NEUMATICOS SIN CAMARA DE AIRE

Introduction

Memory can be impaired by brain damage at any age. Developmental memory disorders can result from genetic factors or genetic/environmen- tal interactions which prevent the brain from developing normally; or from early-acquired brain lesions. Developmental memory problems are likely to resemble adult memory problems with respect to specific deficits in information processing that depend on dedicated brain structures where there is little scope for transfer of function, as may be true of the medial temporal lobes (MTL) (Bachevalier, this volume, Chapter 2).

The effects of developmental memory problems are, however, likely to differ from the effects of adult-acquired memory problems in several ways. On the one hand, they may have a cumulative effect in so far as they slow or prevent the acquisition of knowledge critical for the development of social skills, language and crystallized intelligence. As a result, devel- opmental memory problems may produce much more severe social, linguistic and cognitive disruption than adult memory problems, as appears to be the case in infant-lesioned as opposed to adult-lesioned primates (Bachevalier, this volume, Chapter 2). Conversely, early- acquired disorders could allow the acquisition of better compensatory strategies and coping skills than occurs in adult-acquired disorders. Also, if a structure has never been functional or even present, other brain structures will not have to adapt to working with it and then without it, so that their functioning may be more optimal.

Despite the differences between the effects of early-acquired as opposed to late-acquired memory deficits, the likely similarities of the actual deficits themselves and their brain bases suggest that the rich set of findings and theories relating to adult-acquired memory disorders can be profitably used as a heuristic to guide research into developmental mem- ory impairments and anomalies. This argument applies to any neuro- developmental disorders in which atypical memory function occurs. However, it particularly applies to autistic spectrum disorders (ASDs) for two reasons.

First: despite the fact that high-functioning individuals have normal or superior intelligence overall, they nevertheless have certain subtle mem- ory impairments, as documented in several chapters of the present book. Second: from the 1970s to the present day there have been persistent suggestions concerning possible parallels between the profile of memory strengths and weaknesses in autism and the memory profile associated with the adult-acquired amnesic syndrome. Early versions of this hypothesis (e.g. Boucher & Warrington, 1976; DeLong, 1978) were proposed at a time when autism was defined partly in terms of learning and language difficulties, such as might result from severe memory deficits. Moreover, it was widely argued at that time that the social impairments in autism might be secondary to severely delayed language, or to cognitive processes underlying the acquisition of language (e.g. Rutter, 1968; Churchill, 1972; Ricks & Wing, 1976). Severe memory deficits could therefore be seen as a possible cause of the social, as well as the intellectual and linguistic impairments of autism as then defined.

Changes in the definition of autism in the 1980s to exclude learning and language difficulties as core impairments, and abandonment of the suggestion that social impairments might result from language-related deficits, made the hypothesis that autism constitutes a developmental form of the amnesic syndrome untenable. This conclusion was reinforced by empirical studies of memory in individuals with high-functioning autism (HFA, used here to include Asperger syndrome) showing that memory is predominantly, though not wholly, intact (see Williams, Minshew & Goldstein, this volume, Chapter 7). Moreover, Vargha- Khadem et al.’s (1997) case studies of developmental amnesia showed that bilateral hippocampal lesions of early origin cause severe episodic memory impairments not accompanied by autism (see Salmond et al., this volume, Chapter 4). Finally, many studies have demonstrated structural and functional abnormalities of the amygdala and associated prefrontal structures in people with autism, sufficient to explain the core social impairments (see e.g. Baron-Cohen et al., 1999; Howard et al., 2000). None of these developments, however, rules out the possibility that a developmental condition analogous to the adult-acquired global amnesic syndrome (associated with impaired semantic as well as episodic mem- ory) causes the learning and language difficulties associated with low- functioning autism (LFA), as suggested within the original hypothesis, and as currently proposed by various authors (Bauman & Kemper, 2004; Boucher et al. 2005; Bachevalier, 1994; also Bachevalier, this volume, Chapter 2 and DeLong, this volume, Chapter 6).

The purpose of this chapter is, therefore, to present a summary of empirical findings and theories from the literature on adult-acquired

organic memory impairments, particularly focusing on those aspects of memory that may have most relevance to autism, including an extended section on the amnesic syndrome. Following this summary, we suggest ways in which insights from the literature reviewed may illuminate autism itself, as manifested across the whole spectrum, including individuals with HFA; and also how this literature might be fruitful in guiding research into the causes of learning and language impairments in LFA.

Adult-acquired organic memory disorders

The idea that there are different kinds of memory, each supported by different sets of brain structures, is mainly based on evidence that non- overlapping brain lesions in human adults cause different memory dis- orders. The initial distinction is between forms of memory that only last for a few seconds (short-term or working memory) and longer-lasting forms of memory (long-term memory). Working memory involves several distinct slave systems that briefly hold very specific kinds of information, such as phonological or visuospatial information (perhaps by continued activity of the representing neurons), and a central executive that pro- cesses what is held in the slave systems and so enables complex thinking to occur (Baddeley, 1992). Working memory is disrupted by lesions of particular parts of the neocortex, prominently including the parietal cortex and probably the frontal cortex. When a specific slave system is disrupted by a lesion, long-term memory is not pervasively disrupted, but long-term memory for the kind of information held in the slave system is devastated (see Mayes, 2000).

Within long-term memory, the major distinction is between declara- tive and procedural memory (see Gardiner, this volume, Chapter 1). Declarative memory includes episodic memory, which is memory for personally experienced episodes; and semantic memory, which is memory for the kind of factual information available in dictionaries or encyclope- dias. A defining feature of declarative memory is that those remembering know that they are remembering, i.e. recall or recognition is accompanied by a feeling of memory. Procedural memories are not accompanied by such a feeling and include memory for motor, perceptual and cognitive skills, various kinds of conditioning, and priming. Procedural memories are, therefore, heterogeneous and only share the feature of not being accom- panied by a feeling of memory.

Lesions of the basal ganglia and cerebellum disrupt certain types of procedural memory. In particular, memory for certain kinds of skill is disrupted by basal ganglia and, more controversially, by cerebellar lesions (Mayes, 2000). Cerebellar lesions also disrupt motoric classical

conditioning (Daum & Schugens, 1995; Mayes, 2000). Emotion condi- tioning is disrupted by amygdala lesions (for example, Bechara et al., 1995). Although individuals with lesions of the amygdala can explicitly remember pairings of previously neutral events with aversive stimuli, they do not show learned development of autonomic responses (such as a change in skin conductance). In other words, they do not learn to asso- ciate automatic emotional responses with the neutral event. Although basal ganglia, cerebellar and amygdala lesions disrupt different kinds of skill and conditioning, they do not disrupt priming, some forms of which seem to be disrupted by lesions to those sensory neocortex regions which underlie representation of the primed information (e.g. Keane et al., 1995). Nor is there convincing evidence that any of the above lesions disrupt declarative memory. The one partial exception may be amygdala lesions, which abolish the usual memory advantage shown for emotional over neutral stimuli, because such lesions reduce emotional arousal and hence the boost to the consolidation of declarative memory that emotion- related stimuli produce in typical individuals (Adolphs et al., 1997).

In contrast, declarative memory is disrupted in a small number of memory disorders involving loss of memory for specific kinds of information – the agnosias, including prosopagnosia, or loss of memory for faces (Benson, Segara & Albert, 1974); also, and most notably, in a condition called the amnesic syndrome. The agnosias are associated with lesions of neocortex specific to the kind of information being processed. For example, prosopagnosia is often associated with lesions of the fusiform face area (Barton et al., 2002). The amnesic syndrome, on the other hand, is associated with damage to any of a number of strongly interconnected brain regions, which include the structures of the MTL, the midline diencephalon and the basal forebrain. There is also evidence that damage to fibre tracts that connect these regions, such as the fornix, which interconnects parts of the MTL with parts of the midline diencephalon (for example, McMackin et al., 1995), causes amnesia. This suggests that normal declarative memory depends on a system of connected structures working together in a highly organized way (Tranel & Damasio, 2002).

Evidence and theories relating to the adult-acquired amnesic syndrome are considered in more detail below.

The amnesic syndrome: empirical findings

The amnesic syndrome is sometimes confusingly referred to as ‘organic amnesia’. To ensure clarity, ‘organic amnesia’ is used here to include any kind of memory disorder that is caused by brain damage or dysfunction,

whereas ‘amnesic syndrome’ is used as the name of a specific kind of disorder of declarative memory as outlined above. It is called a syndrome because it comprises a collection of memory disorders that are in princi- ple dissociable, and because it has many clinical causes including stroke, viral or bacterial brain diseases, alcoholism, various causes of hypoxia (e.g. cardiac arrest) and several others.

Anterograde and retrograde amnesia

The amnesic syndrome is thought to affect, either mainly or solely, declarative memory. The syndrome has two main features. First, affected individuals are very impaired at recalling and recognizing per- sonal episodes or facts that they encountered often only seconds pre- viously prior to being distracted. This symptom is known as anterograde amnesia and life, for people with anterograde amnesia, feels as if they are continually waking from a dream the contents of which fade from view almost as soon as they are grasped. Although people with amnesia are very poor at learning new facts or remembering what they have just experienced, many of them can hold information in immediate or work- ing memory quite normally. For example, their digit spans are often normal. They can also show normal intelligence and, consistent with this, they seem to represent and manipulate incoming information fairly normally.

Second, individuals with the amnesic syndrome have impaired recall and recognition for facts and personal episodes that were encountered prior to their brain damage and were learned normally when they were encountered – retrograde amnesia. It is unclear to what extent anterog- rade and retrograde amnesia co-occur and how strongly they are corre- lated with each other. Although it is accepted that anterograde amnesia can sometimes occur in relative isolation, reports of focal retrograde amnesia have been more contentious (Kopelman, 1991, 2002).

Dissociation from procedural memory impairments

Amnesic individuals can show normal procedural memory, whether this is simple motoric classical conditioning or skill learning. However, they are often unable to recall the learning experience and so deny that they will be able to show conditioning or skill memory (Weiskrantz & Warrington, 1979). There is also evidence that amnesic individuals are unimpaired at some kinds of priming, although it is unresolved whether they show impairments with priming for novel information (Gooding, Mayes & van Eijk, 2000).

Possible subtypes based on processes

It has proved difficult to establish whether or not the amnesic syndrome comprises distinct subtypes, each caused by specific kinds of brain dam- age. If the syndrome does subdivide into several disorders, the division could relate to the different kinds of processes that are disrupted, the different kinds of information for which long-term memory is disrupted, or both. Possible subtypes based on processes are considered first.

Evidence related to process dissociations within the amnesic synd- rome is currently inconclusive. As indicated above, there is some evi- dence that retrograde and anterograde amnesia may dissociate from each other in some individuals, although evidence for selective retrograde amnesia, in particular, is difficult to assess (Kapur, 2000; Kopelman, 2000). Furthermore, even if genuine double dissociations between ante- rograde and retrograde amnesia exist, little is known about their neuro- anatomical bases. Also, most cases of the amnesic syndrome show both anterograde and retrograde amnesia even where damage seems relatively focal, so a single processing deficit may often underlie both post- and premorbid memory problems.

In recent years, most attention has been given to the possibility that recollection and familiarity memory can be separately disrupted by differ- ent brain lesions. Familiarity is a feeling that a target stimulus has been encountered before – a feeling that is not accompanied by recall of associated information. Recollection, on the other hand, involves recall of a target stimulus plus recall of associated information, and a feeling of memory for both the target stimulus and the associated information. Recognition depends on both recollection and familiarity, although the relative contribution of recollection and familiarity varies greatly for different recognition tests.

There is currently controversy about whether lesions of the hippo- campus disrupt recollection but not familiarity and, relatedly, disrupt recall test performance, but have little effect on recognition except where it depends mainly on recollection. Some people with relatively selective hippocampal damage have shown good recognition and impaired recall, and tests suggest that familiarity has been preserved in the face of very impaired recollection (for example, Mayes et al., 2002; Aggleton et al., 2005). Others with relatively selective hippocampal dam- age have shown more or less equally impaired recollection and familiarity (Manns et al., 2003). At present, there is no accepted explanation of why these different memory impairments are found in individuals with appa- rently similar lesions. However, one closely adjacent MTL structure that may be affected in cases such as those described by Manns et al. is the

perirhinal cortex. Large lesions that include this structure disrupt famil- iarity, and its activity changes when human and monkey subjects show familiarity (Brown & Xiang, 1998). The fact that lesions of the perirhinal cortex often also disrupt recollection does not prove that this region mediates recollection. This is because it projects to the hippocampus, so perirhinal cortex damage partially disconnects the hippocampus.

Finally, whilst bearing the above problem in mind, it is important to realize that, although selective disorders of recollection and recall, caused by hippocampal lesions, may occur, there is not yet evidence that recol- lection can be impaired without familiarity also being disrupted.

Possible subtypes based on material specificity

As indicated above, research has also examined whether the amnesic syndrome can be divided up according to the kind of information memory for which is disrupted. Related to the controversy discussed above, involv- ing the functional roles of certain MTL structures, some researchers believe that either episodic or semantic memory may be selectively dis- rupted in different subtypes of the amnesic syndrome. Large lesions that include the perirhinal cortex and other parts of the anterior MTLs severely disrupt acquisition of new semantic as well as episodic memories. However, Vargha-Khadem et al. (1997) found that three young people, who had suffered very early hypoxic damage to the hippocampus, not only showed good recognition and clearly impaired recall and recollection, but had also been able to acquire normal or relatively normal levels of seman- tic knowledge during the years in school. This suggested that hippocampal damage may impair the ability to put new episodic experiences into memory, but leave the ability to acquire new factual memories intact.

Subsequent work, however (for example, Holdstock et al., 2002), found that the ability to acquire new semantic memories is not completely preserved after selective hippocampal damage in adults. Nevertheless, there may be some protection for memories that can be acquired slowly across long periods of time and multiple exposures. Such slow acquisition of factual memories may depend less on the hippocampus, which is critical for rapid learning, than on neocortical structures that acquire memories slowly over widely spaced, multiple exposures. People who have relatively selective hippocampal damage should be very impaired at acquiring episodic memories, which typically depend on a single expo- sure, but will be less impaired at acquiring those semantic memories which can be acquired slowly. They will not show completely typical semantic memories, however, because the rapid learning abilities of the hippocampus also contribute to building these memories.

It should also be pointed out that there is no strong evidence that individuals who have a subset of the lesions that cause the amnesic syndrome ever show a selective deficit in the ability to acquire new seman- tic memories. Some researchers have argued, however, that semantic dementia (a condition caused by progressive atrophy of the anterior temporal lobes, involving a gradually worsening impairment of memory for previously well-established semantic information (Snowden, Goulding & Neary, 1989)) can be associated with relatively greater loss of semantic as compared to episodic memories.

The strongest evidence for subtypes based on material specificity is that which supports the view that amnesia is material-specific in a way that depends mainly on the laterality of lesion location. Thus, it is generally accepted that damage to memory structures in the left hemisphere particularly disrupts verbal memory, whereas right-sided damage par- ticularly disrupts nonverbal kinds of memory (e.g. Milner, 1971). It is plausible to argue that small lesions to the association neocortex may disconnect part of these information representation and processing regions from the MTL-midline diencephalon-basal forebrain complex, which is prevented from performing its vital memory operations on very specific kinds of information. Ross (1982) reported several cases of this kind.

Other, more controversial, claims of subtypes of the amnesic syndrome relating to material specificity concern whether individuals with MTL damage have completely typical encoding abilities (which is a processing issue). Although most individuals seem to be able to encode (process and represent information at input) normally, it has recently been argued that specific MTL lesions cause subtle high-level visual processing impair- ments for specific kinds of information. Thus, some believe that perirhi- nal lesions not only disrupt the ability to remember recently encountered facts and experiences, but also subtly disrupt the ability to process feature conjunctions or make high-level viewer-independent visual representa- tions of objects or faces at input; and that hippocampal lesions not only cause memory disruption, but also subtly disturb the ability to make viewer-independent scene, but not object and face, representations (Lee et al., 2005a, b).

The amnesic syndrome: explanatory theories

Any theory of the amnesic syndrome needs to link specific lesions to a functional deficit that leads to some or all of the symptoms of amnesia. Such a theory would have implications both for how the human brain normally mediates episodic and semantic memory, and for how and why