Testing memory word list

IV.B.3 Interference Theories of Memory

Countless examples of word-list learning experiments have demonstrated that if two related pieces of information are learned, practice of one piece of information can interfere with the ability to remember the other piece of information. This can take the form of prospective interference, in which past information interferes with the ability to retrieve new information. One example of this phenomenon is having difficulty remembering a friend’s married name because her maiden name keeps popping to mind. Alternatively, retrospective interference can occur, in which the learning of new information interferes with the ability to recall old information; the new married name gets in the way of recalling the friend’s maiden name. In either case, recall of one set of information “interferes” with the ability to recall the other. When considering traumatic experiences, imagine the child who’s favorite uncle sexually abused him on one occasion and takes him to a ball game on another. The child may rehearse the uncle–ball game association repeatedly while never rehearsing the uncle–abuse experience due to pressure not to disclose, threats or denial from the uncle, or numerous other reasons. According to standard interference theories of memory, the strengthening of the uncle–ball game association would actually decrease the ability to recall the uncle–abuse situation.

Psychiatric

The effects of diazepam on word list recall, prospective memory, sustained attention and subjective ratings of arousal have been studied in 48 healthy participants, aged 19–35 years, who took oral diazepam mean dose 0.19 mg/kg or placebo in a double-blind study [21]. Retrospective memory and prospective memory were assessed by free recall of unrelated word lists and by instructing participants to ask for a hidden belonging at the end of the session. Sustained attention was measured by multiple trials of a digit cancellation task and subjective arousal was assessed by self-ratings of drowsiness. Diazepam impaired performance on all measures. Diazepam reduced prospective memory performance associated with reduced subjective arousal, but unrelated to sustained attention. This is the first report of the effects of benzodiazepines on prospective remembering and further supports the view that the arousal/attentional system is composed of partially independent subsystems that have differential relationships to memory.

4.11.4.8.4 Verbal memory tests

Phonological versus semantic short-term memory deficits

Aphasic patients are almost universally impaired on simple span tasks involving word list recall (Martin & Ayala, 2004). While neurally healthy adults may have a STM span of about 7 digits and 5 words, aphasic patients typically have spans of 1 to 3 items. A large body of research has examined the causes of these STM deficits and their consequences for aphasic patients’ language comprehension and production. In some theoretical positions, the passive storage involved in word span tasks is assumed to be a buffer that maintains phonological representations (e.g., Baddeley, 1986). In other approaches that take a more language-based approach to STM, both phonological and semantic information are thought to support word list retention. R. Martin and colleagues (e.g., Martin, Shelton & Yaffee, 1994) and N. Martin and colleagues (e.g., Martin & Saffran, 1997) have provided patient data supporting the latter view. For instance, Martin et al. (1994) and Martin and He (2004) have shown that some patients show a deficit specifically in maintaining phonological information and others show a deficit specifically in maintaining semantic information. Those with a phonological STM deficit fail to show the standard effects of phonological variables on span performance (e.g., failing to show an effect of phonological similarity of the words in the list) but do show effects related to semantic variables (e.g., performing better with word lists than lists made up of nonwords such as “pem, dat, tur”). Patients with a semantic STM deficit show the reverse pattern, showing effects of phonological variables, but failing to show effects of semantic variables (for instance, performing at the same level on word and nonword lists). Two tasks that have been shown to discriminate the patients with semantic and phonological STM deficits are the category and rhyme probe tasks (Figure 9-3). In both tasks, patients hear a list of words followed by a probe word. In the category probe task, subjects must judge if the probe word is in the same category as any of the list words. In the rhyme probe task, subjects judge whether the probe word rhymes with any of the list words. Patients with a phonological STM deficit perform better on the category than the rhyme probe task, whereas the patients with a semantic STM deficit perform better on the rhyme that the category probe task. (See also Barde, Schwartz, Chrysikou, & Thompson-Schill, 2010; Hoffman, Jefferies, Ehsan, et al., 2009; Wong & Law, 2008, for recent replications of these findings.) N. Martin and colleagues have presented other data showing different effects of semantic and phonological variables on recall of words which are correlated with patients’ semantic and phonological processing abilities. For instance, a composite measure of semantic processing is correlated with imageability effects on span, whereas a composite measure of phonological processing is correlated with frequency effects on span.

Based on these dissociations, Martin, Lesch, and Bartha (1999) presented a model of verbal STM that includes separate capacities for retaining semantic and phonological information (Figure 9-4). On the left hand side of Figure 9-4 are the types of knowledge representations that one has about words. On the right are buffers used to maintain these different types of representations. In addition to separate buffers for semantic and phonological information, as shown in Figure 9-4, separate capacities are assumed for maintaining input and output phonological representations. Input representations are those derived from the perception of speech. Output representations are those used in speech production. Martin et al. (1999) and others (e.g., Allport, 1984; Shallice & Butterworth, 1977) have presented evidence showing that patients can have poor retention of spoken words on the input side but show normal patterns of speech production (implying preserved output phonological retention), whereas other patients can show normal retention of phonological representations during speech perception but have difficulty in maintaining phonological information used in production.

Implications and Conclusions

There has been considerable interest in using linguistic measures to predict risk, aid in diagnosis, and track the progression of dementia, particularly Alzheimer dementia. Performance on tests of verbal memory, such as immediate and delayed word list recall, as well as tests of verbal fluency, appear to be particularly sensitive indicators of risk for dementia and its onset and seem to covary with other known risk factors, such as apolipoprotein E genotype. However, deficits in verbal fluency are not specific to Alzheimer disease, as discussed above.

The relationship between linguistic ability and Alzheimer disease has been the focus of a number of studies. Declining language ability in late life may be a marker of risk for Alzheimer disease, indicating the onset and progression of this disease. For example, clues to the onset of dementia can be found in the writings of well-known authors including Iris Murdoch and Agatha Christie by examining lexical diversity and syntactic complexity. Thus, changes in language ability in late life may signal incipient dementia. Along similar lines, the Nun Study, a longitudinal, epidemiological study of aging found that low linguistic ability in young adulthood, determined from an analysis of language samples written by the nuns at age 18–32 years, was associated with increased risk for poor performance on cognitive and memory tests in late adulthood, increased neuropathology characteristic of Alzheimer disease, and increased all-cause mortality among participants. A similar analysis of oral speech samples collected from individuals at risk for the development of Alzheimer disease also found that linguistic ability in late life predicts cognitive decline. These findings suggest that linguistic ability may offer a general measure of cognitive and neurological development. Low linguistic ability may reflect suboptimal neurocognitive development, which in turn may increase or reflect susceptibility to Alzheimer and other dementing diseases.

Implications and Conclusions

There has been considerable interest in using linguistic measures to predict risk, aid in diagnosis, and track the progression of dementia, particularly Alzheimer’s dementia. Performance on tests of verbal memory, such as immediate and delayed word list recall, as well as tests of verbal fluency, appear to be particularly sensitive indicators of risk for dementia and its onset and seem to covary with other known risk factors, such as apolipoprotein E genotype. However, deficits in verbal fluency are not specific to Alzheimer’s disease, as discussed above. The relationship between linguistic ability and the risk for Alzheimer’s disease and longevity has been the focus of the Nun Study, an ongoing longitudinal, epidemiological study of aging. Low linguistic ability in young adulthood, determined from an analysis of language samples written by the nuns at age 18 to 32 years, was associated with increased risk for poor performance on cognitive and memory tests in late adulthood, increased neuropathology characteristic of Alzheimer’s disease, and increased all-cause mortality among participants. These findings suggest that linguistic ability may offer a general measure of cognitive and neurological development. Low linguistic ability in young adulthood may reflect suboptimal neurocognitive development, which in turn may increase or reflect susceptibility to Alzheimer’s and other dementing diseases.

Verbal Paired Associates Delayed Word Recall

This delayed free-recall task requires an examinee to recall as many of the words from the list of word pairs as he or she can remember. This task does not require the words to be correctly associated, just recalled. Performance on this task is not considered a measure of associative memory; rather, it is a word list recall task. Also, the juxtaposition of the recognition trial prior to the administration of the free recall condition provides the examinee with another exposure to the correct information, as well as competing incorrect information. Performance on this task is not a true delayed recall condition due to the re-exposure of the stimuli just prior to recall. This is best interpreted as a measure of immediate word recall for previously learned material. Low scores suggest that the examinee did not encode many of the words required to perform the association task and did not benefit from re-exposure to the word pairs. The examinee may have a more basic word list learning problem that may have limited the amount of verbal associations encoded. Also, the task is a free-recall measure compared to the Verbal Paired Associates I and II tasks, which are cued recall. Problems with retrieving information from memory may also contribute to low scores on this measure.

Frontotemporal Dementia

There are three clinical presentations of frontotemporal dementia: a neuropsychiatric syndrome, nonfluent progressive aphasia, and semantic dementia.³⁷ Of these, semantic dementia is of most interest with respect to impainment of declarative memory.³⁸ As the name suggests, these patients have progressive semantic memory impairment (factual knowledge, word meanings, and object knowledge), which gives rise to comprehension deficits and fluent aphasia. Unlike post-HSVE cases, significant category-specific deficits in semantic knowledge are rarely encountered, possibly because HSVE is more likely to result in a patchy distribution of cortical damage. In fascinating contrast to amnesic syndromes, patients with semantic dementia have relative preservation of episodic memory, as evinced by their often remarkably rich ability to recount anecdotes from the recent past. An important caveat for the neuropsychological assessment of episodic memory in semantic dementia is that the semantic knowledge deficit confounds performance on verbal memory tasks. Word-list learning or story recall is aided under normal circumstances by the ability to make semantic associations; if semantic knowledge is degraded, then word-list learning is analogous to normal subjects’ learning a list of unfamiliar foreign-language words. Nevertheless, because degeneration is usually maximal in the left temporal lobe, a degree of verbal episodic memory impairment is difficult to rule out. However, on nonverbal tests that minimize semantic associative knowledge (e.g., the Rey-Osterrieth Complex Figure Copy Test [Fig. 4-9]), delayed recall scores are often within the normal range.

The remote memory profile also provides an interesting contrast to that seen in classic amnesic syndromes. Recent episodic memory (dating back weeks to months) is relatively preserved in comparison with memory from more remote time periods,³⁹ a finding that is also true of semantic facts that can be dated to a specific time period (such as knowledge of famous people and events).⁴⁰ These findings suggest that remote autobiographical memories may actually be supported by similar mechanisms to those involved in semantic memory.

The distribution of degenerative change in semantic dementia is typically asymmetrical and involves the anterior temporal lobes (especially the poles and inferior surfaces) (Fig. 4-10). The preservation of episodic memory was initially thought result from sparing of the hippocampi; however, volumetric MRI has shown that hippocampal atrophy in semantic dementia is at least as severe as that seen in Alzheimer’s disease, which suggests that the amnesia in the latter may be more a consequence of damage to other areas. A final contentious issue in semantic dementia is whether the asymmetrical atrophy of the temporal lobes can give rise to material-specific semantic memory deficits. Some authors have suggested that patients with greater right-sided atrophy have more problems with nonverbal semantics⁴¹ (such as prosopagnosia for famous faces). The alternative view is that semantic knowledge is bilaterally distributed but naming is lateralized to the left. Consequently, verbal semantics are more impaired with greater degrees of left temporal atrophy, because this weights word-processing ability.

IV Population Studies

In the laboratory studies of the LE described previously patients were selected on the basis of neurologic complaints. Although these types of studies have been helpful in understanding the nature and severity of neurologic deficits in Lyme disease, they tell us little about the prevalence of these symptoms. Nancy Shadick and colleagues studied the prevalence of persistent neurologic symptoms in a sample of unselected patients with a history of Lyme disease in Ipswich, Massachusetts, a community endemic for Lyme disease. These investigators initially studied 38 people who met CDC criteria for previous Lyme disease and 43 people who did not. There was a slight but statistically significantly difference between groups on the CVLT, with the Lyme disease group performing more poorly. Twelve of the 38 Lyme patients scored two or more standard deviations below mean on the word list test, compared to only 5 of the 43 controls. Patients with residual symptoms, neurologic and musculoskeletal, had longer duration of disease prior to treatment. These findings suggested that a small percentage of patients with previous Lyme disease may have permanent learning and memory deficits, albeit subtle. In a larger subsequent study on Nantucket, another community highly endemic for Lyme disease, these investigators compared 186 people with prior Lyme disease to 167 healthy controls using similar measures including the CVLT. Patients were studied an average of 6 years after infection. Although the patient group reported a higher incidence of nonspecific symptoms, including fatigue, difficulty sleeping, memory impairment, and poor concentration, there were no significant differences between groups on any of the objective tests of memory and concentration. At least two other population studies produced similar findings—namely, that the prevalence of objective measures of LE in previously infected patients who were treated for Lyme disese is very low. Thus, although patients previously infected with Lyme disease may report more neurologic symptoms than never infected controls, there is little evidence of any objective deficits. The relationship between reports of perceived memory dysfunction and performance on memory tests is discussed later.

Memory deficits

Memory deficits are often prominent and a leading complaint. The international consensus paper referred to “working memory deficits” or “short-term memory loss” and did not give an operationalization. The intended meaning was that remote memories usually appear unimpaired, whereas patients perform poorly during one-time tests in the presence of an investigator (“short-term memory”). This terminology has been criticized. It has been pointed out that this “clinical” concept of “short-term memory” has in fact been disentangled into working memory and declarative-episodic memory. Typical tests for “working memory” assess the memory span: How many items—for example, digits—can a proband correctly repeat back immediately after presentation? Episodic memory, by contrast, is typically examined by word list learning tests with delayed recall like the California Verbal Learning Test (CVLT) or the Verbal Learning and Memory Test (VLMT). These word lists are longer than the capacity of the working memory (e.g., 15 items). Nonverbal episodic memory tests are also available. One group of researchers emphasized the dissociation of preserved “working memory” and impaired “long-term anterograde memory” in six patients with LE associated with different abs (Nascimento Alves et al., 2017). Another study performed more sophisticated tests on seven patients with high-concentration “VGKC abs” (probably LGI1 in today’s nomenclature) > 1 year after the disease nadir. All had developed left-sided or bilateral hippocampal atrophy. The authors reported on impaired recent episodic memories but less striking impairment of remote episodic memories. They documented preserved personal semantic memory. There were recall but not recognition memory problems (Lad et al., 2019). One careful long-term study of a patient with relapsing LE due to LGI1 abs illustrates the distinction between an always normal digit and visual span but subnormal performance on a verbal list learning test—more often in the recall than in the learning aspects (Zangrandi et al., 2019). Recently, the boundary between declarative-episodic performance as hippocampal function and working memory as a frontal task has become more permeable: researchers have noted in LE patients a mediotemporal linkage of different types of information, potentially situated in different regions of the brain, regardless of memory duration (Pertzov et al., 2013). For clinical purposes, however, it is recommended to rely on established tests and concepts: in LE, strong deficits are typically observed on delayed recall during list learning tests (episodic memory), whereas the digit span is usually unimpaired (working memory).