Episodic memory Flashcards
Hippocampal vs PFC-controlled encoding
Hippocampus is important in forming new memories. It can do this automatically - even when we actively try not to remember, e.g. in PTSD
When we use strategies to remember, we involve the PFC. Mainly we talk about ventrolateral and dorsolateral PFC.
Strategies used
Practice, elaboration, deeper processing, active retrieval, spacing, encoding-retrieval overlap, organisation.
Practice
Rehearsal retains information in STM and allows transition to LTM (so not so much a strategy as a necessity?)
This idea is core to the Atkinson and Shiffrin 1968 multi-store model and the Baddeley and Hitch 1975 phonological loop
Pirolli and Anderson 1985 found degree of rehearsal correlated with amount of info recalled.
Hanson et al 2000 - rehearsal is associated with activity in ventrolateral PFC
BUT - introducing a filled delay (with arithmetic or a digit span task) after every item in a list of words to remember does not erase the primacy (or recency) effect, i.e. things are still being encoded into LTM. Perhaps this is because these tasks did not require speaking out loud?
Level of processing
Craik and Lockhart 1972 - got people to learn things by simple rehearsal, or by thinking about the words’ meanings and connections to other knowledge. Recall was better in the latter stage.
Wagner et al 1998 - imaged people while studying (half using non-semantic learning, half using semantic studying) then tested success at recall. Those using semantic studying were more successful, and had increased activity in left VLPFC and MTL. They showed that previous studies’ observations of dorsolateral PFC activation were reflecting retrieval attempt, rather than success.
The assumption is that processing deeper makes the memory trace more unique and thus less subject to interference.
Elaboration
Anderson and Bower 1972 - C+L’s results were due to elaboration of info to previous knowledge. When trying to remember the pairing ‘doctor’ with ‘lawyer’, subjects were better when they’d generated an elaboration themselves than when simply given the sentence ‘the doctor hated the lawyer’. This paper also gave their ‘generate-recognise’ model
Frase and Schwartz 1975 - giving students relevant questions to think about during studying (this group were the ‘advance organisers’) improved recall.
Gabrieli et al 1996 - more activity in left VLPFC when subjects used elaborative techniques during encoding
Organisation
Segal and Mandler 1972 - gave some subjects words to remember already sorted into categories, the other subject had to categorise them themselves. Recall was best for the highest organisational demands.
Blumfeld and Ranganath 2006 - Activity in highest level of organisation was seen in the left dorsolateral PFC, which is itself associated with manipulating information as opposed to just maintaining
Encoding-retrieval overlap/Encoding specificity principle
Tulving and Thompson 1973 - Our memory trace includes information about the context in which we learnt the info (this is the basis of episodic memory). This context includes internal feelings and state, which is affected by temperature, light, etc etc. Our recall success is dependent on the similarity between the info in the memory trace and info available at retrieval.
Goodwin et al 1969 - State-dependent memory is where we have more recall success when the retrieval context is the same as the encoding context (e.g. drunk vs sober, wet vs dry, quiet vs noisy). Imaging studies have shown that this overlap also corresponds to overlap in brain areas active during encoding vs retrieval.
Spacing
Bahrick 1979 - Taught people English-Spanish word pairs in massed practice sessions, or in sessions separated by 1 or 30 days. The 30 days separated subjects were more successful in the long term, although massed practice was more successful short-term.
Active retrieval
Tulving 1964 - Subjects learnt a word list either via (study, test, study, test) or via (study, study, study, test) or via (study, test, test, test). He found equivalent success in all three conditions. In fact, repeated testing is better than repeated studying if you assess recall after a 1-week delay. Nelson et al 2013 - Activity during active retrieval in the left ventrolateral PFC and other areas.
Retrieval is actually reconsolidation
Loftus and Palmer - people watched a car crash, and wee asked using ‘smashed’, ‘hit’ or ‘contacted’. Speed estimate was higher for ‘smashed’, as was damage they ‘remembered’ observing. Thus memories were altered during retrieval by the retrieval cue.
English students were taught a Native AMerican folk tale and asked to repeat it. Stories got shorter, gist remained, details replaced by bits from English folk tales. So again, memory altered by retrieval, interacted with existing memories.
Generate-recognise model
Anderson and Bower 1972 - A cue is given, the brain generates candidate memories, these are tested to see if the brain recognises them as things seen before/the desired memory, then either retrieved (if yes) or generate new candidates if no.
This is why cued recall is better than free recall (cue helps generate candidate memories), and why recognition memory is the best (skip candidate generation altogether).
Also accounts for frequency effect on recall and recognition; words encountered less frequently are harder to recall (because they’re less likely to be produced as candidates), but easier to recognise (because they’re more easily distinguishable).
BUT, Tulving and Thompson’s experiment using four phases (weak associates to remember, generate more associates, were those generated words in the first list? what was associated with X in the first list?) showed that recall can be good (phase 4) but recognition bad (phase 3). Instead, proposed the encoding specificity principle.
Recollection and Familiarity
Recognition can be via familiarity (implicit, perceptual, extra-hippocampal?) or recollection (episodic, hippocampal?).
Korsakoff’s amnesics could recognise pictures seen either that day or the day before, but couldn’t say which day they’d seen them (controls could).
Using a concurrent task during learning reduces Remember but not Know responses, and giving masked presentations of the target before the target itself increases Know but not Remember responses.
Hupert and Piercy gave amnesics three presentations then test, or one then test. Much better at the former, presumably because it increased familiarity?
Subjects with selective hippocampal damage are impaired only at Remember responses. Subjects with widespread MTL damage are impaired at both Remember and Know responses.
Imaging has suggested that hippocampus is involved in recollection (more activity in correct Remember vs correct Know responses), perirhinal cortex in familiarity
–BUT note than the Manns/Squire group has found impairments in both Remember and Know in selective hippocampal lesions.–
Source memory - experimental evidence for (5)
Janowsky et al 1989 - patients with frontal lobe damage could recollect facts, but not where they’d learnt them
Jackson et al 1993 - drew a distinction between internally generated memories (e.g. imagination) and externally generated memories (facts learnt).
Jackson et al 1994 - forcing people to respond in 300msec impaired source memory (‘where did you hear this’) but not recognition itself. But does this mean they’re stored separately, or just that retrieving the extra info takes longer?
Glisky et al 1995 - older adults with high frontal function were better at source memory (which voice had read out the sentence). Those with high MTL function were no different to those with low MTL function.
SImons et al 2002 - a little hippocampal damage permitted source memory, but a little frontal damage totally abolished it
Source memory - imaging evidence for (5)
Ranganathan et al 2000 - more activity in left anterior frontal cortex for source memory than recollection.
Cansino et al 2002 - right anterior hippocampus active when recalling spatial position of item
Davachi et al 2003 - hippocampus and posterior parahippocampal gyrus activity correlated with source recollection but not item recognition. Perirhinal cortex activity correlated with item recognition but not source recollection. Image encoding engaged anterior part of left VLPFC
Simons et al 2006 - asked patients to distinguish between an internally generated memory and externally generated. Activity seen in same regions as are dysregulated in schizophrenia, namely anterior medial PFC (valence of reward, reward value)
VLPFC involvement in successful encoding/retrieval (8)
Frontal patients are most impaired when presented with disorganised stimuli, or stimuli similar to each other (remember, hard to test frontal patients bc personality changes)
Gabrieli et al 1996 - elaborative encoding caused more left VLPFC activity
Wagner et al 1998 - ‘subsequent memory methods’ showed that activity in VLPFC and MTL predicted successful recollection (and successful recollection was ‘created’ by using semantic studying)
Kelley et al 1998 - left VLPFC most active for words, bilateral activation for nameable objects, right VLPFC for unfamiliar faces
Poldrack et al 1999 - anterior VLPFC for semantic elaboration, posterior VLPFC for phonological elaboration
Henson et al 2000 - rehearsal caused activity in VLPFC
Davachi et al 2003 - activity in left VLPFC seen when subjects could recollect source
Blumfeld and Ranganath 2006 - DLPFC active in organising material to be remembered
Nelson et al 2013 - repeated active retrieval is associated with VLPFC activity
