Memory codes

1
Memory codes

• Memory codes refer to the representation used to
  store an item in memory
• Craik and Lockhart (1972)
• Todays lecture will focus on whether memory
  should be thought of as consisting of a single
  system or multiple systems

2
Memory codes

• History
• In the late 1960s people believed there was a
  single type of memory system as will be
  described, this view came to be challenged
• The Atkinson-Shiffrin model distinguished between
  different types of memory STM/LTM

3
Memory codes

• History
• Subsequent work by Baddeley and his model of
  working memory has greatly changed our
  understanding of working memory
• Many neuroscientists currently believe there is
  some form of short-term storage buffer, and some
  form of working memory

4
Memory codes

• History
• Currently at least within neuroscientists there
  is a general view that there are multiple systems
  of memory
• It now seems to be generally agreed that
  explicit/implicit memories are two of the
  important systems (other terms for these systems
  are declarative/nondeclarative)

5
Memory codes

• History
• Declarative memory refers to a form of memory
  that is unlimited in storage capacity
  multimodal information can be stored quickly,
  often after a single exposure it can be
  described (e.g., verbally) information is
  accessible flexibly inferences can be made from
  information products of retrieval can be
  consciously thought about
• Tulving Markowtsch (1998) Hippocampus

6
Memory codes

• History
• Nondeclarative or implicit memory will be
  discussed in a subsequent lecture

7
Implicit Memory
Explicit Memory
Priming
Nonassociative Learning
Neocortex
Reflex Pathways
Facts (Semantic Memory)
Events (Episodic Memory)
Skills and Habits
Simple Classical Conditioning
Striatum
Medial Temporal Lobe
Emotional Response
Skeletal Musculature
Amygdala
Cerebellum
Squire and Knowlton (1994) Squire (1987)
Declarative vs. procedural memory
8
Memory codes

• Explicit memory (Declarative memory)
• Episodic memory and semantic memory are both
  types of declarative memory
• Standard view (Squire, 1998, Hippocampus)
• both types of memory are dependent on medial
  temporal structures (hippocampal formation and
  adjacent anatomically-related cortex) and medial
  diencephalic structures
• Episodic memory depends, in addition on frontal
  lobes

9
Hippocampus
10
Memory codes

• Explicit memory (Declarative memory)
• Implication of this view is that when episodic
  memory is impaired semantic memory should also be
  impaired

11
Memory codes

• Explicit memory (Declarative memory)
• Support for standard view
• Hamman Squire (1995) tested this implication by
  teaching new factual knowledge to amnesics (40
  three-word sentences such as MEDICINE cured
  HEADACHE). Training occurred during four weekly
  sessions (two training trials per session). At
  test participants presented fragment (MEDICINE
  cured _______) and participant instructed to
  complete with word studied

12
Memory codes

• Explicit memory (Declarative memory)
• Hamman Squire (1995) Results
• Patients learned slowly- from 0 to 19 after 4th
  training session controls were at about 75 1
  week after first training session
• Event memory was tested in second session by
  asking about events that occurred in first
  training session
• Patients were impaired on event memory and on
  fact learning to a similar degree

13
Memory codes

• Explicit memory (Declarative memory)
• Hamman Squire (1995) Conclusion
• This finding supports the standard view, because,
  as hypothesized episodic memory and semantic
  memory were acquired to a similar degree

14
Memory codes

• Explicit memory (Declarative memory)
• - Opposing the Standard view
• Contrary findings were reported by Tulving et al.
  (1991) this study showed that amnesic patient KC
  acquired semantic knowledge rapidly (it was still
  impaired) even though his episodic memory was
  severely impaired in other words there was a
  dissociation between his episodic and semantic
  memory performance

15
Memory codes

• Explicit memory (Declarative memory)
• - Opposing the Standard view
• Conclusion
• Tulving (1991) proposed that semantic memory can
  be spared or partially spared in amnesic patients
  even though episodic memory is impaired

16
Memory codes

• Explicit memory (Declarative memory)
• Counter-argument from the standard view
• Squire (1998 Hippocampus) suggested KC had some
  functional medial temporal lobe tissue and this
  allowed him to acquire semantic knowledge his
  impairment on episodic memory was a product of
  medial temporal lobe and frontal damage
• therefore no reason to reject standard view

17
Memory codes

• Explicit memory (Declarative memory)
• Further support for standard view
• Remembering and knowing in amnesia
• It is thought that remembering and knowing
  reflect the operation of episodic and semantic
  memory respectively (Tulving, 1989)
• Remember (R) refers to recollection that is
  embedded in some aspect of the episode Know (K)
  refers to a sense of familiarity it is context
  free

18
Memory codes

• Explicit memory (Declarative memory)
• Further support for standard view
• If amnesia affects both semantic and episodic
  memory then both R and K should be affected
• If semantic memory is relatively unaffected then
  K should be less affected than R (Tulving, 1991)

19
Memory codes

• Explicit memory (Declarative memory)
• 13 amnesic px were shown 36 words and then 10
  minutes later were given yes/no recognition tests
• For each word endorsed as a study item
  participants had to make an R or K judgment
  (Knowlton Squire, 1995)
• Amnesic px were impaired in both R and K
  judgments, and they performed like control
  subjects who were tested after a 1-week delay
• Knowlton and Squire concluded there was no
  indication in data to support sparing of semantic
  knowledge in amnesics

20
Memory codes

• Explicit memory (Declarative memory)
• Argument against standard viewdevelopmental
  amnesia
• Vargha-Khadem (1997) reported 3 young
  neurological patients who had sustained
  neurological focal bilateral hippocampal damage
  early in life, but spared perirhinal, entorhinal,
  and paraphippocampal regions

21
Memory codes

• Explicit memory (Declarative memory)
• Px have difficulty remembering everyday events
  and retaining information studied under
  controlled conditions
• Px have normal intelligence or near normal
  intelligence
• Px have made normal or near normal progress
  through school
• Px have considerable capacity for semantic memory
  (speech and language, literacy, and factual
  knowledge) in spite of impaired episodic memory

22
Memory codes

• Explicit memory (Declarative memory)
• Authors proposed that semantic memories were
  stored independently of episodic memories, by way
  of the intact perirhinal and entorhinal cortices.
  This is contrary to standard view.
• In response, standard view (Squire, 1998)
  proposed that each patient with developmental
  amnesia had some residual episodic memory that
  supports the acquisition of semantic knowledge,
  although only after time and repetition

23
Memory codes

• Patients described by Vargha-Khadem, unlike most
  other amnesic patients appear to have
• Impaired recall (similar to amnesics)
• But, relatively spared recognition (unlike
  amnesics)
• Why?
• hypothesis spared recognition of V-K patients
  mediated largely by semantic memory

24
Memory codes

• Patients described by Vargha-Khadem, unlike most
  other amnesic patients appear to have
• Impaired recall (similar to amnesics)
• But, relatively spared recognition (unlike
  amnesics)
• (See Baddeley et al. 2001)
• Why?
• hypothesis spared recognition of V-K patients
  mediated largely by semantic memory

25
Memory codes

• Gardiner et al. (2006)
• Case study of Jon, an individual with
  developmental amnesia, but having normal IQ,
  normal speech and language, and good semantic
  knowledge
• Is Jons recognition memory mediated by semantic
  rather than episodic memory
• Strategy selected independent variables that
  normally increase remember but not know
  responses levels of processing, and task
  enactment

26
Memory codes

• Gardiner et al. (2006)
• levels of processing experiment
• Study. Presented two lists of 30 words for 15
  words participants made pleasantness ratings
  (deep semantic processing) and for other 15 words
  counted number of syllables (shallow processing)
• Test. Presented 45 words, 30 study words and 15
  unstudied words. For each word participants had
  to make an old/new judgment and then for old
  judgments make remember, know, and guess
  responses

27
Corrected recognition (hits minus false alarms)
28
Memory codes

• levels of processing experiment
• Remember, know, guess data showed the following
• Jon and controls were more like to judge old
  items remembered in the pleasantness condition
  than the syllable condition

29
Memory codes

• levels of processing experiment
• Conclusion
• Jons recognition enhanced by deeper levels of
  processing
• Jon does seem to be impaired in his recognition
  performance relative to controls at least when
  deeper levels of processing are required
• At face value, these data suggest that Jon is
  good at remembering and that his recognition
  performance reflects episodic memory (though see
  Duzel, E. 2001 Greves, 2006)

30
Memory codes

• Task enactment experiment
• Study. 48 noun-verb actions phrases were
  presented (e.g., read a book) for 24 phrases
  participants read the phrase, and for the other
  24 participants mimed the phrase (enacted the
  phrase without the object)
• Test. The 48 studied phrases were presented
  together with 48 matched phrases that included
  the same phrase (e.g., read a newspaper) that
  were not presented at study. Participants made
  the same old/new and remember, know, and
  guess judgments required in previous experiment

31
Corrected recognition (hits false alarms)
32
Memory codes

• Task enactment experiment
• Conclusions.
• Jons performance was not influenced by task
  enactment unlike controls. Previous research has
  shown that enactment performance is usually
  higher than read performance and this effect is
  attributed to better remembering (episodic
  memory) see Kouri-Nouri, R. 2005

33
Memory codes

• Task enactment experiment
• In a follow-up experiment Jon performed an
  experiment in which bizarre (e.g., hug a cactus)
  and ordinary phrases were read or were mimed at
  study.
• At test participants were given the studied nouns
  together with 40 unstudied nouns and were
  required to make the same judgments as in the
  previous experiments.

34
Memory codes

• Task enactment experiment
• Previous research has attributed higher
  recognition performance to deeper conceptual
  processing of the bizarre phrases. Hence if Jon
  processes material semantically, one would
  predict higher performance for bizarre than
  ordinary phrases, but no difference for enact
  versus read. This finding is what was observed.

35
Memory codes

• Enactment study (contd)
• In addition, participants had to recall the verb
  with which the noun had been presented.
• Jons recall performance was extremely low
  compared to controls

36
Memory codes

• Enactment study (contd)
• Conclusions.
• enactment studies do not support the notion that
  episodic memory contributes to recognition
• General discussion
• - in general findings support the idea that
  Jons recognition memory is semantically, not
  episodically mediated.
• - deeper levels of processing in the levels of
  processing experiment may be a consequence of
  semantic processing without transfer to episodic
  memory

37
Memory codes

• Enactment study (contd)
• Conclusions.
• In general, Jons memory appears to be somewhat
  impaired, especially following tasks requiring
  deeper levels of processing or more integrative
  processing. However, his recognition performance
  is relatively well preserved in comparison to his
  recall performance.

38
Memory codes

• Challenging the Standard Model
• Serial Parallel Independent Model (SPI)
• This model hypothesizes that encoding is serial
  (S), storage is parallel (P), and retrieval is
  independent (I)
• It is assumed that there are three independent
  memory systems perceptual, semantic, and
  episodic. These systems are hierarchically
  organized and information enters into these
  systems seriallyinfo enters into perceptual
  system, then into semantic system, and finally
  into the episodic system

39
Memory codes

• Challenging the Standard Model
• Serial Parallel Independent Model (SPI)
• Parallel. Information is stored in parallel. This
  means that different aspects of the input are
  stored in different systems
• Independent. Information is retrieved
  independently. This means that what is retrieved
  from one system does not have implication for
  information retrieved from the other systems.

40
Memory codes

• Challenging the Standard Model
• Serial Parallel Independent Model (SPI)
• Note. Unlike the standard model this model allows
  for the possibility that information can enter
  into semantic memory independently of it entering
  into episodic memory
• This provides a straightforward interpretation
  for the findings from px with developmental
  amnesia because it means that acquisition of
  semantic information can proceed in the absence
  of episodic memory (see also results from Jon in
  Gardiner, 2001)

41
Memory codes

• Challenging SPI
• Semantic dementia
• - temporal variant of frontotemporal dementia
• - neuropathological and neuroradiolgical studies
  have reported progessive focal atrophy of the
  inferolateral apsect of the left and/or right
  temporal lobes with relative sparing of
  hippocampal structures
• semantic dementia results in impaired
  performance on any task requiring conceptual
  knowledge about objects, facts, concepts, and the
  meanings of words.

42
Memory codes

• Challenging SPI
• Semantic dementia studies
• Previous research has shown that px with semantic
  dementia may have spared episodic memory for
  recently experienced events suggesting that
  mechanisms for encoding new episodic memories may
  function adequately
• This research would run counter to the SPI model
  because that that model hypothesizes that new
  episodic memory is dependent upon semantic
  knowledge of the items or concepts to be encoded

43
Memory codes

• Challenging SPI
• Semantic dementia studies
• Previous research has shown that px with semantic
  dementia may have spared episodic memory for
  recently experienced events suggesting that
  mechanisms for encoding new episodic memories may
  function adequately
• This research would run counter to the SPI model
  because that that model hypothesizes that new
  episodic memory is dependent upon semantic
  knowledge of the items or concepts to be encoded

44
Memory codes

• Challenging SPI
• Semantic dementia studies
• Graham et al. (2000) performed a study in 8 px
  with semantic dementia. Px showed preserved
  forced-choice recognition memory for pictures of
  objects in a memory experiment although they had
  impaired knowledge about these items.

45
Memory codes

• Challenging SPI
• Based on these findings it has been proposed
  (See Simons et al. 2001) the serial notion of the
  SPI model is inconsistent with this finding. They
  argue that according to SPI theory impaired
  semantic memory implies that episodic memory
  should also be impaired.

46
Memory codes

• Multiple input model to episodic memory
• Instead they propose a multiple input model.
  According to this model, both sensory-perceptual
  information and semantic information can be input
  into episodic memory. This means that episodic
  memory can still be relatively spared even when
  semantic memory is impaired because it can draw
  upon information from the sensory-perceptual
  system