Associative And Dissociative Models Of Amnesia
The debate between associative and dissociative models of amnesia is long-standing and to this day there is no clear answer either way. Squire in 1991 proposed an associative model, stating that a unitary system can explain the way in which declarative memories are stored and therefore this model can account for the pattern of amnesia noted amongst patients. Aggleton and Brown however suggest a dual system, arguing for a dissociative model of memory. This essay will outline both models and discuss the evidence both for and against each one, with the purpose of concluding why this debate is yet to be resolved.
H.M first presented as amnesic after neurosurgery, and being studied for over X years until his death in X. Squire (1991) uses H.M’s amnesic presentation to develop an associative model called the medial temporal lobe memory system (MTLMS). This system suggests that different structures within the medial temporal lobe (MTL) contribute towards memory formation, with this structures being the H++ areas (the hippocampus and adjacent structures). Squire’s theory is that the H++ areas are responsible for the binding of different aspects of a memory, allowing for a memory to be consolidated and then stored elsewhere. Therefore once binding has occurred, the memory becomes independent from the system itself, and therefore damage to the system, such as in the case of H.M would not disrupt memories that have already been consolidated by the H++ areas and stored elsewhere. Squire states that the bigger the damage to this area of the brain the bigger the deficit will be. With the initial paper by Scoville and Milner () stating that the size of H.M’s lesion was 8cm through the hippocampus, his level of impairment was in line with this view by Squire.
It later emerged however that this 8cm lesion as described in the paper was far from accurate, and with evidence that patients with amnesia could score relatively close to normal on tasks of familiarity, Aggleton and Brown (1999) formulated their alternative model. This model states that memory is a dual process, with declarative memory being made up of familiarity and recollection. Tulving (1985) outlines this difference by explaining how remembering involves ‘mental time travel’ with a recollection of what happened. In comparison, knowing/familiarity is the ability to access information but not having specific memory of where this knowledge comes from. Recognition can be performed using either knowing or remembering, whereas recollection requires remembering and is episodic. This dissociation between the two processes is evidenced by Gaffan () showing that amnesics can show familiarity without recognition, as well as cases of YR and KN showing impairment on recollection tasks but ability to show familiarity (Aggleton et al 2005). Aggleton and Shaw (1996) also found that amnesics with hippocampal lesions without adjacent damage to the ++ areas, demonstrate recognise objects or paces, even without recollection. Alongside these, ROC (receiver operating characteristics) curves suggest that lesions of the hippocampus impair recollection without impacting familiarity (Yonelinas et al 1998; Fortin et al 2004). FMRI studies also indicate that during tasks requiring familiarity there is a lack of activity in the hippocampus (Yonelinas et al, 2001; Eldridge et al, 2000), again suggesting the dissociation between the active ++ areas involved in familiarity and the hippocampus inactive during familiarity tasks but active during recall.
In spite of this evidence used by Aggleton and Brown to support their dissociative model, Squire (1991) explains this evidence in terms of the greater cognitive load needed to recall information in comparison to being familiar, causing a ceiling effect on testing. Single dissociations are unable to disprove this explanation, and so it is a reasonable critique. Manns et al (2003) studied patients with specific hippocampal lesions from hypoxia, and found impairment on both recall and familiarity, showing that the results used by Aggleton and Brown in support of a dissociation between the two functions are not always replicable, calling the results and therefore the model based on these results into question. It is clear that until there is a double dissociation able to prove the dissociative model, the evidence isn’t yet strong enough to respond to Squire’s ceiling effect explanation. This serves to highlight the ongoing difficulty encapsulated within this debate where neither side is wholly proven nor disproven.
It could be argued that animal studies provide a double dissociation, potentially solving this debate. Eacott and Gaffan (2005) studied rats, and found a double dissociation; rats with lesions of the hippocampus displayed impairments on what-where-which tasks with normal scores on both what-which and what-where tasks. The opposite pattern was found for rats with ++ lesions. In response to the critique by Squire suggesting ceiling effects being the explanation for results in human testing, this study shows that it is not due to an easier vs harder task, given that the ++ rats showed normal range scores for the what-where-which tasks but impairment on the ‘easier’ tasks of what-which trials. Eacott et al (2005) responded to critiques suggesting results were due to the task being previously learnt, and so a further study was carried out to demonstrate that the what-where-which memory remains, even on unlearned tasks. These studies, demonstrating a double dissociation of familiarity and recollection, support the Aggleton and Brown model that the two functions are structurally distinct.
This type of research is not so simple in its support of this model however. Given the nature of animal studies there are criticisms of using this type of study to support a model referring to human amnesia. Tulving (2002) states that a crucial part of episodic memory is that it has an autonoetic element; one is able to place themselves in the memory. Without language in animals it is impossible to know if they have this conscious autonoetic experience of memory, and in such a way the comparative ability to humans of these studies is called into question. Further to this, animals brain structures are far from identical to humans; how can studies generalise from lesions of animal brains to humans? Whilst the findings from these animal studies do present the dissociation between Tulving’s (2002) what-which memory and episodic-like memory, language would be required to show the dissociation between semantic memory and episodic memory. In such a way, it emerges that one may not reject Squire’s (1987) model in favour of Aggleton and Brown’s (2001) model on the merit of animal research.
Squire’s (1987) model assumes that episodic and semantic memory are linked and inseparable. However there is research to suggest that semantic memories may be constructed even in the absence of episodic memory (Parkin 1982). An example case is that of patient K.C who retained and learnt new semantic information in spite of a severe impairment to their episodic memory (Tulving 1991). Working with K.C to study across a number of weeks, it was found that they could attain semantic information such as three-word sentences. It is acknowledged that the learning rate was not normal, and K.C was far slower, but the ability to acquire semantic knowledge with a severely damaged episodic memory was there nonetheless (Tulving 1991). Tulving (1991) from this concluded that semantic memory may be spared in amnesia; it was argued that the performance on such tasks by H.M was due to the learning method implemented, and that it was this poor method that lead to what looked like both episodic and semantic impairment in contrast to K.C who used a more successful learning method to reveal the true extent of deficit. It is this case that Tulving uses to suggest the sparing of semantic memory. This sparing would therefore mean that the two types of memory are separate and dissociable and therefore support the dissociative model of amnesia.
Squire and Zola (1998) discuss this use of K.C as a demonstration of dissociation. Hamman and Squire (1995) attempted to replicate K.C’s results, using the same learning method (study-only) to also try and teach three-word sentences. While this learning method did increase percentage correct across patients, from 19% to 32%, these results were found to correlate with how much of the previous day patients could recall; those who had more episodic knowledge of the day prior, showed higher accuracy on the tests. In such a way it can be seen how it might be suggested that semantic and episodic memory not be dissociable when there seems to be correlating evidence that increased episodic knowledge links with increased semantic knowledge. Further to this study, Hamman and Squire (1995) tested patient E.P. Patient E.P had severe amnesia with an undetectable episodic memory (Squire and Knowlton 1995). Given several training sessions across a period of two weeks, E.P was unable to show any trace of learning the semantic information, repeatedly scoring 0. Here it emerges that without any episodic memory, semantic memory seems to also be completely non-existent too. In such a way it is not clear how the conclusion can be made that they are dissociable. Squire and Zola (1998) also discuss the way in which there are irregularities in the reporting of K.C’s case alongside his ‘completely dysfunctional’ episodic memory. The nature of K.C’s injury is also discussed within this paper; the damage to the MTL is asymmetric, and there is only a small amount of damage on the right side. There was also generalised damage to other cortical areas on the left side of K.C’s brain. With such problems with the evidence, it would be unwise to draw conclusions and generalise from this research. This research is still unable to disprove the associative model put forward by Squire (1987).
While this discussion has been centred around the idea that amnesia is caused by hippocampal and ++ areas, it has also been suggested that it may be interruptions of other structures or pathways causing amnesia. Gaffan, Parker and Easton (2001) conducted research into this area, to investigate the connections between the MTL and basal forebrain. These areas are connected via acetylcholine projections, with three pathways emerging; the anterior temporal stem, amygdala and the fornix-fimbria. It was found in this study that when either the anterior temporal stem or amygdala pathway was damaged there was a mild impairment. In contrast when the fornix area was damaged a severe impairment was noted. Using immunotoxins, Turchi et al (2005) targeted the cholinergic cells. The results showed identical patterns of impairment to when the white matter connections were damaged. This serves to highlight that anterograde amnesia can be produced solely by disruption of the acetylcholine projections. From this research it can be seen how the matter of which model is correct is further complicated; the cases of patients who have MTL damage alongside basal forebrain damage used to support previous claims on either side of the debate can be called into question given that the extent of the basal damage is never previously considered. In such a way the discovery of these projections make it harder to see clearly which model of amnesia is correct, given that there is now the potential confounding variable of forebrain damage in prior cases.
At present, there is a great body of research on each side of the debate, suggesting one model is more valid compared to the other, but none is completely able to disprove its counterpart. The inaccuracies in reporting alongside the understandable lack of precise lesions in the human research further cloud the argument. With H.M’s case being written up from surgery notes rather than accurate measurements until much later with the introduction of more modern imaging techniques, it is unclear if the foundation for Squire’s model still stands. Equally with the reporting of K.C’s case holding odd statements, such that his scores were ‘comparable to mean scores of amnesic subjects used in experiments in other laboratories’ (Tulving et al, 1991, p 598). Such lack of clarity makes the conclusions drawn lose validity somewhat. Equally the discovery of the acetylcholine projections further confuses the situation as to what side of the associative versus dissociative debate is correct. Prior research in support of either side did not account for forebrain damage and so their conclusions may be inaccurate in their support. This debate essentially comes down to the lack of double dissociations in human cases to prove or disprove either model. It remains unclear whether this lack of double dissociation doesn’t exist due to not discovering such a patient, or because one will never exist due to Squire’s model being correct.