Article Text

Download PDFPDF

Lifting the veil: how to use clinical neuropsychology to assess dementia
  1. James R Burrell1,2,3,
  2. Olivier Piguet1,2,3
  1. 1Neuroscience Research Australia, Sydney, New South Wales, Australia
  2. 2University of New South Wales, Sydney, New South Wales, Australia
  3. 3ARC Centre of Excellence in Cognition and its Disorders, Sydney, New South Wales, Australia
  1. Correspondence to Dr James R Burrell, Neuroscience Research Australia & University of New South Wales, Barker Street, Randwick, Sydney, NSW 2031, Australia; j.burrell{at}


Neurologists often struggle to interpret the results of neuropsychological testing, even though cognitive assessments are an integral component of the diagnostic process in dementia syndromes. This article reviews the principles underlying clinical neuropsychology, background on common neuropsychological tests, and tips on how to interpret the results when assessing patients with dementia. General cognitive screening tools, appropriate for use by general neurologists and psychiatrists, as well as specific cognitive tests examining the main cognitive domains (attention and orientation, memory, visuospatial function, language and executive function) in patients with dementia are considered. Finally, the pattern of deficits, helpful in defining clinical dementia phenotypes and sometimes in predicting the underlying molecular pathology, are outlined. Such clinicopathological associations will become invaluable as disease-modifying treatments for dementia are developed and implemented.

View Full Text

Statistics from


Neurologists are trained to investigate complex medical and personal histories and perform detailed neurological examinations. Seamlessly, they proceed to neurophysiological tests or engage vigorously with neuroradiologists to interpret neuroimaging. Similarly, psychiatrists deftly synthesise complex psychiatric and behavioural histories to determine the presence of affective and anxiety disorders, or psychosis in later life. In contrast, cognitive testing is often deferred to neuropsychologists. Why is it that general neurologists and psychiatrists appear to lack understanding of the benefits of neuropsychological examination? Here, we review the principles underlying clinical neuropsychology, background on common cognitive tests, and tips on how to interpret cognitive test results when assessing patients with dementia (box 1). In addition, we examine some of the more controversial areas in dementia assessment, where clinical assessment and neuropsychology may need to be interpreted in a wider context.

Box 1

What to look for in a neuropsychology report

  1. Premorbid cognitive ability (how was it estimated?)

  2. History

  3. Test conditions

    1. Effort

    2. Medical/medication confounds

    3. Comorbid depression/anxiety

    4. Linguistic/cultural background

  4. Cognitive screening task

    1. Mini-Mental State Examination (MMSE)

    2. Montreal Cognitive Assessment (MoCA), or

    3. ACE-III

  5. Domain-based cognitive testing (with reference to ‘normal’ performance*)

    1. Attention/orientation/working memory

    2. Memory

    3. Language

    4. Executive function

    5. Visuospatial ability

    6. Social cognition

  6. Summary of findings

  7. Neuropsychological diagnosis

*Normal performance may be defined in reference to a population normative value, or to the estimated cognitive ability of the individual patient, based either on a reading task or educational/vocational attainment

In the right context, neuropsychological testing provides reliable information on cognitive function. It helps determine an individual's cognitive strengths and weaknesses, informs functional capacity and documents progression and/or evolution of cognitive deficits over time. The pattern of cognitive deficits provides insights into normal brain function and underlying pathological processes. Nonetheless, the interfaces between neurology, psychiatry and neuropsychology in the assessment of dementia are complex and an understanding of the underpinnings of clinical neuropsychological is valuable and will improve diagnostic accuracy.

What is neuropsychological testing?

A neuropsychological assessment encompasses a series of different tasks designed to probe specific aspects of cognition. Six independent, but overlapping, domains are outlined in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-V): complex attention, learning and memory, perceptual/visuospatial/visuoconstructive, language, executive function and social cognition.1 Each domain can be further subdivided into specific subdomains.

Cognitive functions have more or less precise biological substrates, although evidence from lesion and neuroimaging studies has demonstrated that most involve distributed neural networks and various brain regions. Although important for the understanding of normal brain function, and the mechanisms by which pathological processes can result in specific clinical features, a review of the neuroanatomical regions and networks underlying specific cognitive domains is beyond the scope of the present review.

Although cognitive dysfunction may be suspected following a careful clinical assessment, neuropsychological testing quantifies the extent and severity of change. Furthermore, the pattern of cognitive deficits may evolve over time, prompting revision of the syndromic diagnosis in individual patients.

What are the pitfalls in neuropsychological testing?

Importantly, the neuropsychological assessment is interpreted within the broader context of the medical and personal history (obtained from both the patient and family), psychiatric history, neurological examination, neuroimaging and where available other investigations (eg, blood tests, genetic testing).

A number of medical and non-medical variables have the potential to influence cognitive performance on testing. First, reduced or variable effort, which may be inherent to the underlying disease prompting the assessment, can affect capacity to sustain concentration for long periods of time. Comorbid physical and psychological illnesses (eg, respiratory, renal or hepatic failure, drug intoxication, depression, anxiety) can also affect performance.

Second, most neuropsychological tests are administered using verbal instructions—and often require a verbal or written response. As such, ensuring that patient and examiner can communicate with each other appropriately is important. The use of medical interpreters only partly accounts for this potential confound. Culturally biased tests and level of education are other potential confounds. Additionally, it is important to bear in mind that language deficits are inherent to a number of dementia syndromes. Such deficits may interfere with the assessment of non-language cognitive domains, regardless of the language used to perform testing.

Finally, performance on cognitive testing is not meaningful in itself, and scores need to be compared to a relevant benchmark (ie, comparative or normative scores). Normative scores are usually derived from testing of healthy individuals, generally controlling for such variables as age and/or education. Norms, however, may not be able to control for all relevant variables, for example premorbid cognitive ability in an individual patient (more on this later). It is worth emphasising that, while people perform better on some tests than others, healthy individuals tend to obtain a fairly homogeneous test performance across cognitive domains (eg, memory, language, executive function). A disproportionate impairment in one or more cognitive domains, while not pathognomonic in itself, may reflect a disorder of cognition. It is also the pattern of deficits that is most informative diagnostically.

General cognitive screening tests

A vast array of neuropsychological tasks is available to probe specific cognitive abilities, many of which are time-consuming and labour intensive. Clearly, a comprehensive neuropsychological examination for every individual is not always possible or necessary. Nonetheless, even busy clinicians can administer a brief cognitive screening task, which provide an estimate of overall cognitive function, and help identify patients who require a detailed cognitive evaluation.

Mini-Mental State Examination

The Mini-Mental State Examination, or MMSE, is probably the most recognised cognitive screening test. The MMSE was designed to distinguish cognitive impairment due to dementia, from apparent cognitive impairment in psychiatric illnesses.2 Its administration usually takes less than 10 min, even in patients with dementia and it provides a broad index of cognitive function, where a score of 25 or below, out of a possible 30 points, is indicative of cognitive deficits.

The MMSE is widely used by clinicians from all levels (eg, junior hospital doctors, registrars, specialists) and specialties (eg, general practitioners, neurologists, psychiatrists, geriatricians). Importantly, the MMSE was not designed to distinguish different forms of dementia. It is heavily weighted towards the orientation/attention and memory domains (70% of the available points). Assessments of language and visuospatial function are very brief, and executive function is essentially ignored. As such, the MMSE is insensitive to cognitive deficits encountered in some dementias like frontotemporal dementia (FTD). The MMSE is now subject to copyright, making it an expensive option for routine clinical practice.

The Montreal Cognitive Assessment

The Montreal Cognitive Assessment (MoCA) was first published in 2005.3 Like the MMSE, the MoCA is scored out of 30. Unlike the MMSE, however, attention and memory make up only half of the available points. Tests of language, executive function and visuospatial abilities are also included.

The MoCA offers a number of advantages over the MMSE. First, the assessment of a broader range of cognitive deficits results in a higher sensitivity of the MoCA in detecting mild cognitive impairment, and a cut-off of 26/30 detects mild cognitive impairment with a sensitivity of ∼ 90% and specificity of 87%.3 It has been translated into, and validated in, many different languages and administration time of the MoCA is comparable to that of the MMSE.

The Addenbrooke's cognitive examination—version III

The Addenbrooke's cognitive examination, now in its third iteration (ACE-III),4 is the most detailed cognitive screening test in wide usage. Similar to the MoCA, the ACE-III covers a range of cognitive domains including attention/concentration, memory, verbal fluency (executive function), language and visuospatial ability. It is scored out of 100 points and a cut-off of <88/100 detects dementia with high sensitivity and specificity. Subscores for each of the main cognitive domains can also be obtained (attention/concentration 18 points, memory 26 points, verbal fluency 14 points, language 26 points and visuospatial function 16 points).

One advantage of the ACE-III is its ability to distinguish different dementia syndromes by considering the pattern of cognitive deficits, rather than just the severity of any impairment. In particular, the ACE-III reliably distinguishes FTD and Alzheimer's disease, primarily because of the more detailed language and executive assessments. These benefits are achieved with only a modest increase in the administration time (12–20 min)5 compared to other screening tools. The ACE-III is also useful for tracking the progression of cognitive deficits over time, because it includes a broader range of scores. While originally developed as a pen-and-paper test, the ACE is also available in an electronic version (ACEmobile), which allows administration and scoring on tablet devices (see

Overall, it is important to remember that broad screening measures are just that: they may not be sensitive to subtle cognitive deficits, especially in very intelligent or highly educated individuals. This emphasises the need for specific and cognitively demanding tasks in such individuals.

Domain-specific neuropsychological testing

A neuropsychological assessment aims to probe each cognitive domain independently. In practice, however, most cognitive tasks are not ‘pure’ measures of a single cognitive domain and optimal performance usually relies on a combination of cognitive processes. An understanding of the domains tested by each task is required to interpret the results. Comprehensive neuropsychological testing may take between 2 and 7 h, depending on the presenting symptoms and severity of deficits.

Estimating general premorbid cognitive ability

An estimate of premorbid general cognitive function is an internal benchmark against which performance on cognitive testing is interpreted. An individual of high cognitive aptitude should perform better on neuropsychological tasks than someone with low cognitive capacity, even in the context of early dementia. The simplest method to estimate premorbid ability is to use educational and vocational attainment. Another common approach is to assess word reading ability, which reflects lifelong learning/knowledge capacity and is relatively resistant to the effects of cognitive decline and ageing.6 In English, one such task is the National Adult Reading Task (NART),7 which is composed of 50 irregular words of decreasing frequency. Knowledge and pronunciation of irregular words (eg, ‘cello’), which cannot be read correctly with the application of common phonetic rules, can only be acquired through exposure and correlate with general intelligence.

Attention and concentration

Cognition may be considered hierarchical, with aspects of attention and concentration necessary for other cognitive abilities. Patients must be able to concentrate on the cognitive tasks at hand during the examination or testing will at best uninterpretable and at worst impossible to perform. As such, concentration is often investigated first, either by assessing ability to respond to simple questions or commands (provided language is intact), or by observing distractibility, lethargy and drowsiness during interactions and testing.

Formal tests of attention are varied but include span tasks. Digit span forwards, for example, examines the ability to sustain attention. A series of digits is repeated in order, with the number of digits increasing as the test proceeds. English-speaking adults are generally able to repeat 6–8 digits in the correct order. Separately, ability to divide attention can be assessed by so-called ‘Oddball’ tasks, which require a response to an infrequent (and irregular) target stimulus presented among a series of standard stimuli.8 Other simple tasks include counting from 1 to 20, reciting the days of the week or months of the year, followed by the same tasks in reverse order.

Working memory is sometimes conceptualised as a separate domain that is aligned with, but distinct from, executive function. Clinically, however, it is often examined with attention/concentration abilities with such tasks as digit span backwards. This task requires a sequence of digits to be repeated in reverse order.7 ,9 ,10 Most normal people can repeat at least four digits in reverse, but patients with significantly impaired working memory may be unable to repeat more than two digits.


Given its prominent breakdown in many conditions, memory has attracted more attention than other cognitive domains. Memory is conceptualised in different ways such as ‘declarative’ or ‘explicit’ when referring to conscious recollection of facts, or ‘non-declarative’ or ‘implicit’ when referring to memory for skills and procedures. Declarative memory is further subdivided into episodic memory (eg, where you went to dinner on your last birthday) or semantic memory (eg, knowledge of the Australian states or English counties). Generally, only declarative memory, particularly episodic memory, is routinely tested, with verbal and visuospatial (non-verbal) aspects tested separately.

Most memory tests assess various components necessary for adequate memory performance and can be summarised as follows: (1) encoding (ie, capacity to take in novel information), (2) retention (ie, capacity to hold this information over time); and (3) retrieval (ie, capacity to bring back this information after a delay). Retrieval can be examined further using free recall (no external assistance), cued recall (in response to a general or specific cue) or recognition. In general free recall is more difficult than cued recall, which is in turn more difficult than recognition.

In clinical practice, verbal memory is tested using verbal information that varies in grammatical and semantic structure (see Lezak et al10 for an exhaustive review). In addition to prose passages (eg, WMS Logical Memory), common tasks include pairs, arrays or lists of words with variable semantic relations (eg, Paired-Associate Learning, The Free and Cued Selective Reminding Test (FCSRT), Rey Auditory Verbal Learning Test).7 ,10 Most of these tasks include immediate and delayed (ie, after 30 min) recall (free and/or cued) components, as well as recognition. The FCSRT appears to be particularly sensitive in differentiating episodic memory deficits due to mild cognitive impairment from those due to early Alzheimer's disease,11 ,12 and in the distinction of Alzheimer's disease and FTD.13 The FCSRT has been incorporated into trials of antiamyloid therapies.

Visuospatial memory is assessed by recognition of visual stimuli or by recollection and reproduction of line drawings from memory. For example, in one of four components of the Doors and People test7 patients are presented with 12 pictures of different types of doors to memorise. Correct recognition of doors from an array of four similar doors responses is scored. Reproduction from memory of a complex line drawing (figure 1)6 ,7 ,10 is another common test of visuospatial memory.

Figure 1

The Rey Complex Figure. (A) In the first part of the test, patients are required to copy The Rey Complex Figure. Time taken to produce the copy is recorded, and accuracy is scored for comparison with controls. In the second part of the test patients are required to reproduce the Rey Complex Figure form memory. (B) An example of an impaired Rey Complex Figure from a patient with dementia. The patient took 11 min and 23 s to complete the task.

Semantic memory is generally examined using tests of word knowledge, knowledge of famous faces or less commonly world events. Aspects of autobiographical and procedural memory are seldom tested in routine clinical practice.

One criticism of common memory tasks is that they bear little relevance to functional, day-to-day memory ability. The Rivermead Behavioural Memory Test was developed as an ecological measure of memory in order to address this confound. It comprises tasks such as memory for faces and objects, appointments, messages and location of a hidden object.7


Although speech and language pathologists focus exclusively on language disturbances, neuropsychological assessments also examine integrity of verbal and written language skills. Aspects to consider include speech fluency, prosody (ie, the intonation of verbal output), rate of speech, errors in grammar and motor speech problems (eg, effortful, distorted speech). This is often accomplished by engaging the patient in unstructured conversation, or perhaps by asking them to describe a complex visual scene, such as the ‘Cookie theft’7 or ‘Beach’ scenes from the Western Aphasia Battery.14

Formal language tests include naming, picture-word matching, single word repetition and sentence repetition tasks. A common naming task is the Boston Naming Test, which consists of 60 line-drawn objects of increasing difficulty.7 ,10 Common word-picture matching tasks include the ‘Pyramid and Palm-trees’15 or ‘Camel and Cactus’16 tasks. The Sydney language battery (SYDBAT) is a recent test that combines a number of these tasks.17 The advantage of the SYDBAT is that it consists of four subtests including confrontation naming, word comprehension, semantic association and single word repetition using the same stimulus set. The profile of impairment across SYDBAT subtests is useful in distinguishing subtypes of primary progressive aphasia.17

Other aspects of language, such as production and interpretation of grammar, are less frequently assessed. The test of reception of grammar (TROG) is used to probe grammatical understanding.7 ,10 The TROG requires patients to interpret a number of short sentences, which become increasingly complex grammatically as the test proceeds. Ability to follow 1-step, 2-step or 3-step commands is another simple way to test grammatical understanding, especially if complex sentence structures are used (eg, ‘Hand me the pen after touching the paper’).

Visuospatial and constructional ability

Visuospatial function is assessed by measuring ability to interpret various types of visual information. Simple copy or drawing tasks, such as interlocking pentagons, wire cube, interlocking figure of eights or the reproduction of a clock face (figure 2), are widely used to assess constructional ability.7 These tasks are easily administered even with very impaired individuals. The clock face test requires the individual to draw a clock face from memory, with numbers and hands set at a specific time. Distortions or inability to draw the numbers within the clock face have been found to be sensitive, but non-specific, indicators of cognitive deficits. Another common, and more difficult task of visuoconstructive ability is the Rey-Osterrieth Complex Figure task,7 ,10 where the person is asked to copy a complex line drawing (figure 1). The accuracy of the copy (scored out of 36 points), time taken to copy the figure, and approach to the task are recorded. Obviously, pronounced motor dysfunction (eg, limb weakness, dystonia, parkinsonism or apraxia) may confound performance.

Figure 2

The clock face (A–C). Three examples of impaired clock faces from different individual patients with dementia. Patients are instructed to draw a clock face, including all the numbers and to set the time.

Basic visual processing can be assessed using line bisection or object cancellation tasks.6 ,7 Another common test, the Visual Object and Space Perception Battery (VOSP),7 involves the interpretation of information that varies in visual complexity (eg, dot counting, position discrimination and cube analysis). Cognitively intact individuals are expected to obtain perfect, or near-perfect, scores on VOSP subtests (>90% correct). As such, even a small decline in performance is suggestive of cognitive impairment.

Higher order visual processing deficits are also easily investigated. For example, a verbal description of a complex visual scene (eg, the ‘Cookie theft’) can be used to assess for simultanagnosia (the inability to interpret a complex visual scene). Similarly, visual agnosia—the impaired ability to recognise objects despite intact basic visual perception—may be suspected by observing how a patient approaches naming common objects. For example, the patient may have difficulty naming an object by sight, but produce the correct name when other sensory modalities (eg, touch) are employed. Prosopagnosia, is a special case of visual agnosia and refers to the difficulty or inability to recognise faces. This can be assessed using facial recognition tasks which require the matching of a target face with an array of similar looking faces presented from different angles.7

Executive function

Executive function comprises different cognitive skills including the ability to abstract, shift set, plan, organise and adapt behaviour to current circumstances,6 and is tested using a combination of approaches. For example, abstraction can be tested by concept formation or similarities tasks (eg, “what do ‘bicycle’ and ‘train’ have in common?”). The type of responses will inform the examiner as to the capacity of the patient to reason in abstract terms (eg, ‘They are both modes of transport’) or in concrete terms (eg, ‘They both have wheels’).

Set shifting can be tested using the Trail Making Test.7 ,9 Part A of the Trail Making Test requires the participant to draw lines between circles labelled with consecutive numbers (ie, ‘1’, ‘2’, ‘3’, etc). In Part B, the task is made more difficult by alternating consecutive numbers with consecutive letters (ie, ‘1’, ‘A’, ‘2’, ‘B’, ‘3’, ‘C’, etc). Individuals with executive impairment may take longer to complete these tasks, or make errors or both.

Fluency tasks, such as letter (eg, F, A, and S), or category (eg, animals, vegetables) fluency, also assess capacity to follow specific rules and to modify behaviour flexibly (ie, set shifting). Fluency tasks require the generation of as many words as possible in 1 min according to the rule set. Patients with executive dysfunction produce fewer correct responses on verbal fluency tasks than normal controls, although language proficiency and/or deficits need to be considered. Other, non-verbal equivalents (eg, design fluency) also exist, but are not commonly used.

Planning and organisation deficits might become apparent by observing the approach taken to complete a task. For example, a slow and disorganised approach to a copy task might suggest executive impairment. Disinhibition, and behaviour modulation more generally, is infrequently tested in clinical practice. One option is the Hayling Sentence Completion test, which requires suppression of a prepotent responses by completion of sentences with non-sensical endings.7 ,9

A number of different aspects of executive functioning including temporal judgement, set-shifting, planning and strategy can be tested formally using such tasks as the Behavioural Assessment of the Dysexecutive syndrome (BADS).9 More complex tasks such as the Iowa Gambling Test, or the Wisconsin Card Sorting test, are used infrequently, either in a research setting or sometimes clinically, to detect subtle executive dysfunction.9

Social cognition

Disturbances of social conduct and cognition in dementia, and their interactions with other cognitive domains, have been increasingly recognised over the past 20 years. Although still not widely used in clinical practice, tests of social cognition are gaining ground in the assessment of dementia syndromes.

Tests of social cognition investigate emotion recognition, disinhibition or theory of mind. Emotion recognition can be tested using the Ekman 60, where participants are required match photographs to one of six basic facial emotions (anger, disgust, fear, sadness, surprise and happiness).18 The Awareness of Social Inference Test (TASIT) uses videotaped vignettes to assess evaluation of emotions and social interference.18 Other tasks have been developed to test recognition of social faux pas19 and theory of mind. Theory of mind is the ability of an individual to imagine the inner thoughts of another person, an ability normally acquired throughout childhood and adolescence.20 This field is fast evolving, with particular interest in the interactions between social cognition, neuroeconomics and complex decision-making in individuals with dementia.

Behaviour and functional assessment

Behavioural disturbances and functional capacity are often measured together, through the use of carer-orientated questionnaires. For example, the occurrence and severity of behavioural disturbances are probed with instruments like the Neuropsychiatric Inventory (NPI) or the Cambridge Behavioural Inventory (CBI).16 ,21 The NPI is a structured carer interview that uses screening questions, as well as detailed subquestions, to probe neuropsychiatric symptoms like delusions, hallucinations, dysphoria, anxiety, agitation, euphoria, apathy, irritability, disinhibition, aberrant motor behaviour, night-time behaviour disturbances and changes in appetite and eating.22 Partly based on the NPI, the CBI examines a wider range of symptoms21 and may be administered in written form, either before or during the clinical assessment. Responses to the NPI and CBI are graded in terms of frequency and severity and can be converted into scores to indicate the level of behavioural disturbance. In addition, the CBI includes several items designed to measure functional abilities (eg, use of electrical appliances, handling money, self-grooming, etc).

A number of tools have been developed to probe different aspects of behavioural disturbance more specifically. For example, the Geriatric Depression Scale, composed of 30 ‘yes’/‘no’ questions, was developed to explore affective symptoms in older adults, both healthy and demented.23 Other measures examine apathy,24 stereotyped behaviours25 and changes in eating and appetite26 through the use of structured questions, posed to carers or patients themselves. The BADS is a test battery, which includes six components designed to probe rule set shifting, problem solving, planning and judgement. It was developed to assess executive dysfunction in a range of everyday activities, and thereby provide a measure of functional performance.7 Its usefulness is mitigated by its length of administration and is therefore rarely administered in its entirety to patients with dementia.

Several instruments have been developed to specifically assess functional impairment in patients with FTD. For example, the Frontal Behavioural Inventory (FBI) is a 24-item face-to-face caregiver interview designed to document the types of behavioural disturbances commonly seen in the behavioural variant of FTD.27 The Functional Rating Scale (FRS) is another commonly used assessment that includes a functional assessment. The FRS examines performance in: behaviour, outings and shopping, household chores and telephone use, finances, medications, meal preparation and eating and self-care/mobility.28 Importantly, measures of functional impairment such as the FBI or FRS may detect behavioural changes in patients with FTD who perform well on cognitive testing initially.

The progression of functional impairment in dementia can be measured using tools such as the Clinical Dementia Rating Scale (CDR), which was specifically developed for use in Alzheimer's disease.29 The CDR incorporates patient and informant derived information to grade performance in: memory, orientation, judgement and problem solving, community affairs, home and hobbies and personal care. Each section is rated individually then combined to produce a global CDR. A CDR score rates dementia as 0 (no dementia), 0.5 (questionable), 1 (mild), 2 (moderate) and 3 (severe).29 The CDR has been modified for use in FTD (CDR-FTD) to include behavioural and language sections.30 By comparing current with previous patient ability, the CDR may be less influenced by education, language and culture than traditional neuropsychological testing.29

Diagnosis of dementia and patterns of impairment

Reliable in vivo clinical biomarkers of dementia do not currently exist, despite significant advances in neuroimaging, pathology and genetics in recent years. As such, the diagnosis of dementia is made primarily on clinical grounds. History, examination, neuroimaging and performance on neuropsychological assessment all contribute towards a diagnosis. Numerous clinical diagnostic criteria for dementia exist and have been revised over the years. These include broad indices, such as the DSM-5, as well as those focusing on specific entities, including Alzheimer's disease,31 FTD,32 primary progressive aphasias.33 The criteria for the diagnosis of Alzheimer's disease, proposed by McKhann et al,31 emphasise the need for supporting evidence from biomarkers (eg, imaging, blood, cerebrospinal fluid), which may improve diagnostic accuracy in early or even prodromal cases.34

Clinical diagnostic criteria for the main dementia syndromes are an important reference but, like criteria for other clinical diagnoses, may be difficult to implement in individual cases. A discussion on their merits and limitations is, however, beyond the remit of the present review. Rather, we provide a brief outline of the various neuropsychological profiles observed in the common dementia syndromes, each of which might be associated with a different pathological process. This discussion is not intended to be comprehensive, but will provide a foundation for the general neurologist or psychiatrist and illustrate the clinical utility of neuropsychological assessment (table 1).

Table 1

Patterns of cognitive impairment across the range of dementia syndromes

Patients with the typical, or amnestic, presentation of Alzheimer’s disease usually exhibit variable concentration, marked deficits in episodic memory (verbal and visual) and visuospatial ability, but relatively intact language, behaviour and social cognition, at least in the initial stages.35 This pattern of cognitive deficits predicts underlying Alzheimer's pathology with >85% accuracy. In a small proportion of cases, Alzheimer's disease may present in an atypical fashion with predominant deficits in language (ie, logopenic progressive aphasia (LPA)), executive function, motor function (ie, corticobasal syndrome) or vision (ie, posterior cortical atrophy).31 ,36 The role of specific memory testing in patients with mild cognitive impairment, often considered a prodromal form of Alzheimer's disease remains controversial. Older individuals often present to clinic with concerns about failing memory and may even demonstrate subtle memory disturbances on cognitive screening. Importantly, only a proportion of these individuals will progress to develop Alzheimer's disease or another dementia, and it remains difficult to predict with certainty the outcome in any individual patient. The pattern of deficits on memory tasks may help, whereby impaired encoding of information together with impaired recall and recognition appears to associate with the early hippocampal pathology seen in Alzheimer's disease. In contrast, impaired recall but relatively preserved recognition is suggestive of an alternative diagnosis.11 ,37 ,38 Unfortunately, such memory deficit profiles may not be as specific as initially thought for the early diagnosis of Alzheimer's disease.39

At a clinical level, vascular dementia can be difficult to distinguish from Alzheimer's disease; the two conditions often coexist pathologically, particularly in very old individuals. A history of multiple strokes and focal neurological signs on examination is suggestive of vascular dementia, as is marked white matter signal change on MRI.35 Neuropsychologically, patients with vascular dementia demonstrate variable memory and executive impairment, like patients with Alzheimer's disease, although subtle differences may be detectable.35 For example, patients with vascular dementia may show relatively spared ability to encode new information but impaired retrieval of information. In contrast, patients with Alzheimer's disease demonstrate impairment in encoding and retrieval of new information. In addition, patients with vascular dementia may demonstrate impaired cognitive processing speed, thought to reflect the subcortical burden of pathology.6 ,35

Another important dementia syndrome is dementia with Lewy bodies (DLB), which exists clinically in a disease continuum with idiopathic Parkinson's disease. DLB is often associated with a mixed pathology, with features of both Alzheimer's and Parkinson's diseases. Clinically, DLB is characterised by impaired and fluctuant concentration throughout the day and visual hallucinations,35 although such fluctuation may be difficult to assess neuropsychologically. Importantly, cognitive dysfunction in DLB often responds to treatment with anticholinesterase inhibitors.40

Atypical parkinsonian disorders are associated with variable cognitive deficits.35 ,41 Progressive supranuclear palsy is often associated with a poverty of verbal output, as well as subtle executive dysfunction.41 Corticobasal syndrome, on the other hand, is associated with a wide range of cognitive deficits including memory, language and visuospatial impairments.41 Perhaps the most consistent abnormality identified in corticobasal syndrome is visuospatial impairment, which is often closely related to limb apraxia.42 In addition, patients with corticobasal syndrome may develop a non-fluent speech disorder, which can be difficult to distinguish from progressive non-fluent aphasia (PNFA) in the early stages.

FTD is the most common cause of dementia in younger patients after Alzheimer's disease, and presents with a range of behavioural and cognitive deficits.43 Roughly half of patients with FTD present with the so-called behavioural variant (bvFTD), which presents with disinhibition, apathy and marked deficits in interpersonal conduct, while cognitive performance may appear intact initially, although detailed testing usually demonstrates evidence of executive dysfunction and disinhibition.44 Other deficits in language and memory may become more prominent as the disease progresses.45

When patients present with marked and progressive behavioural disturbance, as well as frontotemporal atrophy on MRI or hypometabolism on functional imaging, the diagnosis of bvFTD is relatively straightforward. Patients with atypical Alzheimer's disease (eg, the frontal variant) can present with apathy and executive deficits, despite relatively preserved memory. Alternatively, some patients with a clinical presentation identical to bvFTD may have no cortical atrophy on MRI or perform normally on neuropsychological testing. Whether such patients have FTD, or a neuropsychiatric mimic of FTD, remains to be resolved. Finally, a atypical or slowly progressive bvFTD phenotype has been described in carriers of the C9ORF72 repeat expansion, now recognised as an important cause of familial and apparently sporadic FTD.46 ,47

Almost half of patients with FTD present with striking, but differing, patterns of language impairment.17 ,35 For example, the semantic dementia (SD) phenotype is characterised by severe deficits in word and object knowledge, surface dyslexia (ie, difficulty reading irregular words) and behavioural disturbances, with preserved single word and sentence repetition.48 Once encountered, SD is easily recognised, but may be missed by the uninitiated. Pathologically, SD is tightly associated with underlying TDP-43 positive intraneuronal inclusions,49 so recognition of the syndrome is very helpful diagnostically. In contrast, non-fluent speech characterised by motor speech errors (called apraxia of speech) and grammatical errors or agrammatism is typical of PNFA, which is more commonly associated with underlying τ-positive pathology.50

Another non-fluent speech syndrome called LPA has been described.33 LPA is characterised by marked word finding difficulties and impaired sentence repetition, despite relatively preserved single word repetition and only minor phonological or grammatical errors in spontaneous speech.51 While it resembles PNFA, LPA is almost always associated with underlying Alzheimer's pathology.52

In clinical practice, the distinction of different aphasia syndromes can be difficult, especially in non-fluent cases and reliance on clinical neuropsychological features alone can be misleading. For example, syntactic or grammatical errors or may occur in PNFA and LPA, as can disturbances of word and sentence repetition. Even for experienced cognitive neurologists, it may be difficult to confidently classify some non-fluent cases. Similarly, impaired naming and word-finding difficulties can occur in LPA and SD, even though the two disorders are quite distinct clinically and neuroradiologically. Despite the publication of diagnostic criteria based on clinical features,33 more specific markers of underlying pathology are required to better define clinicopathological correlations.

Final remarks

Like many other clinical tools, a detailed neuropsychological assessment can play a central role in the diagnosis and grading of cognitive deficits in dementia syndromes, provided the benefits and limitations are clearly understood by referring clinicians (box 2). An understanding of cognitive (eg, attention and orientation, memory, language, visuospatial function and executive function) and behavioural domains is central to the interpretation of neuropsychological reports. It is important to understand how ‘normal’ performance on cognitive tests is defined, whether this is relative to an estimate of premorbid function, or to a set of population norms. Finally, the pattern of deficits can be helpful in defining clinical phenotypes, which can sometimes accurately predict the underlying molecular pathology. Such clinicopathological correlation will become invaluable as disease-modifying treatments for dementia are developed and implemented.

Box 2

Key points in the use of clinical neuropsychology for the assessment of dementia

  • Neuropsychological testing is helpful for diagnosis and management of dementia.

  • Patients need to be able to perform at their best for the testing to be meaningful; medical illnesses, medications, anxiety/depression and testing in the patient’s second language can all confound the results.

  • Testing may not be reliable in patients with severe deficits—especially language deficits.

  • All main cognitive domains should be considered.

  • Cognitive screening tasks (eg, Montreal Cognitive Assessment (MoCA) or ACE-III) are helpful in the diagnosis of dementia in the clinic.

  • The pattern of cognitive deficits is helpful in establishing a syndromic diagnosis, which may help define the molecular pathology.

  • The assessment of behaviour and social cognition is an important part of the neuropsychological assessment.


The authors wish to thank Professor John R Hodges for helpful comments in the preparation of this manuscript. This work was supported in part by funding to Forefront, a collaborative research group dedicated to the study of frontotemporal dementia and motor neuron disease, from the National Health and Medical Research Council of Australia (NHMRC) programme grant (APP1037746) and the Australian Research Council Centre of Excellence in Cognition and its Disorders (CE110001021). JRB is supported by an NHMRC Early Career Fellowship (APP1072451) and OP by an NHMRC Career Development Fellowship (APP1022684).


View Abstract


  • Contributors JRB and OP both conceived the idea for this review article and were involved in drafting and editing the manuscript.

  • Competing interests None.

  • Patient consent Obtained.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.