Complementary learning systems Flashcards
(17 cards)
3 parts of multistore model of memory
- Sensory information input.
- Short-term store (conscious thought, active rehearsal).
- Long-term store (knowledge, passive).
Luck et al. (1997) dual task to demonstrate that there are 2 types of stores in WM
People show a sharp drop-off when set size >4 (visual), but this does not depend on verbal load.
Logie et al. (1990) dual task paradigm with primary task and secondary tasks visual or phonological
Interference only occurred if the two tasks used the same mode of information. Separate stores in WM can operate separately and in parallel.
Deficits with left hemisphere lesions
Verbal WM deficits.
Deficits with parietal cortex lesions
Spatial WM deficits (neglect).
Smith et al. (1996) healthy participants shown letters on the screen tested on either the letter or the position
Even for the same stimulus, different parts of the brain were activated depending on the task.
Owen et al. (1999) central executive maintenance (rehearse span) task vs. manipulation (n-back) task brain activation
DLPFC activity increase relative to baseline only in manipulation task (site of organising contents of WM). VLPFC active in both tasks relative to control - site of maintenance (where the memory is stored).
Tulving (2002) patient KC amnesia EM and SM
- Anterograde amnesia: both EM and SM affected.
- Retrograde amnesia: loss of EM (specific encounters with people etc.) but spared SM (still had general knowledge and language).
Why EM vs. SM dissociation in KC is not simply due to difference in memory strength (SM stronger memory)
- KC retained the distinction of stalactites and stalagmites (general knowledge) but could not recollect the event of a toxic spill from a train crash near his house.
- KC was able to learn semantic knowledge (eg. amusing definitions: ‘performs daily massage..?’) but no memory of visits to the lab.
Semantic dementia site of damage
Anterior parts of the temporal lobe.
Warrington (1975) patient SM 4 impairments and 3 spared abilities.
- Picture naming.
- Naming from description.
- Picture/word matching.
- Picture/word properties.
BUT intact WM and problem solving. Also spared autobiographical info (does not remember nuclear even but remembers that a cloud came over and plants died).
2 reasons against episodic memory = only MTL
- Amnesia gradient (patient PZ’s autobiography). More memories retained for remote than recent memories (not all EM in HC).
- Semantic dementia. Reversed gradient (more semantic knowledge for recent than remote events).
Maguire et al. (2000) taxi drivers and HC
Taxi drivers had enlarged HC relative to age-matched controls. Storage of spatial general knowledge making HC larger.
2 examples of HC not for EM
- Spatial navigation (taxi drivers).
- HC in WM for novel info (Ranganath and Blumenfeld, 2005).
O’Reilly and Norman (2002) plastic memory system model
Need both specifics and generalities (stability-plasticity dilemma).
HC: separation of info.
NC: integration of info.
McClelland et al. (1995) complementary learning systems
Not based on EM/SM, but rather fast/slow learning. HC memory gateway and info slowly gets transferred and consolidated in NC.
HC: fast learning. Separate memories with arbitrary links. Remember specifics.
NC: slow learning. Integrated memories with consistent links. Extract stable generalities.
Vargha-Khadem et al. (1997) evidence against CLS
Effects of early HC damage. Patients had childhood amnesia:
1. Impaired EM: couldn’t remember activities or stories.
2. Spared SM: verbal IQ, reading, and digit span were ok.