memory systems Flashcards
O’Keefe and Nadel (1976)
- place cells in HC fire when animal moves over place field
- positional and contextual information
- form basis of cognitive map
- allocentric (whereas egocentric coded for by cells outside HC e.g. grid)
population activity of HC place cells encode whole environments - formed basis of argument that HC was selectively specialised for processing of spatial info, as HC was required for spatial tasks but not non-spatial
- although, some of these studies also differed in not just spatial/nonspatial but also relational/flexible memory vs rigid/response-only memory
- research now shows that HC is also required for nonspatial aspects of episodic memory (e.g. Eichenbaum 2000)
Wilson 1993
- recordings from 80 CA1 neurons
- well-defined place fields
Eichenbaum 2000
- HC not just for spatial navigation!
- damage to HC impairs both spatial and nonspatial info
- T maze spatial alternation task: most HC cells that fired to location on arm of T (common area) only fired if subsequent turn was in specific direction i.e. either left or right
- so also encodes non-spatial aspects of events e.g. intended direction of movement
separate HC networks encode sequences of behaviours and places separately for left and right turns - episode-specific encoding of aspects including spatial location = “memory space”
- consistent with neuropsychological findings that show HC is required for factual info in memory episodes
Kesner (2005)
- rats learned to associate objects with reward, separated by temporal gap
- CA1 lesion impaired associations between objects over time (CA3 lesion had no effect)
Eichenbaum (2017)
- plasticity mechanisms of highly interconnected HC circuits integrate events that neighbour in space and time
- creates continuous mapping of adjacent elements that have proximity in space (place cells), time (time cells) or both at same time
- HC network creates maps of arbitrary spaces by adding neurons that code for specific events
- elements are linked when they occur together, and linkages extended between elements that have similar attributes (allows for relational memory)
- these direct and indirect associations form basis of complex memory space where memories are linked in space, time, context etc
- space and time initially processed by overlapping brain networks and coded in different scales but then signals are integrated in HC to create spatiotemporal organisation of memory
Moscovitch (2017)
component process model
- during perception, MTL integrates objects and contexts, and objects bound together in HC via spatiotemporal context
- during encoding, part of representation is transferred to long-lasting format in HC and neocortex (supported by schematic relational processes in vmPFC and semantic processes in vlPFC)
- during retrieval, integrated event representations in HC are activated which reactivate MTL and cortex representations etc (explains how similar / adjacent memories can reach consciousness)
= creates multidimensional memory trace / engrams
Eichenbaum (2014)
- HC time cells that fire at successive moments in temporally structured experiences
- not caused by external events, specific behaviours or spatial dimensions of an experience
- instead represent flow of time within a specific memory
- provide additional dimension that is integrated with spatial mapping etc, helps organise elements of an event into coherent memories i.e. combine time and space
- so episodic memory involves embedding a record of events in a representation of spatiotemporal context
Libby (2014)
- fMRI showed HC activity patterns predicted accurate memory for specific object-location relationships
- demonstrates HC role in spatial memory
Lehn (2009)
- subjects recalled order of movie scenes
- strong fMRI activation in HC related to retrieval of temporal order (and predicted accuracy)
Kyle (2015)
- virtual reality game where subjects visited stores in specific spatial layout and in particular temporal order
- made near or far judgements either related to spatial layout / distance, or how close they were in temporal order
- comparable levels of activity throughout HC so space and time both processed throughout / not localised to particular regions of HC
- but space and time judgements were characterised by distinct patterns of neural activity, so suggests they are processed via different neural networks within HC
Packard + McGaugh (1996)
- rats trained on T maze = place strategy when tested after 1 week
- overtrained = response strategy
- so initially, place memory and cognitive map guided acquisition of memory (learning task) then switched to response memory as habit developed
- lidocaine into HC abolished place memory
- lidocaine into striatum abolished “response” memory in week 2 (but still could do task as could use place memory)
cued radial arm (olton) and morris water mazes
- use response strategy if they are cued e.g. with a light (associative learning only)
- not impaired by HC lesion
Cook + Kesner (1988)
- rats with striatum lesion = impaired on response tasks e.g. visual discriminant water maze, or turning right on arm of maze through habit
- normal performance on place tasks using cognitive map e.g. normal radial arm maze / spatial discriminant water maze
Kernadi (1995)
- monkeys trained to follow certain sequence of dots then repeat pattern by fixating each location in order of presentation, then reaching to target position
- some striatal neurons respond to particular location but only within certain sequence
Thompson (1994)
- rabbit pavlovian eye blink conditioning
- tone/light (CS) then air puff to eyelid (US) = reflexive eye blink (UR)
- several CS-US pairings = CR to CS
- lesion interpositus nucleus in cerebellar cortex = CR not learnt
- inactivation of red nucleus (between cerebellar cortex and motor cortex) = CR learnt but can’t be produced until inactivation reversed (only prevents motor output from cerebellum to cortex)
association builds up in cerebellum and feeds out to motor nuclei
Scoville + Milner 1957 / Milner 1968
- HM had intact procedural learning e.g. mirror drawing
- also intact priming, both perceptual e.g. Gollins picture task and semantic e.g. word association
Cohen + Squire
distinction between declarative and procedural memory
Baddeley + Warrington (1970)
STM vs LTM
Squire + Zola-Morgan 1991
- HC lesioned monkeys impaired on DNMTS task (except for very short delays)
- showed RA gradient for associations before lesion
- can still acquire procedural skills (independent of MTL)
Roof 1993
Gender-specific impairment on Morris water maze task after entorhinal cortex lesion
Yin 2004
anterior vs posterior striatum lesions - only anterior impaired
Hallock (2013)
- conditional discrimination T maze: turn left if mesh, right if wood (simple association between cue and spatial location, learned as habit)
- delayed spatial alternation T maze: remember last trial and choose alternate arm
- transiently inactivated either dorsal striatum or dorsal HC using muscimol
- inactivation of striatum (but not HC( impaired CD task while inactivation of HC (but not striatum) impaired DA task performance
- double dissociation between roles of DS and DH in each task
Sternberg (1966)
- mean RT for digit span increases by 38 ms per digit added (length of gamma cycle so time-matched)
- serial scanning mechanism (must scan whole list)
- MEG recordings: theta power increases incrementally with task load, but doesn’t increase once limit of 7 has been reached
- supports role of theta in maintaining WM
theta + gamma
- single items in an episodic memory e.g. components A-G activated sequentially as a “fast” list on different gamma cycles (30-80Hz)
- 7 +/- 2 gamma cycles on each theta cycle (Ebbinghaus STM capacity)
- theta cycles (4-10 Hz, slow list) with repeated activation of A-G on each one = represent complete episodic memory
represents STM encoding at cellular level - as rat moves through place field, place cell fires earlier and earlier in theta phase i.e. phase-specific firing
- centre of place field = fires at trough (time of firing dictates location)
- theta phase precession and population code = predict exact location from temporal code
Sutherland (1989)
- associations between memories relies on MTL
- trained on paired associates, one rewarded and one not, e.g. AB pair = A rewarded, B not
- then present novel pair BD (but both individual components have appeared before and been rewarded depending on situation and rat has memory of this) = tests transitive interference. should choose B as rewarded over C whereas D is not (AE doesn’t need transivity as A always rewarded and E never rewarded)
- lesion HC (fornix) or paraHC or perirhinal cortices = impair transivity for BD, but AE fine
- rat can still make associations / memories, but cannot compare episodic memories to produce relational behaviour
Rempel-Clower (1996)
- RB, LM, WH
- more extensive damage = more severe AA + RA goes back further
- STM/LTM distinction
- also declarative/non-declarative distinction
Aggleton (1996)
HC damage = impaired explicit recall (but recognition okay)
Bowles (2007)
no HC damage but damage to perirhinal + entorhinal cortices = impaired familiarity but normal recall
Moscovitch (2016)
- regularities of memories stored in neocortical traces and bound together in relational representations mediated by the MTL
- explains why memories can be triggered by situations which have similar elements to a stored memory
- as memories become older and more remote, they become more semantic / abstract and rely less on MTL
MTL damage = no memory traces are transferred to neocortex to be stored long-term
Alvarez (1994) + Lynch (2004)
models of LTP / consolidation by MTL
Kan (2007)
- cued recognition paradigm
- controls showed enhanced performance for same-cue as opposed to different-cue items but amnesics did not, even when they managed to show single item recognition
- suggests the HC is required to bind together information to form relational memory
Verfaellie (2017)
- relational memories are more stable than familiarity info processed in perirhinal cortex and tend to be forgotten through decay rather than interference
- semantic memory can be more well-preserved than episodic memory in amnesics
Dewar 2010
- HC lesions = increased interference effects
Hales (2014)
- MEC lesions only partially disrupt HC place cells and specific types of HC-dependent memory
- bilateral lesion of whole MEC in rats caused lower proportion of active HC cells
- remaining cells had place fields but had decreased spatial precision and decreased long-term stability
- impaired on morris water maze task (like HC lesioned rats)
- combined MEC and HC lesion = even more impaired
- MEC lesioned rats not impaired on other HC-dependent tasks e.g. those where object location or context was remembered
- MEC input not required for all types of spatial coding / HC-dependent memory, but is necessary for normal acquisition of place memory