Theories of hippocampal function Flashcards
Dissecting hippocampal from amygdala function
Mishkin et al 1978 - surgically H+A+ lesioned monkeys, to mimic HM. Found more severe amnesia than in H+ lesioned monkeys. This led to the dual route hypothesis - A+ and H+ lesions performed as control, H+A+ were impaired. However, en passant fibres (of which there are many more in the amygdala) were also damaged, and Mishkin damaged rhinal cortex too (because he lesioned from below)
Later studies found both H+ and H+A+ lesions impaired performance on simple object discrimination, concurrent object discrimination, and DNMTS, the latter showing delay-dependence (so not a simple perceptual failure, must be memory involved).
Murray and Mishkin 1998 - Excitotoxic lesions of HA perform as controls on DNMTS!
None of the H, A, or HA lesions, using any method, had motor skill learning impairment.
Overall, it seems hipocampal lesions caused mnemonic impairments, additional damage to adjacent cortex caused object discrimination impairments, and amygdala damage did not contribute to the amnestic syndrome.
DIssecting hippocampal from PRC function
PRC contains cells that fire for a particular object, and fire less after repeated presentation of that object. So they may be novelty detectors.
Malkova et al 2001 - PRC required for DNMTS task.
Baxter and Murray 2001 - reviewed 3 papers, found positive correlation between PRC damage and DNMTS impairment, but negative correlation between H damage and DNMTS impairment.
Penfield called PRC the ‘silent cortex’, because it didn’t do anything when poked in awake patients, whereas poking the hippocampus or amygdala caused patients to imagine symphonies, remember childhoods etc
Bussey et al 2003 - PRC lesions disrupted object recognition only when discriminations were perceptually difficult and slowly learnt, esp if they involved feature ambiguity.
Bussey et al 2004 - PRC but not H lesions affected object discrimination; H but not PRC lesions affected radial maze test scores.
PRC probably involved in semantic memory - Vargha-Khadem’s kids had more selective hippocampal lesions, and intact semantic memory.
Yonelinas et al 2002 - H lesions were impaired more at recall than familiarity. H+ lesions were impaired equally at both.
BUT Manns et al 2003 - H lesions were impaired at both.
Yonelinas et al 2004 - Manns’s results were because patients were of varying etiologies, whereas his own were all cardiac arrests and mild hypoxia.
Aggleton et al 2005 - Reported patient KN, post-meningitis H damage (no PRC damage), impaired equally at recall and familiarity.
PRC lesions impaired visual and tactual memory, but only after a delay of at least 10s. TE lesions impaired it after 1s.
Declarative theory
Proposed by Squire and colleagues - there is no dissociation within the hippocampus, it’s required for semantic AND episodic memory (though not necessarily storage)
The hippocampus encodes new memories, then consolidates them to the neocortex.
Manns et al 2003 - H damaged patients had semantic impairments that were NOT simply a byproduct of episodic impairments, because controls’ semantic memory was not improved by remembering the episode where they’d learnt it.
Evidence against (e.g. vargha-khadem’s kids) may be because their damage was sustained early, so there was rerouting. But what about semantic dementia/alzheimer’s dissociation?
Note: non-declarative memory is thought to reside in the basal ganglia.
Associative learning theory
The hippocampus stores simple, specific stimulus-response associations
Some studies have shown a specific deficit in associative learning with hippocampal damage
BUT some of these studies had varying etiologies, and many others have found both associative and single-item learning impaired.
Familiarity
Yonelinas et al 1998 - patients with selective H lesions were impaired on recollection but not familiarity
In rats, you can use an innate preference for familiarity that means the rat will always pick the old object if it can recognise it. Rats are impaired on this task when 50% of hippocampus volume is damaged. C.f. morris water maze, performance impaired when 30% of H volume is damaged.
These two are contradictory! Maybe both familiarity and recollection rely on hippocampus, but former requires less circuitry to is less likely to be impaired in any given H lesion.
Difficulties with imaging studies
Activity can been seen in an area that, when lesioned, does not impair activity (e.g. eyeblink conditioning in rabbits).
Depends on the baseline used - e.g. looking at novel pictures causes no hippocampal activity when rest is the baseline, but robust bilateral activity when judging if digits are odd or even is the baseline.
Individual anatomical differences - can collapse across multiple voxels but lose resolution, or maximise overlap in the region of interest at the expense of whole-brain overlap, or use cortical unfolding technology to produce a 2D map.
Cognitive map theory - FOR
MacGuire et al 1997, 1998, 2000, 2006 - London taxi drivers had larger posterior hippocampi and smaller anterior. This effect is greater when subject has been a taxi driver for longer. Parahippocampal and posterior hippocampal activity when learning one’s way through a virtual or imagined environment. Stronger during wayfinding than travelling a wellknown route, stronger during route-learning than aerial view. Patient TT was a taxi driver with bilateral hippocampal damage, who was impaired at navigating through a virtual London, but fine when only using the main roads - so he’d lost his allocentric map.
Morris et al 1982 - water maze performance impaired when H lesioned
Pearce et al 1998 - lesions don’t impair ability to head in a particular direction to a stimulus that bears a fixed relation to the platform
Cognitive map theory - spatial context and a closer look
The taxi driver experiments (and others) showed activity in right hippocampus and parahippocampus only, not the left - patients with left hippocampal lesions can navigate and path integrate as controls, but are impaired at recalling location of objects on a table.
Lesions impaired navigating without visual cues (so it’s not just using signposts)
Place cells are most easily observed in an empty room
Gaffan et al - H lesions imair DNMTS performance on object-in-place tasks, but not when context varied on each trial. So they were required to link spatial context to object in space.
Dorsal (but not ventral) lesions impair Morris water maze performance. Ventral (but not dorsal) lesions impair context-dependent fear conditioning
Lesioning the hippocampus soon after context fear conditioning lessens the fear response. Lesioning 28 or more days afterwards does not. Context, lesion, training rescues the behaviour. BUT H lesions cause hyperactivity, maybe this is enough to overcome context-dependent freezing, while leaving CS-dependent freezing intact?
Lesions impair tasks other than spatial tasks, so this can’t be the whole story
Cognitive map theory - egocentric/allocentric dissociation
Patient TT (taxi driver, bilateral H lesions) was able to navigate using main roads, i.e. egocentric navigation was unimpaired. Rats whose hippocampi were injected early in training with lidocaine made more response-based navigation choices (in a double T maze). Late in training, more normal responses were egocentric and injecting H had no effect, but injecting caudate caused an abrupt shift to allocentric navigation
LTM (a mixed picture)
Takehara et al 2002 - mice H lesioned 1 day after eyeblink conditioning lost their response. Mice lesioned 4 weeks after conditioning were intact. The opposite pattern is seen with frontal lesions.
In humans, remote spatial memory is spared by H lesions
In rats, remote spatial memory is NOT spared by H lesions.
Egocentric vs allocentric map - definitions, and evidence that rats have an allocentric map
Egocentric - The map only exists in relation to you, there is no route apart from the one you take, you navigate by landmarks
Allocentric - The map is separate from your place in it. You can be dropped anywhere in it and figure out your route from there.
Tolman - trained a rat to go into a single portal (which led ahead then right) to a reward chamber. When the rat was put into a radial maze, instead of going straight for the same portal it instantly took the ‘shortcut’ portal. This suggests it was using an allocentric map, and defies behaviourism.
Place cells - discovery, properties etc
Discovered by O’Keefe and Dostovsky 1971
Large, consistent firings = place fields
Place cells rotate with visual cues, fire independently of directions, and ‘remember’ locations
Moser et al 2008 - discovered ‘grid cells’, which converge on a single place cell, perhaps giving it its place specificity
Place cells can lose and change their place association if the environment is abruptly changed, or gradually becomes more similar to a different, familiar environment.
Place cells’ firing is affected by odour, rat’s behaviour, and the passing of time
Place cells show pattern completion and pattern separation, indicative of an attractor network.
Humans have cells that behave like place cells but for celebrity faces - maybe the idea of ‘place cells’ is generalisable.
Multiple trace theory
the hippocampus and PRC and PFC create memory traces, which get strengthened every time they’re recalled and eventually get combined with pre-existing knowledge and ‘semanticised’ to the neocortex. Whilst remote memories are better protected from hippocampal lesions, they’ll still be weakened until fully semanticised.
This differs from declarative theory, because declarative theory suggests that remote memories are totally safe from hippocampal lesions.
BBB model
H - path integration, codes location using self-perceptual cues
PHC - boundary vector cells fire based on bearing and distance of edges from us, allocentric NSEW respresentation
PRC - cells fire to specific visual objects
~reciprocal connections between all three constantly update location in space~
Medial parietal cortex - head direction cells tune all three, determines our ‘view’ of an imagined space, egocentric representation
Frontal lobes - send simulated self-motion efferent copy
Posterior parietal, retrosplenial, parieto-occipital sulcus, head direction cells in Papez’s circuit - translate PHC allocentric to medial parietal egocentric representation
This whole process can perhaps be generalised to episodic retrieval.
Representational-hierarchical theory - assumptions
- hierarchical organisation of representations - they get more complex as you go down the processing stream (just like how visual processing is thought to work already)
- Content not process - there are no substantial processing differences between the modules, and any memory dysfunctions can be explained purely by a failure of content
- A brain region is useful when the representation stored in that region is useful for a task (rather than when the function that region carries out is useful for a task)
