OUP user menu

Cognitive and behavioural deficits associated with the orbitomedial prefrontal cortex in amyotrophic lateral sclerosis

Sandra L. Meier, Alison J. Charleston, Lynette J. Tippett
DOI: http://dx.doi.org/10.1093/brain/awq254 3444-3457 First published online: 1 October 2010


Amyotrophic lateral sclerosis, a progressive disease affecting motor neurons, may variably affect cognition and behaviour. We tested the hypothesis that functions associated with orbitomedial prefrontal cortex are affected by evaluating the behavioural and cognitive performance of 18 participants with amyotrophic lateral sclerosis without dementia and 18 healthy, matched controls. We measured Theory of Mind (Faux Pas Task), emotional prosody recognition (Aprosodia Battery), reversal of behaviour in response to changes in reward (Probabilistic Reversal Learning Task), decision making without risk (Holiday Apartment Task) and aberrant behaviour (Neuropsychiatric Inventory). We also assessed dorsolateral prefrontal function, using verbal and written fluency and planning (One-touch Stockings of Cambridge), to determine whether impairments in tasks sensitive to these two prefrontal regions co-occur. The patient group was significantly impaired at identifying social faux pas, recognizing emotions and decision-making, indicating mild, but consistent impairment on most measures sensitive to orbitomedial prefrontal cortex. Significant levels of aberrant behaviour were present in 50% of patients. Patients were also impaired on verbal fluency and planning. Individual subject analyses involved computing classical dissociations between tasks sensitive to different prefrontal regions. These revealed heterogeneous patterns of impaired and spared cognitive abilities: 33% of participants had classical dissociations involving orbitomedial prefrontal tasks, 17% had classical dissociations involving dorsolateral prefrontal tasks, 22% had classical dissociations between tasks of both regions, and 28% had no classical dissociations. These data indicate subtle changes in behaviour, emotional processing, decision-making and altered social awareness, associated with orbitomedial prefrontal cortex, may be present in a significant proportion of individuals with amyotrophic lateral sclerosis without dementia, some with no signs of dysfunction in tasks sensitive to other regions of prefrontal cortex. This demonstration of variability in cognitive integrity supports previous research indicating amyotrophic lateral sclerosis is a heterogeneous disease.

  • amyotrophic lateral sclerosis
  • orbitomedial prefrontal cortex
  • social cognition
  • decision-making
  • emotional-processing


Amyotrophic lateral sclerosis (ALS), the most common form of Motor Neuron Disease, is a devastating neurodegenerative condition characterized by progressive paralysis due to loss of both upper and lower motor neurons. Onset can occur in any body region, and death is typically due to respiratory failure. ALS has traditionally been viewed as affecting motor neurons only, but recent research has demonstrated that it can cause variable alterations in cognition and behaviour. Recent studies report rates of mild to moderate cognitive impairment from 32 to 36% (Massman et al., 1996; Ringholz et al., 2005), with dementia occurring in an additional 19–23% of cases (Lomen-Hoerth et al., 2003; Ringholz et al., 2005). Cognitive dysfunction is not an inevitable part of the disease process, however, as cases with extreme levels of disability may still maintain apparently normal cognitive function (Fuchino et al., 2008; Lakerveld et al., 2008). Variability in cognitive integrity, together with the variability of clinical presentation, genetic causation and pathology, indicate ALS is a heterogeneous disease (Rosen et al., 1993; Ince et al., 1998; Munch et al., 2004).

The most frequently reported cognitive deficits in ALS without dementia involve attention and executive functions mediated primarily by dorsolateral regions of prefrontal cortex (Massman et al., 1996; Abrahams et al., 2000; Lomen-Hoerth et al., 2003; Phukan et al., 2007; Pinkhardt et al., 2008). Deficits in verbal fluency are the most common finding, even when methods are used to control for the confounding effects of motor slowing or disability (Abrahams et al., 1996, 2000). Dysfunction of memory and language systems has also been reported (Massman et al., 1996; Strong et al., 1999; Moretti et al., 2002; Bak and Hodges, 2004; Ringholz et al., 2005). Visuoconstructional, visuoperceptual and visuospatial abilities may also be affected although less frequently (Massman et al., 1996; Ringholz et al., 2005; Flaherty-Craig et al., 2006).

The common finding of poor performance on tasks associated with dorsolateral prefrontal cortex in part reflects the widespread use of conventional measures of prefrontal function, which typically tax these regions (Stuss and Levine, 2002). In contrast, little research has focused upon functions associated with orbital and medial prefrontal regions. The neuroanatomical term ‘orbital and medial prefrontal cortex’ is used to refer to areas defined by Öngür et al. (2003) following histological and immunohistochemical analyses of human prefrontal cortex. Orbital regions include subdivisions of medial orbital [Brodmann area (BA) 11, 13, 14] and lateral orbital (BA 47/12) surfaces; the frontal pole comprising subdivisions of BA 10 extending into the medial wall; and medial regions include anterior cingulate cortex (subdivisions of BA 25, 24 and 32). Individuals with dysfunction limited to these regions typically do well on standard tests of prefrontal cortex function, yet have profound and debilitating changes in behaviour and social functioning (Eslinger and Damasio, 1985; Gregory et al., 2002). Orbitomedial prefrontal cortex plays a key, but not exclusive, role in multimodal emotional processing (Phan et al., 2002; Kawasaki et al., 2005); maintaining reward-value associations (O'Doherty et al., 2001); altering behaviour in response to environmental reinforcers, as in reversal learning (Fellows and Farah, 2003; Kringelback and Rolls, 2003; Hornak et al., 2004); decision-making (Rogers et al., 1999; Clark and Manes, 2004; Fellows, 2006), and social cognition, e.g. affective Theory of Mind and empathy (Eslinger and Damasio, 1985; Eslinger, 1998; Stone et al., 1998; Shamay-Tsoory et al., 2005).

Converging lines of evidence suggest dysfunction of orbitomedial prefrontal cortex may occur in ALS. The most common dementia diagnosis accompanying ALS is the frontal variant of frontotemporal lobar dementia (FTLD) (Lomen-Hoerth et al., 2003; but see Ringholz et al., 2005), a disease in which the earliest and most severe degeneration occurs in orbitomedial prefrontal cortex (Rosen et al., 2002; Broe et al., 2003). Interestingly, motor neuron disease-type ubiquitinated inclusions are the most common pathology type in FTLD and may present either with or without co-morbid motor neuron disease (Lipton et al., 2004). Recently TDP-43 was identified as the protein present in both FTLD, and the majority of ALS, ubiquitinated lesions (Neumann et al., 2006; Maekawa et al., 2009). Geser et al. (2008) found that in 31 ALS cases without dementia, abnormal TDP-43 was widely distributed in cortical regions, including the orbital prefrontal region (albeit at a lower rate than was found in dorsolateral prefrontal cortex).

Preliminary clinical and neuropsychological findings also suggest dysfunction of orbitomedial prefrontal cortex may be present in ALS. Reports of marital separation, dysfunctional relationships and financial stresses prior to an ALS diagnosis, suggest early alterations in personality, behaviour and social cognition may be common (Lomen-Hoerth, 2004). Consistent with this, aberrant behaviour in ALS has been reported using carer interviews such as the Neuropsychiatric Inventory (Strong et al., 1999; Lomen-Hoerth et al., 2003; Murphy et al., 2007), Frontal Systems Behavioural Scale (Grossman et al., 2007) and carer questionnaires designed to assess FTLD (Gibbons et al., 2008). Strong et al. noted high levels of individual variability; some patients showed no behavioural change whilst others developed significant agitation, anxiety, delusions, disinhibition, apathy and irritability/lability. Whilst not all behaviours assessed by such measures are indicative of orbitomedial prefrontal cortex dysfunction, several have been associated with early orbital degeneration in frontal variant FTLD such as apathy, irritability and disinhibition (Broe et al., 2003; Wittenberg et al., 2008).

In ALS the most commonly studied cognitive function relevant to orbitomedial prefrontal cortex is emotional processing. This brain region is central to the multi-sensory integrated processing of both emotional expressions and their reward-value associations (Wildgruber et al., 2004; Kawasaki et al., 2005; Kober et al., 2008), regardless of emotion type (Keane et al., 2002; Phan et al., 2002). Results from studies in ALS are mixed. Papps et al. (2005) found participants with ALS were unimpaired at recognizing facial expressions and judging the approachability of emotional faces but they failed to demonstrate the normal enhancement of recognition memory for emotional words. In contrast, Zimmerman et al. (2007) found participants with ALS were impaired at recognizing facial emotional expressions and Schmolck et al. (2007) reported significantly poorer judgement of the approachability of emotional faces.

Theory of Mind abilities have also been investigated in ALS (Gibbons et al., 2007). This is the ability to infer other people’s mental states, including their thoughts and feelings (Baron-Cohen et al., 1985), enabling us to respond not only to the behaviour of others, but also to what we theorize are their intentions, beliefs and desires. Individuals with lesions to orbitomedial prefrontal cortex are impaired on Theory of Mind tasks, particularly affective Theory of Mind, i.e. knowledge about the emotional states of others (Stone et al., 1998; Shamay-Tsoory et al., 2005, 2007) as are individuals in the early stages of frontal variant FTLD (Gregory et al., 2002; Torralva et al., 2007). Neuroimaging studies of normal subjects provide converging evidence for the critical role of medial and orbitolateral prefrontal cortex in Theory of Mind abilities (Gallagher and Frith, 2003; Mitchell et al., 2006; Völlm et al., 2006). Gibbons et al. (2007) found ALS participants were not differentially impaired at inferring mental states within stories or cartoons, and concluded that any difficulties present were likely to reflect general executive dysfunction.

In summary, current evidence of cognitive deficits associated with dysfunction of orbitomedial prefrontal cortex in ALS is mixed. To our knowledge, however, no studies to date have systematically investigated performance of ALS individuals without dementia on a range of cognitive and behavioural tasks in which orbitomedial prefrontal cortex plays a key role. Establishing the likelihood of dysfunction of orbitomedial prefrontal cortex is important as the behavioural, cognitive and personality changes associated with even subtle impairments of this region can have significant implications for quality of life not only for the individual but also carers and other family members (Cullen et al., 1997; Riedijk et al., 2006; Davis and Tremont, 2007). Additionally, such deficits can complicate treatment and care (e.g. Lomen-Hoerth et al., 2003; Olney et al., 2005).

In the following study, we carried out a systematic evaluation of behaviour and performance on cognitive tasks sensitive to regions within orbitomedial prefrontal cortex in a group of individuals diagnosed with ALS (without dementia) and a closely-matched control sample. Selection of measures was critical. We included published measures that assessed emotional processing, reversal learning, decision-making and social cognition, with task selection guided by findings from a significant lesion and neuroimaging literature. Not surprisingly, tasks that assess these complex functions also recruit distributed neural networks that include subcortical and posterior cortical regions (Rudebeck et al., 2008), and sometimes processes linked with dorsolateral prefrontal cortex, such as working memory (e.g. Clark and Manes, 2004; Fellows and Farah, 2005). Where possible our tasks include conditions to control for confounding variables, but even so, it must be acknowledged that no cognitive task yet designed is a pure assay of orbitomedial prefrontal cortex functioning.

The Neuropsychiatric Inventory was selected to measure the presence and severity of aberrant behaviour. As orbitomedial prefrontal cortex has been implicated in the appraisal of visually and vocally-expressed emotional expressions, regardless of emotion type (Phan et al., 2002; Kawasaki et al., 2005; Schirmer and Kotz, 2006; Kober et al., 2008), we selected a well-published task involving prosody recognition to test emotional processing (Ross et al., 1997). To measure the ability to alter behaviour in response to changes in stimulus-reward associations, we included a reversal learning task based on the task of Fellows and Farah (2003), with the addition of a probability factor designed to move performance from ceiling. Social cognition was assessed using the Faux Pas Task (Stone et al., 1998), a Theory of Mind task that involves both representing the knowledge or beliefs of others (cognitive Theory of Mind) and appreciating the emotional state of others (affective Theory of Mind).

Decision-making is a complex process with many component processes, making task selection difficult. Although performance on gambling tasks that require decision-making in the presence of uncertainty and risk has been linked with the integrity of orbitomedial prefrontal cortex (e.g. Bechara et al., 1994; Rogers et al., 1999), other findings suggest poor performance on these tasks may not be specific to dysfunction of orbitomedial prefrontal cortex. Fellows and Farah (2005) found that patients with damage only to dorsolateral prefrontal cortex were also impaired, while Manes et al. (2002) found only patients with large lesions involving dorsal and ventral prefrontal cortex were impaired across gambling tasks, while patients with lesions restricted to orbitofrontal cortex were unimpaired.

A different approach to assessment of decision-making utilizes multi-attribute decision-making tasks. These entail choosing between known alternatives that differ across multiple dimensions, (e.g. choosing which of three schools to send your child). Fellows (2006) adapted analyses of the strategic information-acquisition process developed in studies of decision-making in healthy participants (e.g. Payne et al., 1976), to examine multi-attribute decision-making in individuals with focal frontal lobe lesions. Their task required selecting between rental apartments with complete availability of information. Importantly, only patients with focal damage to ventromedial prefrontal cortex searched apartment-by-apartment; healthy control subjects and individuals with dorsal and/or lateral prefrontal damage used an attribute-based strategy. This altered method of information acquisition was thought to reflect difficulty determining the relative value of alternatives that are compared during decision-making (Fellows, 2007; Fellows and Farah, 2007). In this study we adapted the rental apartment task used by Fellows (2007) so that participants had to select an apartment for an upcoming holiday, an ecologically valid decision task that avoids the confounding influences of ambiguity, risk and working memory.

A secondary aim of this study was to use verbal and written lexical and category fluency and a planning task (One-touch Stockings of Cambridge) to assess functions in which dorsolateral prefrontal cortex plays a key role. Although the number of tasks provides a less extensive evaluation of functions linked with this brain region, the most consistently reported deficit in ALS involves verbal fluency (Ludolph et al., 1992; Abrahams et al., 1996, 2000; Massman et al., 1996). Fluency impairments in ALS have been most strongly (but not exclusively) linked with executive dysfunction and functional abnormalities in dorsolateral prefrontal cortex (Abrahams et al., 2000, 2004; Lomen-Hoerth et al., 2003).

Overall we predicted the ALS group would perform worse on tasks linked with orbitomedial prefrontal cortex than the control group, but not on control versions of tasks that do not recruit this region. Similarly we expected the ALS group would perform less well on tasks sensitive to executive dysfunction linked with dorsolateral prefrontal cortex. We also predicted, however, that there would be heterogeneity in the patterns of impaired and spared cognitive performance on tasks sensitive to these two broad regions of prefrontal cortex in individual subjects.

Materials and methods


ALS participants

The ALS group comprised 18 participants, 12 male and 6 female, all with a neurological diagnosis of either probable or definite ALS (Brooks et al., 2000). ALS participants were recruited during the period February 2007 to July 2008 from patients attending an outpatient Motor Neuron Disease clinic at Auckland City Hospital. Inclusion criteria were ability to either speak or write with ease, proficiency with the English language, no diagnosis of dementia, and either a forced vital capacity measure above 50 or no clinical evidence of nocturnal hypoventilation. All patients who met inclusion criteria were invited to participate. Participants were screened via a clinical interview (that included family members), by the motor neuron disease neurologist (A.C.) experienced in dementia diagnosis. If the clinical history or interview revealed concerns about possible dementia, the Addenbrooks Cognitive Examination-Revised test was administered. No participant met the consensus criteria for a frontotemporal dementia (Neary et al., 1998) or other dementia. Three participants did not have forced vital capacity measures above 50: one participant with a score of 46, had respiratory onset and used a bi-level positive airway pressure machine nightly resulting in good ventilation; two had mouth muscle weakness preventing accurate measurement of forced vital capacity, however there was no clinical evidence of nocturnal hypoventilation. Participants were also excluded if they had uncontrolled hypertension or psychiatric illness, a history of alcohol abuse, neurological insult or disease other than ALS, or a score on the Hospital Anxiety and Depression Scale (Zigmond and Snaith, 1983) exceeding the mild level. Twenty-four patients were invited to participate in the study: three declined, one was excluded due to a psychiatric condition, and two were unable to cope with testing sessions, leaving 18 in the study. A carer or close family member of each participant with ALS was also invited to take part in the study as an informant, and all agreed to participate. Sixteen were spouses who had daily contact with the ALS participant, the other two (daughter and granddaughter) had weekly contact.

Control participants

The control group comprised 18 participants, 11 males and 7 females, and included other family members and friends of ALS participants, and other residents from the greater Auckland area. The same exclusion screening criteria were applied. Of the 31 people willing to participate, two were excluded due to level of alcohol consumption, two due to moderate levels of anxiety, one due to history of significant head injury, and one who was unable to attend the second interview and test session. One participant withdrew for personal reasons, and one withdrew as testing triggered traumatic memories. Five control participants were used to pilot test procedures prior to formal testing of the main body of participants and, due to subsequent modifications of test procedures, their data were excluded from analyses, leaving 18 participants in this group.

Participant demographics and disease characteristics are shown in Table 1. Participants in the two groups were closely matched with no significant difference in age, t(34) = 0.27, P = 0.79, years of education t(34) = 0.19, P = 0.85, Hospital Anxiety and Depression Scale scores t(34) = 0.19, P = 0.85 and t(34) = 0.63, P = 0.53, respectively, gender X2(1) = 0.12, P = 0.73, or ethnicity X2(2) = 0.00, P = 1.00.

View this table:
Table 1

Demographic characteristics of the ALS and control groups

ALS (n = 18) mean (SD) rangeControls (n = 18) mean (SD) range
Age (years)64.50 (11.48)63.56 (9.45)
Education (years)13.22 (2.83)13.06 (2.37)
Gender M/F12/611/7
HADS anxiety4.72 (2.52)4.89 (2.85)
HADS depression3.06 (2.62)2.56 (2.09)
Forced vital capacity78.63 (17.37)
Time since onset (months)34.56 (24.75)
ALS Functional Rating Scale—Revised35.78 (5.23)
  • a Ethnic groups = NZ Pakeha/NZ Maori/White British/White South African/White Australian; HADS = Hospital Anxiety and Depression Scale.


Tasks sensitive to orbitomedial prefrontal cortex function

Neuropsychiatric Inventory

The Neuropsychiatric Inventory is a semi-structured interview conducted with informants to assess twelve types of aberrant behaviour commonly associated with neurodegenerative disorders: delusions, hallucinations, depression, anxiety, irritability/lability, aggression/agitation, euphoria, apathy, disinhibition, aberrant motor behaviour, sleep disturbance and eating/appetite change (Cummings et al., 1994). An additional subscale was added to assess stereotypical behaviour often seen in frontal variant FTLD (Nyatsanza et al., 2003). Interviews were conducted using protocols described by Cummings et al. (1994) with two alterations. First, respondents were instructed to identify behaviours present during the past month that had developed either since diagnosis, or within four years prior to diagnosis. Second, care was taken to exclude changes in behaviour more likely due to physical disability or a culturally-appropriate reaction to their diagnosis (e.g. comments about death). Each behaviour was rated for severity (1–3) and frequency (1–4); the product of these ratings was the score for each behaviour (maximum score of 12). Maximum total Neuropsychiatric Inventory score was 156.

Faux Pas Test

The Faux Pas Test (Stone et al., 1998; Gregory et al., 2002) requires the recognition of a faux pas in a variety of social situations. A faux pas was defined as when a person unintentionally makes a comment they shouldn’t have made, not knowing or realizing that they shouldn’t be making it because it could hurt or upset another person. Participants were read 20 brief stories (with written copies in front of them), 10 containing a faux pas (Theory of Mind condition) and 10 control stories in which the faux pas was removed. Stories from the two conditions were randomized and the order counterbalanced across participants. Minor modifications were made to culturally-based content (e.g. a school ‘dinner lady’ was changed to ‘teacher’). Test administration was based on the procedure of Gregory et al. (2002). After each story the participant was asked whether anyone said something they should not have said. If participants indicated a faux pas was present, clarifying questions were asked—who had said something they should not have, why they should not have said it and why the participant thought they had done so. Finally a general non-Theory of Mind question was asked for all 20 stories to assess basic story comprehension.

Probabilistic Reversal Learning Test

This task involved a computerized card game with fictitious monetary reward. Participants chose cards one at a time from two decks of different colours; choosing from one deck resulted in monetary gain (the winning deck) and choosing from the other resulted in monetary loss. A ‘reversal’ of the winning deck occurred after 13 consecutive selections were made from the current winning deck. The built-in probability factor (6:1 correct:incorrect feedback) meant that occasionally, picking from the winning deck resulted in a loss and vice versa. An initial learning phase was followed by 75 further trials allowing up to five reversals to be performed. If the learning phase was unsuccessful (failure to pick consistently from the winning deck after 50 trials) the task was terminated. Participants were told about the ‘winning’ and ‘losing’ nature of the decks and that their goal was simply to accumulate as much money as possible. They were also told that the ‘winning’ deck might change, that when this occurred they should change to the ‘new’ winning deck to keep accumulating money and that trying to guess when a change would occur was a risky strategy. Outcome measures were the average number of learning trials required, average number of incorrect trials after deck reversal and total number of reversals achieved.

Holiday Apartment Task

The Holiday Apartment Task, a computerized decision-making test adapted from the rental apartment task used by Fellows (2006), assesses strategic processes involved in multiple-attribute decision-making without ambiguity or risk. Participants were instructed to imagine they were going on a 4-week holiday, and that they should choose the apartment that would best suit them from a range of apartments with different attributes (all of which catered for physical disability). Information was presented on a computer touch-screen, in a grid with apartment letters as column headings and apartment attributes as row headings. Information was initially masked by white rectangles, which, once touched, revealed the information for that particular apartment and attribute. This remained visible until a decision was made. Participants were advised that there was no correct answer and that they did not need to examine all the information before making their decision, but should view as much information as needed to make their decision as efficiently as possible. They were encouraged to think aloud and their comments were recorded to focus their attention on the decision-making process.

A practice trial was completed that required deciding which of two cars to purchase, each with two attributes. This was followed by three trials of apartment decision-making of increasing complexity; two apartments with four attributes (2 × 4 grid), four apartments with six attributes (4 × 6), and six apartments with seven attributes (6 × 7). Attribute values ranged from negative to positive, for example the attribute ‘views’ could have values of ‘limited’, ‘good’ and ‘superb’. No apartment was obviously better than the others.

The main outcome measure was the search index, or pattern of information acquisition, in each trial. This was formulated as: (movements across – movements down)/(movements across + movements down). The result is a variable ranging from −1 (all movements were down, searching by apartment alternatives) to +1 (all movements were across, searching by attribute). The search index, regardless of how much information is acquired, is an indicator of the primary direction of the search strategy, i.e. primarily attribute-based (horizontal search looking at an attribute across apartments), or primarily apartment-based (vertical search looking at an apartment for all or most attributes). Additional outcome measures were time to reach a decision and amount of information viewed.

Aprosodia Battery

Processing of affective prosody was tested using the comprehension and discrimination tasks of the Aprosodia Battery (described in detail in Ross et al., 1997 and Testa et al., 2001). Stimuli were presented using computerized audio recordings and comprised six emotions—happy, sad, disinterested, angry, surprised and neutral.

Comprehension task: This comprised three stimulus sets: prosody combined with a semantically-neutral sentence ‘I am going to the other movies’; prosody combined with repetition of a syllable ‘ba ba ba ba’; and prosody combined with a single long syllable ‘aaaahhhhh’. Each stimulus set contained four trials per emotion, two with emphasis on the first part of the expression and two with emphasis on the second part, giving 24 trials per set, with randomized presentation. Participants were given a set of six faces, each depicting one of the emotions, with the expression name printed beneath. They were asked to indicate which of the emotions in front of them best matched the emotional expression in the recording, with accuracy being the main outcome measure.

Discrimination task: Sentences from the comprehension subtest were filtered to remove phonetic information whilst preserving the pitch, intonation and stress patterns that convey prosody, and then randomly combined into pairs. Participants were instructed to indicate whether the emotional expression in the two sentences was the same or different.

Tasks sensitive to dorsolateral prefrontal cortex function

Verbal fluency

Oral and written verbal fluency tasks followed the method recommended by Abrahams et al. (2000) to minimize the effects of physical disability. Oral tasks comprised letter fluency (P, R, W) and category fluency (animals), each trial with 1 min to generate as many items as possible. Written tasks comprised letter fluency (letter S for 5 min, four letter C words for 4 min) and semantic fluency (colours for 2 min).

A control condition involved reading aloud, or copying, the exact words produced in each generation task. A fluency index was calculated to represent mean time to think of each word: time allowed for the generation condition – time taken in the control condition/total number of words produced. Fluency indices were calculated for each fluency task.

One-touch Stockings of Cambridge

The One-touch Stockings of Cambridge (Owen et al., 1995) is a modified computerized touch-screen version of the standard Tower of London task (Shallice, 1982), which is a spatial planning task that assesses the ability to plan a series of sequential moves of variable complexity. The One-touch Stockings of Cambridge places fewer demands on motor performance than the standard Stockings of Cambridge, but still reliably activates the dorsolateral prefrontal cortex (Rasmussen et al., 2006; Unterrainer and Owen, 2006). In both the Stockings of Cambridge and the One-touch Stockings of Cambridge tasks, two sets of three stockings are presented on a computer screen, each set containing three coloured balls. The goal is to rearrange the balls in the bottom display so that the position of each ball matches the pattern of balls in the top half of the screen. In the One-touch Stockings of Cambridge, instead of moving the balls manually, participants calculate the minimum number of moves required to achieve the solution.

Participants were initially trained on the standard version of this task. Once task rules were learnt, the One-touch version of the task was administered (see Owen et al., 1995 for details of training and administration). Trials required one to six moves to complete. Participants were told their primary goal was to get the correct solution the first time. Following an error they were to re-think their solution and make another response. There were 10 ‘easy’ trials in which the patterns could be matched in one to three moves, and eight ‘difficult’ trials requiring four to six moves. Outcome measures were proportion of perfect solutions, mean attempts required to obtain a correct solution and mean time taken.


The study was approved by the Northern Y Regional Ethics Committee of the New Zealand Ministry of Health. All participants gave informed written consent. Testing was conducted in a minimum of two sessions at participant’s homes, with length and number of sessions dictated by the ALS participants’ level of fatigue. Computerized tests were administered using a laptop and touch-screen. For the two ALS participants with insufficient upper limb strength to easily manage touching the screen, instructions were given verbally and the screen was touched by the researcher. The order of test administration for each session was counterbalanced for both participant groups. Interviews with ALS informants were conducted independently of the ALS participant.

Statistical analysis

Statistical analysis was performed with Statistical Package for the Social Sciences (SPSS v14.0.2). For normally distributed data, group comparisons were made using either one-way analysis of variance (ANOVA), or repeated measures ANOVA with post hoc pairwise comparisons using Bonferronni adjustment. Welch statistics were used if the assumption of equality of group variances was violated. Independent-sample t-tests were used to compare groups on demographic variables. Categorical data were analysed using Pearson’s chi-square tests. Individual impairment on tasks was evaluated using a modified t-test recommended by Crawford and Howell (1998) for comparing an individual with a control sample. The criteria for classical dissociations recommended by Crawford and Garthwaite (2005a) were used to determine whether individuals showed a dissociation between performance on individual tasks. This involved establishing whether an individual had a deficit on an individual task (described above) then applying the Revised Standardized Difference Test, which compared the individual’s standardized difference between two tasks with the distribution of standardized difference obtained from the controls. This method is robust to Type 1 errors when sample sizes are small and control data are non-normal (Crawford and Garthwaite, 2005a, b). For the Neuropsychiatric Inventory, a significant behavioural impairment was determined by the presence of a score of ≥3 on 2 or more behaviours as per Murphy et al. (2007).

One control participant refused to complete the Holiday Apartment task as requested. One participant with ALS refused to attempt the Stockings of Cambridge task. The physical disabilities of some participants with ALS precluded them from completing either the oral or written versions of verbal fluency tasks. Sixteen of the 18 participants with ALS completed the oral fluency tasks, and 9 of 18 completed the written fluency tasks.


Tasks sensitive to orbitomedial prefrontal cortex function

Neuropsychiatric Inventory

Scores on the Neuropsychiatric Inventory ranged from 0 to 57, with 9 (50%) participants scoring 12 or higher. The number of domains in which significant behavioural changes (score ≥3) were reported for the same participant ranged from one to eight. Nine participants (50%) had significant behavioural impairment (a score ≥3 on two or more behaviours). Behavioural changes reported in order of frequency were apathy, irritability/lability, aggression/agitation, stereotypy, delusions, sleep disturbance, eating/appetite change, depression and disinhibition. There were no significant correlations between Neuropsychiatric Inventory scores and performance on cognitive tasks.

Faux Pas Test

Responses were scored by two independent raters blind to group status. Inter-rater agreement was high (91%). Instances of disagreement were negotiated with a third rater (also blind to group status) until consensus was reached. Table 2 shows the mean number of stories correctly identified in faux pas and control conditions, composite scores for the two groups and the number of comprehension questions answered correctly. Composite scores were calculated as follows: the sum of the number of stories correctly identified for both conditions plus the number of correct answers on faux pas detail questions (i.e. correct identification of character’s mental states) divided by the total number of possible answers (i.e. 50). The ALS group accurately classified fewer stories overall on this task, F(1,34) = 9.28, P < 0.01, but importantly there was a significant interaction between group and story-type, F(1,34) = 5.29, P = 0.03. Pairwise comparisons revealed that the ALS group identified significantly fewer faux pas stories than did controls (P < 0.01), but there was no group difference in accuracy at identifying control stories (P = 0.84). Consistent with this, composite scores of the ALS group were significantly lower than those of the control group, F(1,34) = 16.38, P < 0.01. In contrast, the two groups did not differ in their accuracy at answering comprehension questions about the stories, F(1,34) = 1.00, P = 0.32. We subsequently repeated the analyses co-varying out performance on oral letter fluency to control for possible effects of dorsolateral prefrontal cortex dysfunction. The findings were largely unchanged, with the ALS group performing more poorly overall on the task, F(1,31) = 9.72, P < 0.01, and a significant story-type by group interaction, F(1,31) = 6.28, P < 0.01, indicating the ALS group had differential difficulty on faux pas stories. Nine of the 18 individuals with ALS were classified as impaired on the faux pas accuracy score. The Revised Standardized Difference Test was then applied to the faux pas and control story measures: six patients with ALS showed classical dissociations between these two task conditions.

View this table:
Table 2

Performance by ALS and control groups on tasks related to orbitomedial prefrontal cortex

ALS group mean (SD)Control group mean (SD)
Faux Pas Test
    Faux pas story (Theory of Mind condition)8.39 (0.98)9.39 (0.61)
    Control story9.17 (0.79)9.22 (0.81)
    Composite score0.65 (0.08)0.77 (0.01)
    Comprehension question19.94 (0.24)20 (0)
Probabilistic Reversal Learning Test
    Number of learning trials required18.89 (10.95)16.17 (5.68)
    Number of reversals achieved3.22 (1.00)3.44 (1.25)
    Mean number of trials to reverse response8.69 (10.60)6.02 (5.15)
Holiday Apartment Task: decision complexity
    (search index)
        Trial 1 (2 × 4 grid)0.004 (0.850)0.425 (0.731)
        Trial 2 (4 × 6 grid)0.218 (0.767)0.650 (0.587)
        Trial 3 (6 × 7 grid)−0.056 (0.886)0.542 (0.434)
Holiday Apartment Task: information viewed
        Trial 190.28 (15.19)96.32 (8.57)
        Trial 281.73 (24.04)73.74 (22.54)
        Trial 373.98 (32.36)65.83 (25.49)
Aprosodia Battery: emotion identification
    Comprehension subtest trials:
        Word18.83 (4.25)20.44 (1.79)
        Monosyllable16.44 (3.71)18.89 (1.61)
        Asyllabic14.22 (3.35)15.67 (3.48)
Aprosodia Battery: discrimination
    Discrimination subtest19.72 (1.97)20.39 (1.72)

Probabilistic Reversal Learning Test

All participants completed the initial learning phase and one or more reversals successfully. The ALS and control groups did not differ significantly for either the initial number of learning trials required, F(1,25.53) = 0.88, P = 0.36, the number of reversals achieved, F(1,34) = 0.35, P = 0.56, or the mean number of trials to reverse, F(1,34) = 0.03, P = 0.87 (Table 2). When considering individual performances, one participant with ALS was impaired on the mean number of trials required to reverse responding.

Holiday Apartment Task

Table 2 shows the pattern of information acquisition for the two groups at each level of decision complexity. Separate analyses were conducted for each level. Although the search index of the control group was more positive than the ALS group at all levels, the groups were not significantly different at the 1st level of complexity, F(1,34) = 2.43, P = 0.13 and only approached significance at the 2nd level, F(1,34) = 3.46, P = 0.07. On the third and most complex level, however, there was a significant difference in the search index of the two groups, F(1, 25.04) = 6.54, P = 0.02; the control group searched predominantly by attribute, and the ALS group had a more apartment-based search pattern. There were no significant differences between groups for the amount of information viewed, P = 0.16, 0.32 and 0.41, or time taken to choose an apartment, P = 0.76, 0.73 and 0.41 for complexity levels 1–3, respectively. Since other decision-making tasks have been found to involve dorsolateral prefrontal cortex as well as orbitomedial prefrontal cortex, analysis of the search index for the third level trial was repeated, co-varying out performance on oral letter fluency. The search index of the ALS group remained significantly different from the controls although only marginally, F(1,30) = 4.13, P = 0.05.

Individual analysis of Holiday Apartment Test performance was based upon level 3 index scores. Eight (44%) patients with ALS used a predominantly apartment-based search pattern in which all, or most, attributes for one apartment were examined before moving to the next apartment. This search pattern was not seen in any of the control participants, 11 of whom used a predominantly attribute-based search strategy, with the remaining six using a mixed method combining apartment- and attribute-based searches. The difference was significant for all eight patients with ALS using Crawford and Howell’s (1998) modified t-test and is clearly evident in Fig. 1. Additionally five of the eight participants with ALS, who used apartment-based search strategies, examined all information available before making a decision, compared to only 24% of controls, a difference that approached significance, χ2(1) = 3.59, P = 0.06.

Figure 1

Search indexes on Holiday Apartment Task on Trial 3 (six apartments with seven attributes) of individuals in ALS and control groups. An index of −1 indicates a completely apartment-based search and an index of +1 indicates a completely attribute-based search.

Aprosodia Battery

The ALS group identified emotional expressions from prosody less accurately than the control group on the Comprehension task, F(1,34) = 4.41, P = 0.04. There was also a significant effect of stimulus-type, F(2,68) = 40.47, P < 0.001, but no significant group by stimulus-type interaction, F(2,68) = 0.53, P = 0.60. Across all participants, prosody was recognized more accurately with words than with monosyllables, which were more accurately recognized than the asyllabic stimuli. In a separate analysis there was no evidence that the ALS group had differential difficulty identifying particular emotions [emotion by group, F(5,170) = 0.37, P = 0.87]. On the same-different discrimination task there was no significant difference between the ALS and control groups, F(1,34) = 1.17. P = 0.29. Taken together, these results indicate that the ALS group was significantly less able to identify emotional expression by prosody than control participants but had no difficulty discriminating between different emotional expressions. Six participants with ALS were classified as impaired on emotion identification, with two of these also impaired on the same-different discrimination task.

Correlations between performances on tasks sensitive to orbitomedial prefrontal cortex by patient with ALS revealed a single significant correlation, between identifying emotions on the prosody task and identifying social faux pas, rrho = 0.48, P = 0.045.

Tasks sensitive to dorsolateral prefrontal cortex function

Verbal fluency

The ALS group performed worse (i.e. longer mean thinking time per word) than the control group on oral fluency tasks, F(1,32) = 4.35, P = 0.045. Performance was poorer on letter fluency than category fluency overall, F(1,32) = 23.18, P < 0.001, but the task by group interaction was not significant, F(1,32) = 0.64, P = 0.43. This pattern was repeated for written fluency tasks in which the ALS group performed worse overall, F(1,25) = 8.68, P < 0.01. Once again performance was poorer over both groups on letter fluency than category fluency, F(1,25) = 19.77, P < 0.001, but the group by task interaction approached significance, F(1,25) = 3.62, P = 0.07 (Table 3). Thus, patients with ALS were impaired on oral and written verbal fluency tasks, even when using fluency indices that control for the effects of physical disability. Individual analysis revealed five individuals with ALS with impaired letter fluency performance and seven individuals with impaired category fluency performance, with three individuals impaired on both letter and category tests.

View this table:
Table 3

Performance by ALS and control groups on tasks related to dorsolateral prefrontal cortex

ALS group mean (SD)Control group mean (SD)
Verbal Fluency
    Oral Fluency Letter Indexa4.06 (1.55)3.19 (1.18)
    Oral Fluency Category Indexa2.84 (1.15)2.32 (0.61)
    Written Fluency Letter Indexa10.14 (5.48)5.93 (2.54)
    Written Fluency Category Indexa5.18 (2.22)3.94 (1.72)
One-touch Stockings of Cambridge
    Mean proportion of perfect solutions (mean)
        Easy trials (1–3)0.89 (0.09)0.94 (0.06)
        Hard trials (4–6)0.49 (0.31)0.63 (0.28)
    Mean no. of attempts for correct solution
        Easy trials (1–3)1.24 (.24)1.08 (0.08)
        Hard trials (4–6)2.15 (1.22)1.57 (0.51)
    Time to correct solution
        Easy trials (1–3)4.12 (0.18)4.06 (0.17)
        Hard trials (4–6)4.83 (0.20)4.69 (0.23)
  • a Index scores = thinking time per word (lower scores are better).

One-touch Stockings of Cambridge

Trials were grouped according to difficulty (easy: 1–3 moves, hard: 4–6 moves) for analyses. There was no significant difference between groups for the proportion of perfect solutions achieved, F(1,33) = 3.07, P = 0.09. The ALS group did, however, require more attempts to determine the correct solution than the control group, F(1,33) = 5.05, P = 0.03. Both groups required more attempts for accurate performance on hard trials than easy trials, F(1,33) = 20.68, P < 0.001, but there was no significant interaction between group and difficulty level, F(1,33) = 1.86, P = 0.18. In addition both groups took significantly longer to solve difficult trials (P < 0.001), but there was no significant effect of group, or group by time interaction. The performances of four patients with ALS were classified as impaired on mean attempts to determine the correct solution.

Summary of individual performance of ALS group on tasks sensitive to orbitomedial and dorsolateral prefrontal cortex

Table 4 presents the Z-scores of individual participants with ALS for tasks sensitive to orbitomedial and dorsolateral prefrontal cortex that were classified as impaired [using Crawford and Howell’s (1998) modified t-test] and also presents informant ratings on the Neuropsychiatric Inventory. Tests for classical dissociation between an impaired task primarily associated with one region of prefrontal cortex and tasks primarily sensitive to the other prefrontal region were conducted using the Revised Standardized Difference Test (Crawford and Garthwaite, 2005a). The Z-scores for those tasks meeting the criteria for classical dissociations are marked with an asterisk in Table 4. The performance of 13 patients with ALS showed classical dissociations on at least one task, with six individuals showing only one dissociation. Five of the ALS sample (28%) did not show any classical dissociations. There was no consistent pattern regarding impaired and dissociated tasks—some individuals showed impairment and dissociations only on tasks sensitive to functioning of orbitomedial prefrontal cortex (6/18, 33%), others showed impairment and dissociations only on tasks predominantly sensitive to functioning of dorsolateral prefrontal cortex (3/18, 17%), despite the inclusion of fewer measures associated with the latter brain region. Finally some individuals showed impairment and dissociation on both sets of tasks (4/18, 22%). Although there was considerable variability in time since disease onset between patients with ALS there were no significant correlations between this measure, or a measure of disease severity (ALS Functional Rating Scale-Revised), and any cognitive tasks.

View this table:
Table 4

Summary of impaired performance of participants with ALS on individual tasks (Z-scores) and on tasks showing classical dissociations (*)

ALS participant number
Orbitomedial prefrontal cortex tasks
    Faux Pas Testa: faux pas condition−3.9*3.92.3−2.3*−3.9*−3.9*
    Reversal Learning: mean trials to reverse2.3
    Holiday Apartment: Decision trial #3−3.6*3.0−3.6*3.63.0−3.6*−3.6*−3.6*
    Aprosodia: Battery emotion identification2.2−5.9*−2.6*−3.0*1.842.76
Neuropsychiatric Inventory+++++++++
Dorsolateral prefrontal cortex tasks
    Fluency indexes: letter
    Oral (PRW)1.94n/an/a−4.3*
    Written (S, four letter C)n/an/a−2.2n/an/an/an/an/a−4.0*n/a−5.7*2.19n/a
    Fluency indexes: category
    One-touch Stockings of Cambridge
    Mean attempts for correct solutions−3.0*−3.7*−5.3*−7.1*n/a
  • a All listed Faux Pas scores were impaired and classically dissociated from control stories [using Revised Standardized Difference Test (Crawford and Garthwaite, 2005a)]; Z-scores listed for all other tasks are impaired [modified t-test (Crawford and Howell, 1998)]. Z-scores in bold followed by an asterisk were classically dissociated from tasks sensitive to other prefrontal region. + on Neuropsychiatric Inventory indicates score ≥3 on two or more behaviours on Neuropsychiatric Inventory; n/a = task not administered.


This study is, to the best of our knowledge, the first to examine systematically cognitive abilities and behavioural changes associated with orbitomedial prefrontal cortex in patients with ALS. We found impairments in cognitive abilities and behaviour linked with orbitomedial prefrontal cortex in a typical group of patients with ALS attending a Motor Neuron Disease outpatient clinic who did not meet clinical criteria for dementia. Relative to a carefully matched control group, the ALS group identified fewer faux pas in stories, identified emotions from prosody less well, and adopted a similar strategy on a multi-attribute decision-making task to that used by patients with lesions to ventromedial prefrontal cortex, but not by patients with damage to dorsal and/or lateral prefrontal cortex or healthy controls (Fellows, 2006). Individually, 10 patients with ALS met criteria for individual impairment on at least one of these tasks and for classical dissociation with tasks linked with dorsolateral prefrontal cortex, although five of these individuals showed dissociation on a single impaired task. In addition, 50% of the participants with ALS met criteria for significant behavioural impairment on the Neuropsychiatric Inventory (Murphy et al., 2007).

The ALS group had difficulty identifying the presence of a faux pas in social situations and in answering questions that clarified the nature of the faux pas. They had no difficulty, however, identifying stories that did not contain a faux pas or in answering comprehension questions about the stories that indicated the stories had been understood. This pattern of performance indicates that the ALS group had difficulty in the Theory of Mind components of the task only. Identifying faux pas involves representing the knowledge of others (that a person unintentionally says something they shouldn’t have said) and also the emotional state of others (that someone hearing this would feel hurt or upset). Previous studies have indicated that affective Theory of Mind is particularly reliant on orbitomedial prefrontal cortex, whereas cognitive Theory of Mind may depend on more extensive prefrontal regions including dorsolateral prefrontal cortex (Stone et al., 1998; Shamay-Tsoory et al., 2005, 2007). Although it was not possible to determine accurately whether the cognitive, or affective, Theory of Mind component of this task was differentially affected in our ALS sample, when we covaried fluency performance to control for deficits linked to dorsolateral prefrontal cortex, the ALS group were still impaired at recognizing faux pas and answering specific questions about these. This suggests their difficulty was less likely due to functions associated with dorsolateral prefrontal cortex and more likely to lie primarily with the affective Theory of Mind component of the task. Analysis of individual data revealed that six patients with ALS were significantly impaired on the faux pas condition and showed a classical dissociation with the control condition, although the magnitude of the impairments was mild. Our results contrast with those of Gibbons et al. (2007), who used a Theory of Mind task involving humorous cartoons and stories and found no differential difficulty with Theory of Mind. The use of humour in both Theory of Mind and control conditions may, however, have confounded results given the recent finding that humour is processed in a distributed network that includes orbitomedial prefrontal cortex (Samson et al., 2008). Theory of Mind is a critical component of cognitive processes that contribute to successful functioning within social contexts and in the maintenance of personal relationships. Even subtle deficits in these functions have the potential to cause noticeable difficulties in day-to-day social interactions, which may be heightened for families and caregivers during the difficult physical progression of this disease.

The ability to recognize and respond appropriately to the emotions of others is an important moderator of social behaviour, and important in reinforcing and influencing future behaviour. Our ALS group was significantly worse than the controls at recognizing emotional prosody on the Aprosodia battery, although they had no difficulty on the discrimination task. This indicates the difficulty lay primarily with the identification of the emotion itself. Previous studies investigating emotional processing in ALS populations have produced mixed results (Papps et al., 2005; Schmolck et al., 2007; Zimmerman et al., 2007), although there is little consistency across studies in terms of stimulus modality or task used. Zimmerman et al. (2007) conducted the only other study testing prosodic recognition in ALS and found no impairment using a task in which all stimuli were complete sentences. The prosodic stimuli in this study, however, comprised asyllabic utterances, syllables and a complete sentence, probably making our task more difficult and more sensitive. Individually, the scores of 6/18 patients with ALS were significantly impaired. Impaired identification of emotions can interfere with both interpretation of the behaviour of others and the association of emotions with internal and external stimuli.

The focus of the Holiday Apartment decision-making task is the mental heuristics, or strategy, used by participants to reach their decision in the context of known and limited options and complete availability of information (Fellows, 2006). Control participants used more attribute-based strategies at all levels of complexity. Patients with ALS followed this trend at the moderate level of complexity, but their pattern of responding changed significantly from the control group on the most complex trial, with a shift towards a primarily apartment-based search pattern by a sizeable subgroup (44%) of patients with ALS. Fellows (2006) found that only patients with focal damage to the ventromedial prefrontal cortex used an apartment-based search pattern performing a comparable decision-making task; healthy control subjects and individuals with dorsal and/or lateral prefrontal damage primarily used an attribute-based strategy as did our control participants.

Despite this finding that strategy type was not influenced by dorsolateral prefrontal cortex lesions, given evidence that gambling tasks recruit both orbitomedial and dorsolateral prefrontal cortex (Manes et al., 2002; Fellows and Farah, 2005), we controlled for the possible contribution of impaired dorsolateral prefrontal cortex to task performance by co-varying the effects of oral letter fluency. The difference in the strategy used by the ALS group remained significant, consistent with the interpretation that use of this different approach to measure decision-making is more likely to reflect altered functioning of orbitomedial than dorsolateral prefrontal cortex.

Attribute-based searching in these types of paradigms is thought to reflect use of a ‘maximizing strategy’ where the best alternative is sought, whilst an apartment-based search is argued to involve a ‘satisficing’ strategy where an acceptable, ‘good enough’ alternative is selected (Simon, 1956). Fellows (2006), has argued that use of a satisficing approach by individuals with orbitomedial prefrontal cortex lesions might reflect difficulty representing the value of stimuli. Although by no means the only possible mechanism underlying decision-making difficulties, recent studies have highlighted the importance of determining the relative value of alternatives that are compared during decision making. Converging evidence indicates that orbitomedial prefrontal cortex probably has a critical role in this process, including findings from lesion studies (Fellows and Farah, 2007), single-unit recording in primates (e.g. Padoa-Schioppa and Assad, 2006) and functional MRI in normal humans (McClure et al., 2004; O’Doherty, 2004). Neuroimaging studies have similarly suggested a role for orbitomedial prefrontal cortex in representing values when making economic choices between goods; Chib et al. (2009) found activation in a single area in ventromedial prefrontal cortex correlated with participant’s valuations for different categories of goods.

In summary, the Holiday Apartment Task used in this study is a realistic reflection of many real-life decision-making situations, as these often involve choices between multiple options in the presence of complete information, yet are sufficiently difficult and complex that strategies are required to assist in the decision-making process. Our data indicate that a large subgroup of patients with ALS used a strategy that can produce decisions that are ‘good enough’ rather than ‘the best’, whereas no controls used such a strategy. A change in approach to decision making of this sort could have serious implications when patients with ALS are making financial arrangements for their care and their families’ future welfare, as well as their involvement in care decisions and adherence to treatment guidelines.

The ALS group had no difficulties with the Probabilistic Reversal Learning Task, the sole exception to the general pattern of impaired performance on tasks sensitive to dysfunction of the orbitomedial prefrontal cortex. Although impaired reversal learning in humans has been linked with this region, control participants normally perform at ceiling on this type of task and impaired performance has followed significant, frequently bilateral, damage in a fairly specific region of the orbitomedial prefrontal cortex (Fellows and Farah, 2003; Hornak et al., 2004). As degeneration in typical ALS patients is unlikely to be this severe, it is possible that probabilistic reversal learning tasks are not sensitive enough to detect relatively subtle changes in orbitomedial prefrontal cortex that may be present in patients with ALS.

Examination of the performance of individuals across tasks indicated that the between-group differences were not the result of one or two outlier participants. Not only did a considerable number of participants with ALS have significant task deficits using a modified t-test that compared them with the control sample, a surprising number of patients with ALS (10/18) also showed classical dissociations between impaired performance on tasks sensitive to orbitomedial prefrontal cortex and performance on tasks linked with other prefrontal regions. The magnitude of the deficits remained mild, however, indicating that functional impairments are likely to be subtle. Nevertheless even subtle impairments of social cognition, decision-making and emotional processing can affect interpersonal relations profoundly, particularly when such changes are not recognized as consequences of neural processes. Performance across tasks did not indicate a hierarchy of task sensitivity, with no systematic ordering of task impairment. The measures selected are thought to involve cognitive functions that recruit separate subregions within this rather broad anatomical area, thus differing patterns of impairment may reflect variable distribution of ALS-related neuropathology within orbitomedial prefrontal cortex. While these tasks are sensitive to orbitomedial prefrontal cortex dysfunction, however, they are not specific to this region, which means variable impairment across tasks may also reflect dysfunction of other brain regions that make crucial contributions to these functional domains (Clark and Manes, 2004; Rudebeck et al., 2008). Whilst our findings reveal a novel and important pattern of cognitive and behavioural changes in ALS without dementia, a clearer understanding of the neural bases of these changes will only emerge following studies that also incorporate measures of structural or functional brain changes.

Our secondary aim was to assess performance on tasks linked with dorsolateral prefrontal cortex. Our ALS group took longer to think of words in verbal fluency (even when controlling for effects of physical disability), replicating the results of many other researchers (Ludolph et al., 1992; Abrahams et al., 2000; Ringholz et al., 2005) and supporting the suggestion that verbal fluency is a sensitive measure of cognitive change in ALS. Patients with ALS also required more attempts to solve both easy and hard trials on a planning task.

One alternative account of these findings that must be considered is that this group of patients with ALS comprises a biased subset in which cognitive impairments are generally more severe, even though the individuals have not received a diagnosis of dementia. Close scrutiny of the data suggests this is unlikely to be the case. Firstly, the incidence of impairment on fluency measures in this sample (28% on letter fluency and 39% on category fluency) is similar to that found in previous studies (Abrahams et al., 1995; Flaherty-Craig et al., 2006). A second important characteristic of our data is that impaired performance on measures sensitive to orbitomedial and dorsolateral prefrontal cortex was dissociated within this sample. Some participants performed worse, and had classical dissociations, on tasks linked with orbitomedial prefrontal cortex (33%), while other participants (17%) had more difficulty, and had classical dissociations, on measures sensitive to dorsolateral prefrontal cortex despite the smaller number of measures. Whilst 22% of patients with ALS had classical dissociations suggesting impairment of both regions, the remaining patients with ALS (28%) showed no classical dissociations or were not significantly impaired on any tasks. This pattern of performances does not indicate that the ALS sample is more globally-impaired than those included in previous studies. Rather, these data suggest that there is heterogeneity in cognitive and behavioural impairments, with clusters of patients with ALS with different patterns of spared and impaired cognitive abilities associated with different regions within prefrontal cortex. Amongst that heterogeneity is a subgroup (28%) with no marked cognitive difficulties who, given the absence of associations found between impaired task performance and disease progression or severity, may never develop impaired cognition. If cognitive heterogeneity is disease-related, this suggests possible heterogeneity in the neural expression of the disease across broad regions of prefrontal cortex, or more extensive cortical regions. Neuropsychological measures do not provide a direct probe of neural functioning, however, and ultimately neuropathological studies will be needed to confirm this hypothesis, as well as replications of these findings with other ALS samples.

In summary the main finding of this study is that the ALS group was mildly, but consistently impaired on most measures linked with the functional integrity of orbitomedial prefrontal cortex, even though almost 30% of the sample had no significant cognitive impairment. Whilst it is known that a proportion of patients with ALS develop frontal variant FTLD, which causes complex behavioural changes including poor decision-making, changes in personality and altered social understanding, there is little awareness that subtle changes in these functions may be present in a significant proportion of typical ALS patients. The presence of such difficulties, albeit mild, will inevitably strain interpersonal and familial relationships and occurs at a time when carers and family members are already experiencing increased stress through having to cope with a truly debilitating illness. Alterations in decision-making and ability to manage finances can be exceedingly problematic at a time when individuals with ALS are making financial arrangements for their care and their families’ future welfare. In addition, these changes in cognitive processing and behaviour can interfere significantly with the patient’s ability to participate in care decisions, engage in self-care practices and adhere to treatment guidelines (e.g. Olney et al., 2005). The first step to managing the consequences of these changes is recognition of their likely occurrence as part of the disease process and appropriate assessment. As the impairments are usually mild, once identified there are opportunities to develop management methods, most likely with a combination of education and rehabilitation techniques. This provides an opportunity to take an important step towards improving management and quality of care and support for individuals with ALS and their families.


We express our appreciation to the individuals with ALS and their families in Auckland; and Linda Oliver and Frances Kissel of the New Zealand Motor Neuron Disease Association for their generous and invaluable assistance during the course of the study. We are grateful to Lesley Fellows of the Montreal Neurological Institute for supplying us with two tasks, the Probabilistic Reversal Learning Task and the Apartment Task; and to Donna Addis for reading a version of this article. This study was supported by a grant from the Julius Brendel Trust, and a New Zealand Top Achiever Doctoral Scholarship to the first author.

amyotrophic lateral sclerosis
frontotemporal lobar dementia


View Abstract