing
• implicit retrieval tests are data-driven, they
  rely on reading and perceptual operations for
  successful performance
• explicit memory tests are conceptually-driven,
  they rely on elaboration and organization for
  successful performance
• performance on a given test will depend on how
  the material was processed at encoding
  successful performance will result if the
  encoding processes successfully transfer to
  retrieval (i.e., match)
• implicit retrieval tends to be data-driven, it
  will be aided by data-driven encoding processes
• explicit retrieval tends to be conceptually-driven
  , so it will be aided by conceptually-driven
  encoding processes

22
Smith and Branscombe (1988)

• according to the transferappropriate-processing
  account, generation is a conceptually-driven
  encoding process that will transfer better to a
  conceptually-driven retrieval task like free
  recall
• explains why generated items were recalled better
  than read items encoding and retrieval processes
  matched
• reading is a perceptual (i.e., data-driven)
  process that transfers to a data-driven retrieval
  task like word fragment completion
• consistent with this prediction, priming in
  word-fragment completion was higher for words
  that had been read, relative to words that had
  been generated
• this view would not necessarily predict that
  generating would only help explicit memory
  generating (which is a conceptually-driven task)
  would help any conceptually-driven task, explicit
  or implicit

23
Smith and Branscombe (1988) cont

• phase 3
• category accessibility task looked at
  descriptions of ambiguous behaviors and provided
  a one-word description of the person
• task is implicit, no reference is being made to
  the trait words earlier encoded
• task is conceptually-driven, relying on meaning
  and association
• this condition pits the memory systems and
  transfer-appropriate-procedures view against one
  another
• memory systems approach would predict that the
  pattern of results from category accessibility
  should be similar to word fragment completion
  (read items remembered better), because both are
  implicit tasks
• transfer-appropriate-processing approach predicts
  that the results from category accessibility
  should look like free recall (generated items
  remembered better), because both are
  conceptually-driven
• Results --see
  Table 6.10 p. 241
• performance in the category accessibility test
  paralleled performance in free recall
• priming was higher for generated items relative
  to read items
• This is consistent with transfer-processing
  account of dissociations

24
Conscious and Nonconscious Access to the Past

• remember-know paradigm asking about ones sense
  of awareness during retrieval
• participants are asked to recognize events
  (typically words) that occurred in an earlier
  list
• for items that are recognized, participants are
  asked if they remember seeing the word, or do
  they just know that they saw it earlier
• a remember judgment means that participants can
  vividly recall the presentation of the word,
  basically reliving the experience--conscious and
  effortful memory retrieval
• a know judgment means that there is no experience
  of recollection, or reliving the person just
  knows the word appeared earlier--nonconscious
  automatic memory retrieval
• research employing this task has demonstrated a
  number of dissociations
• Rajaram (1993) --see
  Table 6.11 p. 244
• presented words (e.g., cat) to participants and
  had them generate either a rhyme associate (bat)
  or a semantic associate (dog)
• memory was tested with recognition if
  recognized, a remember-know judgment was made
  The Results
• words with semantic associate were remembered
  better (levels-of-processing effect) but this
  effect was limited to remember judgments