Amnesia is a deficit in, usually episodic, memorywithout any other cognitive deficits. It can be an inability to recall newinformation (anterograde amnesia) and/or an inability to recall information learnedbefore the amnestic event (retrograde amnesia). Retrograde amnesia will have atemporal gradient, i.e. events that occurred relatively recently around theamnestic event are not remembered. It has been difficult to determine a modelwhich best describes amnesia due to differing lesions and symptoms betweenpatients.
Furthermore, there is debate surrounding the definition of amnesia,and which components of memory are affected. Given the difficulties indescribing amnesia, there is no “perfect” explanation. However, an associativemodel may provide the best explanation because it attempts to account for mostof these issues, despite its shortcomings. This essay will focus on Squire andZola-Morgan’s (1991) associative model, and Aggleton and Brown’s (1999)dissociative model, and the difficulties they both encounter in describingamnesia.
Patient HM (Scoville & Milner, 1957) is one of themost famous memory patients due to his impairment. He had bilateral lesions ofthe medial temporal lobe (MTL), including resection of the hippocampus andamygdala, to reduce seizures. This led to an inability to recollect events andrecognise staff from his hospital stay, however his intellectual abilities and proceduralmemory appeared to be unaffected. E.g. HM was able to learn to trace a disc’smovements, despite an inability to report this experience (Corkin, 1968).
Thisled to Squire and Zola-Morgan (1991) proposing that the MTL is responsible forall declarative memory.The MTL memory system (MTLMS) comprises ofthe entorhinal cortex, perirhinal cortex, and parahippocampal cortex whichforms a reciprocal network with the hippocampus to combine and consolidateinformation across the MTLMS before it is stored in the neocortex. Thus, damageto any part of the system should impair declarative memory. This explains HM’sdeficit because the hippocampus had consolidated memories in his distant past,but his recent past, and was unable to consolidate memories from after hissurgery. Therefore, the memories of the events could not be stored in theneocortex due to the damage in the hippocampus. Zola-Morgan and Squire (1984)attempted to replicate the lesions found in HM in monkeys, and assess their memoryin visual discrimination tasks, tasks in which amnesic patients are impaired,compared to motor skills acquisition. It was found that the monkeys with MTLlesions were impaired on visual discrimination tasks, but not on motor skilllearning. This implies that the hippocampus and amygdala, and thus the MTL, areinvolved in declarative memory acquisition.
Contrary to this, RB (Zola-Morgan,Squire & Amaral, 1986) suffered severe anterograde amnesia resulting from damagerestricted to the CA1 region of the hippocampus, due to an ischemic episode. Itwas, therefore, concluded that damage to the hippocampus alone is sufficient toproduce amnesia. However, there was some damage to the basal ganglia andsomatosensory cortex which may have resulted in cell death without structuraldamage. This could have further contributed to the observed memory impairments.However, human lesions are rarely restricted, and damage to areas outside ofMTL may disrupt connections within the cortex, and produce memory impairments,highlighting the potential difficulties that MTLMS encounters in explainingamnesia.
This model proposes that greater damagecauses a larger functional impairment in memory due to the entire MTLMS being involvedin memory. This has been found by Zola-Morgan, Squire and Ramus (1994), whoshowed that damage to hippocampus and surrounding cortex (H++) impaired memoryon delayed non-match to sample (DNMTS), and object discrimination tasks morethan hippocampal (H+), and hippocampal and amygdala (H+A) lesions. Thissuggests a single memory unit in MTL, so damage to any part impairs declarativememory. This is further supported by Rempel-Clower, Zola-Morgan, Squire andAmaral (1996) who showed that more extensive hippocampal damage in humans (GD,LM, and WH, compared to RB) resulted in more severe amnesia. WH had anextensive bilateral lesion in the hippocampus, with cell loss in CA1, CA2, CA3,dentate gyrus, and entorhinal cortex, and suffered severe anterograde andretrograde amnesia.
This further supports the MTLMS, and the proposal thatgreater structural damage results in larger functional impairments. However, Baxterand Murray (2001) found that monkeys were more impaired in the DNMTS task,which models humans episodic memory, following smaller hippocampal lesions, butlarger rhinal cortex lesions resulted in larger impairments. This implies thatthe different components of MTL may have distinct functions within memory, andcortical regions may play a larger role than was first proposed.
While there is some evidence supporting asingular MTLMS, there are some issues with this model. Firstly, there is debatesurrounding the definition of amnesia. This model assumes that all declarativememory is affected by amnesia.
However, evidence from developmental amnesiapatients implies a dissociation between episodic and semantic memory.Vargha-Khadem et al. (1997) investigated which abilities are spared in threecases of developmental amnesia – amnesia resulting from trauma early in life.These cases all had bilateral damage to the hippocampus, and were impaired inepisodic memory tasks, but not on semantic memory tasks, when tested aged 12-19years.
This suggests that amnesia is a deficit in episodic memory only, andthat the hippocampus is involved in consolidation of episodic memories.However, caution must be taken when interpreting the findings of developmentalamnesia due to the development of strategy use to recall information, as wellas neural plasticity, i.e. neurons change in form and function in response tochanges in their environment, such as damage to a structure. Therefore, it maybe that the hippocampus is also involved in semantic memory and this preservedsemantic knowledge is due to other cortical structures adapting the function ofthe impaired hippocampus.
Squire, Wixted and Clark (2007), therefore, suggeststhat episodic memory recall is more sensitive to impairment, as recall is moredifficult than recognition/semantic knowledge, thus episodic memory traces mustbe stronger to resist impairment.Furthermore, the conflicting evidenceregarding a single entity suggests that this associative model of amnesia is asimplistic view. E.g., BJ (Dusoir, Kapur, Byrnes, McKinstry & Hoare, 1990)suffered from damage to the basal regions of the brain, including the mamillarybodies, due to a snooker cue entering his nose. He suffered severe anterogradeand retrograde amnesia. While his anterograde amnesia improved, verbal memory remainedas impaired as other amnesic patients, and retrograde amnesia covering a periodof 6 months, persisted. This suggests that the hippocampus is not the onlystructure involved in amnesia, and thus causes further difficulty in using anassociative model to describe amnesia.
Moreover, there are a wide range ofaetiologies which causes damage in the human brain, and due to this, damage israrely limited to one region. Annese et al. (2014) reviewed HM’s lesion,finding that damage was not complete in MTL, with CA4 region of the hippocampusbeing spared, and damage extending beyond MTL, including into the left frontallobe being damaged. Likewise, Corkin (2002) suggests that the use ofanti-epileptic drugs, such as that used by HM, can cause corticaldisintegration. This could account for HM’s extensive retrograde amnesia. Therefore,the memory loss observed in all patients may not be a direct result of damageto MTL structures, and further highlighting the difficulties that arise inattempting to formulate a model to describe amnesia.While this model provides an account ofretrograde amnesia, there is difficulty in explaining this deficit due todiffering temporal gradients between patients, and limited evidence on thetemporal nature of consolidation in the hippocampus. E.
g. HM’s temporalgradient was 11years, while BJ’s was 6months. This model attempts to explain retrogradeamnesia in terms of severity of hippocampal damage (Squire, 1994).
However,animal literature shows that consolidation can be in the order of minutes, e.g.long-term potentiation. Dudai, Karni and Born (2015) propose that consolidationand underlying mechanisms differ depending on the task, and how new informationrelates to previously learned information. This may account for theinconsistencies, though further research is needed to confirm this. Overall,conflicting ideas on consolidation, and the differing temporal gradients furtheremphasise the difficulties which arise in explaining amnesia. In sum, there does seem to be a wealth ofsupport for an associative model of amnesia wherein consolidation of declarativememory is dependent on hippocampal function.
However, there is evidence ofamnesia occurring without damage to the hippocampus which highlights thedifficulties in using this model to describe all cases of amnesia. Aggleton and Brown (1999) propose a dissociationbetween semantic and episodic memory. This extends the associative model,proposing that the MTL and the extended hippocampal system, including thethalamus, are involved in declarative memory. They suggest that the medialdorsal thalamus and perirhinal and parahippocampal cortex are involved infamiliarity (analogous to semantic memory), while the anterior thalamus andhippocampus are involved in recollection (akin to episodic memory).
This issupported by single dissociations in animal and human lesion data. Patients KNand YR (Aggleton et al., 2005) have shrinkage of the hippocampus, and sparedsurrounding cortex. They have consistently poor recall ability, but intactrecognition. This suggests that there is a dual system underlying episodicmemory that could account for differing symptoms evident in different patients.Furthermore, patient NA (Squire, Amaral, Zola-Morgan, Kritchevsky, & Press,1989) had damage localised to the thalamus, and suffered severe anterogradeamnesia, especially in recall of verbal material. According to an associativemodel, this patient should not suffer anterograde amnesia, as the hippocampuswas reportedly intact, so consolidation of memories should not be affected.However, this was not found, further emphasising the importance of thediencephalic system in memory and amnesia.
However, pure diencephalic amnesiais rare and further investigation of NA’s pathology showed cortical damagearound the hippocampus in the MTL, so further research into the role of thethalamus in memory is needed.Further research relies on animals,allowing for more precise lesions to be completed, and for potential doubledissociations between recollection and familiarity memory. Using thewhat-where-when (Tulving, 1983) and what-where-which (Eacott & Norman, 2004)compounds, animals’ memories can model human episodic memories, wherein human episodicmemory contains contextual information (e.g.
time) linked with an event and location.E.g. scrub-jays can locate worms and peanuts after short (24hr) and long(124hr) delays, respectively (clayton & Dickinson, 1998). This was interpretedas showing that animals also possess episodic-like memory as they could recall what they had cached where, and when (i.e.
how long ago) theyhad cached items. Eacott and Norman (2004), however, showed that the “when”component of episodic-like memory is difficult for humans and animals asmemories are not date-stamped, so propose that “which occasion” (i.e. context) isa more accurate term that can aid recall of temporal information. They showedthat rats with fornix lesions explored a novel what-where-which combinations lessthan rats with perirhinal and postrhinal lesions, while controls couldsimultaneously retain similar but distinct what-where-which information for twoevents over periods up to an hour. This highlights the importance ofhippocampal projections in recognition memory. However, recognition tasks canbe solved using recollection or familiarity, so results only indicate a single dissociationbetween recollection and recognition. Eacott, Easton, and Zinkivskay (2005)modified this set-up to allow for dissociation between recollection andfamiliarity judgements by using an E-shaped maze.
Following familiarisation ofobjects’ locations in the backbones of different backgrounds, the objects werehidden in the arms of one of the contexts. A rat was, therefore, usingrecollection if they turned towards the correct arm in test phases, but wereusing familiarity when the objects were seen in training. Eacott & Gaffan (2005)found that rats with hippocampal lesions had unimpaired recognition memory in thistask, as well as performing well on components of the what-where-which combination,but were severely impaired on recollective components of this task, showingthat hippocampal projections are involved in recognition, and not recollection.Parker, Eacott and Gaffan (1997) have also found that mediodorsal thalamiclesions in macaque monkeys are impaired in DNMTS task for large stimulus setsonly, suggesting that the mediodorsal thalamus is involved in recognitionmemory. Combining Eacott and Gaffan’s (2005) and Parker et al.’s (1997) studiesgives a double dissociation that supports a dissociative model of amnesia.However, there is conflicting evidence that presents further difficulties indescribing amnesia. Merkow, Burke and Kahana (2015), for example, usedhigh-frequency activity measurements and showed that the hippocampus supportsboth recollection and familiarity in humans, thus does not support thedissociative model, and adds further support to the associative model.
Furthermore,Pergola et al. (2012) who found that the deficits in recollection andrecognition memory correlated with the loss in stroke patients’ mediodorsalthalamus. This does not support the dissociative model, and lends furthersupport to an associative model of amnesia.
It is difficult to present a doubledissociation study investigating the neural underpinnings of recollection andfamiliarity, due to both processes being involved in recognition, and purediencephalic amnesia being rare. However, Bowles et al. (2007) investigated NBusing the remember/know paradigm, which allows for measurement of degree of recollectionand familiarity for events. NB had damage to the entorhinal and perirhinal corticesbut spared hippocampus, and it was found that NB’s recollection memory(“remember”) was unimpaired, while familiarity memory (“know”) wassignificantly impaired. While this study does support a dissociative model, itis only a single case study, but when compared with RB (Zola-Morgan, Squire& Amaral, 1996), a double dissociation in support of the dissociative modelis found. Additional support is found in developmental amnesic patients,wherein the shrinkage/damage to mamillary bodies correlates with the deficit inrecollection, despite preserved recognition and semantic memory (Dzieciol etal., 2017).
There are several limitations with studies of developmentalamnesia, as discussed previously, that present further issues in describingamnesia using a prescribed model. Wixted and Squire (2010) have reviewedhippocampal involvement in recognition, concluding that evidence is convergingon a less sharp distinction between familiarity and recollection, and between thefunctions of the perirhinal cortex and hippocampus, because patients are rarelyonly impaired in recollection or familiarity, implying that the MTL cannot bedivided up as crudely as proposed by a dissociative model.Unlike an associative model, this model doesnot attempt to account for retrograde amnesia, and so cannot describe all casesof amnesia. This presents a further difficulty in describing amnesia using thismodel because patients who only suffer anterograde amnesia, such as NA, can beexplained. This implies that an associative model provides a better account ofamnesia as it can account for both anterograde and retrograde amnesia, despitebeing simplistic in its account.
Overall, there is a wide body of support for bothassociative and dissociative models of amnesia, though no one model adequatelydescribes all cases of amnesia, due to differing symptoms and pathologies in patients.An associative model, compared with a dissociative model, seems to bestovercome these difficulties by explaining amnesia in terms of consolidation, accountingfor retrograde and anterograde amnesia, and explaining recollection/familiaritydifferences in terms of difficulty.