Amnesia,as an organic clinical disorder caused by brain damages, refers to permanentdisturbance of declarative memory despite preserved cognitive functions(Markowitsch, 2001; Moscovitch, 2004). Patients with anterograde amnesia (AA)consistently show different degrees of impairment in forming declarativememory, with normal implicit and working memory. Studying amnesia providesinsights on cognitive and biological mechanism underling memory, and arepresentative model is vital for understanding existing data and makingpredictions. Although researches on amnesia have progressed a lot since thefamous case of H.
M., a convincing and widely accepted model has yet to beestablished. One of the main debates concerning it is whether the majorimpairment in amnesia, declarative memory, is relating to the damage of anassociative or dissociative neural system. The presentation of amnesia variesacross patients, and data are inconsistent.
Hence, it causes a great challenge ofmodel’s generalizability. Since both of them are rather inconclusive inexplaining all the existing data, it is difficult to tell whether one issuperior to the other. Apart from the models themselves, external factors havealso posed difficulties in determining whether one of them is better. Although declarativememory could be subdivided, the interdependent nature of the subcomponentsmakes it difficult to test them separately in a clean way. Furthermore,difficulty in making accurate predictions is also a barricade. To predictimpairments from brain lesions, it is essential to localize the neuralsubstrates of memory in the corresponding brain regions. However, theplasticity of brain and adoption of compensatory strategies make it difficultto map the impairments and preserved abilities in brain.
In this essay, the associativemodel proposed by Squire and the dissociative model proposed by Aggleton andBrown (A) would be used to illustrate the difficulties in determiningwhether one of them better describes amnesia. It is argued that both of them areinconclusive in explaining all the of existing data. Moreover, theinterdependent nature of subtypes within declarative memory and plasticity ofbrain make both of the models not testable or predictable. Tocompare on the same ground, Squire’s model has relative strengths in explainingsingle dissociation in human cases, while double dissociation in animal studiesseems to support A&B’s model instead.
According to the medial temporal lobememory model proposed by Squire, declarative memory issupported by a unitary system which includes hippocampus, entorhinal,perirhinal and parahippocampal cortex (Squire & Zola-Morgan, 1991). As a single entity, degrees ofimpairment should be proportionated to the extent of lesion in the system,regardless of location (Squire et al., 2004). Whereas, A advocates adissociative model concerning the hippocampal-diencephalic system. As claimedby this model, recollection and familiarity are two distinct pathways (Aggleton& Brown, 1999).
To recall episodic memory, recollection supported by thePapez circuit is required. This circuit is connected by fornix andmammillothalamic tract, and includes hippocampus, mamillary bodies (MB) andanterior thalamus nuclei (ATN). Whereas, recognition is subserved byrecollection and/or familiarity. As long as recognition does not involve relationalbinding between elements, familiarity pathway that supported by cortices(perirhinal, entorhinal and parahippocampal cortex) and medial dorsal thalamicnuclei (MDTN) is sufficient (Brown & Aggleton, 2001). In order to validateA’s idea, double dissociation between recognition and recall is required.
The clinical evidence that A&B’s model initially built upon were casesshowing selective impairments in recall, which were suggested to be the resultsof sole damages in recollection but not familiarity pathway (Aggleton , 1999). Further support was provided by subsequent cases thatconsistently presented similar selective impairments. For instance, patient KNand YR who had selective bilateral lesions in hippocampus performed far belowaverage in recall tasks, but relatively normal in recognition of both verbaland visual materials (Aggleton et al.
, 2005; Kopelman et al., 2007). Similarly,patient DN had recollection pathway disrupted by fornix compression, alsopresented selective impairments in recall (Vann et al., 2008). Nevertheless, aswhat endorsed by A in their later review (2006), to date there is norecorded case showing the opposite selective impairments.
Although receiver operating characteristic(ROC) analysis has successfully differentiated recollection and familiaritypathway underlying patients’ preserved memory in recognition (Aggleton et al.,2005; Cipolotti et al., 2006), such statistical dissociation is stillqualitatively different from clinical dissociation. This single dissociationdoes not provide convincing evidence for A&B’s model, but is more likely togo with Squire’s instead. According to his model, recognition could beindependently preserved since it is cognitively less demanding, and would beimpaired only when pathology is severe enough (Squire et al., 2004). In otherwords, it is not the location, but the extent of lesion that is important forexplaining patterns of impairments.
If this is the case, then impairments inrecognition and recall should be proportionate to each other since they belongto one single system. Consistent with this, studies adopted MRI scanning found volumein medial temporal lobe and hippocampus negatively correlated with impairmentsin both recall and recognition (Kopelman et al., 2007; Kramer et al., 2005).However, the discussed evidence is still inconclusive since anatomical lesionmay not be equivalent to functional lesion (Bachevalier, 1996; Aggleton, 2008).
Animal studies enable more confined focal brain lesion, and they couldcomplement lesion studies in human to investigate the neuroanatomy underlyingpatterns of memory impairment. Incontrast to human cases, animal studies have clearly presented doubledissociation between recall and recognition. Eacott and Norman (2004) definedepisodic-like memory in animals as the interaction of eventual (what), temporal-spatial(where) and contextual (which) information. Without either where or whichcomponent, animal’s memory of “what” could be supported by familiarity alone.The speciality of episodic-like memory is probably subserved by therelational-binding function of hippocampus, which enables mental constructionof scenes (Maguire et al.
, 2010). Supporting this idea,