Abstract
Behavioural-variant frontotemporal dementia is characterized by a number of ostensibly disparate clinical features, which have largely been considered independently. This update proposes an integrated conceptual framework for these symptoms, by bringing together findings from animal studies, functional neuroimaging and behavioural neurology. The combined evidence indicates that many of the clinical symptoms––such as altered eating behaviour; overspending and susceptibility to scams; reduced empathy and socially inappropriate behaviour; apathy and stereotyped/ritualistic behaviour––can be conceptualized as a common underlying deficiency in goal-directed behaviour and the concomitant emergence of habits. This view is supported by similarities between the characteristic patterns of frontostriatal and insular atrophy in behavioural-variant frontotemporal dementia and the circuitry of homologous brain regions responsible for goal-directed and habitual behaviour in animals. Appreciating the impact of disturbance in goal-directed behaviour provides a new, integrated understanding of the common mechanisms underpinning prototypical clinical symptoms in behavioural-variant frontotemporal dementia. Furthermore, by drawing parallels between animal and clinical research, this translational approach has important implications for the development and evaluation of novel therapeutic treatments, from animal models through to behavioural interventions and clinical trials in humans.
Introduction
Frontotemporal dementia refers to a heterogeneous group of neurodegenerative disorders primarily affecting the frontal and temporal lobes of the brain. While three main clinical variants of frontotemporal dementia are recognized, including two language variants (semantic dementia and progressive non-fluent aphasia) and the behavioural variant, the current Update focuses on the latter. Clinical features of behavioural-variant frontotemporal dementia (bvFTD) have largely been considered independently. This is not surprising, given that symptoms such as increased preference for sweet foods, reduced empathy towards family members and stereotypical/ritualistic behaviour appear, superficially at least, to have little in common. Consequently, research into the symptoms of bvFTD has lacked an integrated behavioural framework, despite the shared involvement of frontostriatal-insular brain regions. In this Update, we bring together evidence from animal studies, functional neuroimaging and behavioural neurology to provide support for a new way of conceptualizing the behavioural changes in bvFTD as due to a disturbance in the capacity for goal-directed behaviour and the concomitant emergence of habits. As such, we highlight the intersection of animal and clinical research in studying goal-directed behaviour and outline a path for future research to systematically investigate goal-directed behaviour in this neurodegenerative syndrome.
Goal-directed behaviour in animals
The capacity for goal-directed behaviour allows us and other animals to control our environment in order to achieve our basic needs and desires. Theories of goal-directed behaviour emphasize knowledge of the ‘causal relationship’ between an action and its consequences, and the ‘desirability’ of those consequences—referred to as ‘belief plus desire’ (Goldman, 1970). This ‘belief plus desire’ account has led to two criteria that can be applied across species, which must be met for behaviour to be considered goal-directed (Balleine and Dickinson, 1998). From this perspective, goal-directed behaviours must be sensitive to manipulations of both (i) the action-outcome contingency; and (ii) the value of the outcome.
Behavioural paradigms used in animal studies. During instrumental training, animals are trained on two action-outcome associations. Following (A) contingency degradation or (B) outcome devaluation procedures, the animal’s performance on a choice test (where frequencies of A1 and A2 are measured) indicates whether their actions are controlled by ‘belief’ or ‘desire’, respectively. A1 = action 1 (left lever), A2 = action 2 (right lever), O1 = outcome 1 (sucrose), O2 = outcome 2 (food pellet).
Laboratory-based tests have been developed to reliably detect these features of goal-directed behaviour in rodents, with the two most common being tests of (i) contingency degradation; and (ii) outcome devaluation (Fig. 1 and Box 1). In contingency degradation tests (Fig. 1A), the instrumental performance of both rats and mice is sensitive to selective changes in the causal relation between their actions and their consequences i.e. changes in ‘belief’. In contrast, the outcome devaluation test (Fig. 1B) is designed to measure changes in ‘desire’, and demonstrates that rats and mice are sensitive to selective changes in the value of the outcome. At the neural level, the contingency degradation and outcome devaluation tests have been used to demonstrate that the acquisition and ongoing performance of goal-directed behaviour involves frontostriatal and insular regions.
In rodents, encoding of action-outcome associations occurs in the posterior dorsomedial striatum (cf. Balleine and O'Doherty, 2010) (Fig. 2A). During acquisition, this encoding requires direct cortical inputs to the posterior dorsomedial striatum from the prelimbic prefrontal cortex (Balleine and Dickinson, 1998) and limbic inputs from the basolateral amygdala (Balleine et al., 2003). Once goal-directed actions have been acquired, however, prelimbic prefrontal cortex input to the posterior dorsomedial striatum is no longer required (Ostlund and Balleine, 2005). The orbitofrontal cortex (OFC) is necessary to recall the outcome associated with a specific action (Ostlund and Balleine, 2007; Bradfield et al., 2015). Interestingly, changes in the value of outcomes involve the basolateral amygdala (Wassum et al., 2009), with this information subsequently encoded in the anterior insular cortex (Balleine and Dickinson, 2000; Parkes and Balleine, 2013). Finally, the insular cortex inputs to the nucleus accumbens for subsequent value-related changes in instrumental performance (Parkes et al., 2015).
Animal and human homologies in goal-directed and habitual behaviour. Brain regions involved in goal-directed (dark blue) and habitual (light blue) behaviour in rats (A) and in humans (B). Areas shaded in red stripes depict the typical pattern of brain atrophy in bvFTD. BLA = basolateral amygdala; DL Str = dorsolateral striatum; DM Str = dorsomedial striatum; IC = insula cortex; NAcc = nucleus accumbens; PL = prelimbic prefrontal cortex.
Contingency degradation test (Fig. 1A): Animals are trained on two action-outcome associations (i.e. A1→O1, A2→O2), after which the relationship between one action and its consequence is selectively degraded, such that performing the action or not, leads to equal probability of the outcome being obtained. As they learn that the specific action (A1) is no longer causal, both rats (cf. Balleine and Dickinson, 1998) and mice (Wiltgen et al., 2007) selectively reduce their performance of action 1 (A1), without changing their performance of action 2 (A2). This demonstrates that animals have learned the specific action-outcome associations.
Outcome devaluation test (Fig. 1B): Animals are trained on two action-outcome associations (i.e. A1→O1, A2→O2), after which the value of one of the two outcomes (e.g. O1) is altered––either via ‘sensory specific satiety’ (e.g. by allowing the animal to consume the food outcome until satiated) or via a ‘conditioned taste aversion treatment’ (e.g. by pairing consumption of the food outcome with an injection of lithium chloride, causing the animal to feel nauseous) (for a review see Balleine, 2009). Both approaches lead to a subsequent selective reduction in the performance of the outcome’s associated action (A1) relative to the other action (A2) (Balleine, 2001; Carvalho et al., 2002). This demonstrates that animals have integrated knowledge of the action-outcome associations with knowledge of the altered value of the outcomes to modify their subsequent behaviour.
Overtraining: Overtraining paradigms have been used to change goal-directed actions into habitual actions. Here, a rat learns that A1→O1, and is subsequently allowed to continue performing A1 for O1 for many days without changes in the action-outcome contingency. As a result, the performance of A1 then becomes habitual, no longer controlled by ‘belief’ or ‘desire’ for O1, which may be assessed by contingency degradation and outcome devaluation tests, respectively (Balleine and O'Doherty, 2010; Dezfouli and Balleine, 2012).
Once goal-directed actions have been established, their continued invariant use often leads to habitual behaviour, such that the action is performed without reference to its outcome (Box 1). At the neural level, habits appear to be mediated by striatal regions that are distinct from those involved in goal-directed actions. Whereas the initial acquisition of action-outcome associations is encoded by the medial prefrontal cortex and dorsomedial striatum, the development of habits depends on a region of the dorsolateral striatum (Yin et al., 2004). During the course of overtraining, critical changes in dopamine D2 receptor activity in the dorsolateral striatum are associated with habits, and reflect changes in postsynaptic plasticity (Shan et al., 2015).
Importantly, it appears that goal-directed and habitual actions form a continuum: as the capacity for goal-directed control diminishes, habits emerge (cf. Balleine et al., 2009). Should conditions change suddenly and unexpectedly, however, the goal-directed circuitry can re-engage and inhibit habitual behaviour. Notably, habits can emerge not only as a consequence of overtraining, but also as a consequence of damage to the neural system that mediates goal-directed action (Ostlund and Balleine, 2008). As such, habits are actions that continue to be performed despite no longer being associated with favourable outcomes. In this case, the habits that emerge are clearly maladaptive; they are no longer subject to regulation by the goal-directed system and often continue (i.e. perseverate) in situations where such performance is not optimal (Corbit et al., 2014; Furlong et al., 2015). In sum, a large body of work using animal models has led to the development of paradigms that reliably assess goal-directed and habitual behaviour and their associated neural correlates.
Goal-directed behaviour in humans
Findings from functional neuroimaging studies in humans complement the animal literature, highlighting the importance of regions such as the OFC, insular cortex, amygdala and striatum in coding and maintaining reward representations that guide flexible, goal-directed behaviour (O'Doherty, 2004; Balleine and O'Doherty, 2010) (Fig. 2B). These regions have been compared in terms of their connectivity and function in a variety of species, including rodents, primates and humans. Although there have been controversies in this literature (cf. Wise, 2008), there is a developing consensus that regions of interest in goal-directed and habitual behaviour are homologous across species. While a full review of this literature is beyond the scope of this Update, evidence indicates that the rodent dorsomedial and dorsolateral striatum are homologous to human caudate and putamen, respectively (Balleine and O'Doherty, 2010; Woolley et al., 2013; Heilbronner et al., 2016); that rodent medial and ventrolateral OFC regions are homologous to the visceromotor and sensory OFC networks described in primates and humans, respectively (Öngür and Price, 2000; Heilbronner et al., 2016); and that the anterior insula, anterior cingulate cortices and the cortico-striatal emotional network are all also homologous across species (Schwarz et al., 2013; Woolley et al., 2013; Vogt and Paxinos, 2014; Heilbronner et al., 2016).
Contingency degradation tests, which vary the causal relationship between button presses and monetary reward over time, have been used in healthy adults (Tanaka et al., 2008; Liljeholm et al., 2011). These studies showed greater activity in the medial OFC and dorsomedial striatum (anterior caudate nucleus) when consequences are highly contingent on specific actions compared to when the action-outcome contingency is low. In addition, outcome devaluation procedures have demonstrated increased OFC activity in healthy adults during the selection of valued compared to devalued actions (Valentin et al., 2007), and impaired value-directed choice in patients with focal lesions of the ventromedial prefrontal cortex, encompassing the OFC (Reber et al., 2017).
Regions involved in habitual behaviour in humans have also been identified. In a functional MRI study, we have previously shown differential effects of training duration on sensitivity to outcome value (Tricomi et al., 2009). Some participants were given relatively little training to press buttons to earn snack foods, whereas others were overtrained. Behaviourally, the less trained participants showed a typical outcome devaluation effect, whereby sensitivity to devaluation was associated with activity in the medial prefrontal cortex and caudate nucleus. In contrast, the performance of overtrained participants was insensitive to outcome devaluation and activity in the right posterior putamen was predictive of overtraining-induced performance. These areas are homologous to those that mediate goal-directed and habitual action in rodents.
Of relevance here, in healthy humans, the application of the abovementioned contingency degradation and outcome devaluation paradigms (Valentin et al., 2007; Klossek et al., 2008; Tanaka et al., 2008; Tricomi et al., 2009; Liljeholm et al., 2011; Friedel et al., 2014; Eder and Dignath, 2015) and other related behavioural paradigms (de Wit et al., 2009, 2014; Eryilmaz et al., 2017) has demonstrated how tests of goal-directed and habitual actions can be translated from animal models to human behaviour. While such paradigms have been used in clinical settings to investigate drug addiction (Hogarth and Chase, 2011), obsessive compulsive disorder (Gillan et al., 2011), autism spectrum disorder (Geurts and de Wit, 2014; Alvares et al., 2016), schizophrenia (Morris et al., 2015) and Parkinson’s disease (de Wit et al., 2011), this translational approach is yet to be applied in frontotemporal dementia.
The clinical phenotype of behavioural-variant frontotemporal dementia
The behavioural variant is the most common clinical variant of frontotemporal dementia (Coyle-Gilchrist et al., 2016). Patients with bvFTD present with progressive deterioration in social behaviour and personal conduct. Clinical diagnostic criteria for bvFTD (Rascovsky et al., 2011) describe six core symptoms, including: (i) behavioural disinhibition (e.g. socially inappropriate behaviour, loss of manners or decorum and impulsive behaviour); (ii) early apathy or inertia; (iii) early loss of sympathy or empathy; (iv) early perseverative, stereotyped or ritualistic behaviour; (v) hyperorality and dietary changes; and (vi) executive dysfunction. At least three of these six features are necessary for diagnosis, though the presence and severity of each symptom may vary across patients and disease stages (Banks and Weintraub, 2008; Mioshi et al., 2010). Considerable pathological heterogeneity also exists, such that those presenting with bvFTD have an almost equal chance of having underlying tau or TAR DNA-binding protein 43 (TDP-43) pathology, with a small proportion of cases having fused in sarcoma (FUS) pathology (Chare et al., 2014; Mackenzie and Neumann, 2016). However, a number of patients presenting with clinical features of bvFTD reportedly show underlying Alzheimer’s disease pathology (Forman et al., 2006; Chare et al., 2014). While up to 40% of patients have a positive family history of dementia, pathogenic gene mutations––including microtubule-associated protein tau (MAPT), granulin (GRN) and C9orf72––account for only 10–20% of cases (Wood et al., 2013; Po et al., 2014). Despite this pathological and genetic heterogeneity, however, anatomical profiles of brain atrophy appear to be relatively similar across tau and TDP-43 subtypes of FTD (Harper et al., 2017). Typically, atrophy is present in the frontal and/or anterior temporal cortices (Rascovsky et al., 2011), as well as anterior insular (Seeley, 2010) and striatal regions (Halabi et al., 2013). Of relevance here, these brain regions overlap with those implicated in goal-directed behaviour in rodent studies (Fig. 2B). Here, we propose that many of the clinical symptoms of bvFTD can be conceptualized as a deficiency in goal-directed behaviour, with a concomitant increase in habitual behaviour (Fig. 3).
New framework for conceptualizing symptoms of bvFTD in terms of goal-directed and habitual behaviour. Goal-directed behaviours must satisfy the ‘belief plus desire’ criteria, whereas habitual behaviours are insensitive to changes in action-outcome contingency and/or changes in outcome value. Altered goal-directed behaviour for primary and secondary rewards may underlie symptoms such as overeating, hypo- or hypersexuality, overspending or risky financial decision making, reduced empathy, socially-appropriate behaviour and apathy. As the capacity for goal-directed control of behaviour diminishes, habits emerge and may manifest as stereotypical or perseverative behaviours.
Goal-directed behaviour in behavioural-variant frontotemporal dementia
Disturbance in goal-directed behaviour in bvFTD may be due to altered sensitivity to reward or deficits in the ability to update actions according to changes in reward value. In bvFTD patients, increased pursuit of primary rewards––including food, drugs and sex––is related to atrophy in the putamen, globus pallidus, insular cortex and thalamus, which are typically implicated in reward-processing (Perry et al., 2014). Similarly, symptoms of overeating are associated with atrophy in right orbitofrontal-insular-striatal regions (Whitwell et al., 2007; Woolley et al., 2007). A wider network of regions has been implicated in increased caloric intake and preference for sweet foods, including the cingulate cortices, thalamus and cerebellum for the former, and bilateral OFC and right insular-striatal regions for the latter (Ahmed et al., 2016). Despite this involvement of OFC regions, which have been implicated in failures of inhibitory control (Hornberger et al., 2010), overeating does not appear to be simply related to disinhibition (Whitwell et al., 2007; Ahmed et al., 2016). In the context of goal-directed behaviour, overeating may reflect insensitivity to the decreasing reward value of food or an inability to flexibly integrate updated reward values with goal-directed actions. Furthermore, while decreased sex drive is commonly reported in bvFTD, a subset of patients present with hypersexuality and aberrant/unusual sexual behaviours (Mendez and Shapira, 2013; Ahmed et al., 2015). Although the causes for altered pursuit of sex in bvFTD are likely multifactorial, involving endocrine dysfunction as well as reduced empathy and increased apathy (Ahmed et al., 2015), the potential contribution of altered sensitivity to rewards or changes in rewards has not been explored.
Given the overlap in brain networks involved in primary and secondary rewards (Sescousse et al., 2013), it is unsurprising that patients with bvFTD also show changes in their pursuit of secondary rewards, such as money and social acceptance. Increased financial risk-taking behaviour is well documented using monetary reward-related decision-making tasks (e.g. Iowa Gambling Task, Cambridge Gambling Task), where patients tend to choose risky options that lead to monetary losses (Rahman et al., 1999; Torralva et al., 2009; Kloeters et al., 2013). However, as failure on these gambling tasks can also reflect broader executive deficits (Perry and Kramer, 2013), targeted measures of monetary reward sensitivity are necessary. In terms of everyday behaviour, patients with bvFTD often spend excessively (Chiong et al., 2014) and are highly susceptible to financial exploitation (Wong et al., 2017). Although cognitive factors including impulsivity, disinhibition, and executive dysfunction may contribute to overspending or risky financial behaviour, such behaviours may also be due to reduced sensitivity to the negative reward value of financial losses, or an inability to flexibly integrate changes in reward value with actions. As such, experimental paradigms that enable specific assessment of components of goal-directed behaviour will help disentangle the reward-related versus cognitive contributions to poor financial decision-making in bvFTD.
Social cognitive dysfunction is well established in bvFTD, with widespread impairments in theory of mind, emotion recognition, empathy and complex social reasoning (Lough et al., 2006; Bertoux et al., 2012; Kumfor and Piguet, 2012; Kumfor et al., 2013; Dermody et al., 2016; Melloni et al., 2016; O'Callaghan et al., 2016). Considering the multi-faceted nature of social cognition, it is plausible that social cognitive deficits in bvFTD are influenced by altered processing of social rewards, although this has been relatively unexplored. Using an Incentive Delay paradigm that required patients to respond to a target cue in order to gain or avoid social or monetary rewards or punishment, Perry et al. (2015) demonstrated greater sensitivity to monetary compared to social rewards. BvFTD patients also show reduced sensitivity to negative social outcomes in social scenarios (e.g. cutting in line at a movie theatre) (Grossman et al., 2010) and reduced prosocial behaviour (e.g. sharing resources when personal gain is not at stake, or giving money to less-fortunate others) (O'Callaghan et al., 2016; Sturm et al., 2017). Moreover, we recently demonstrated that although bvFTD patients may learn and remember aspects of socially rewarding/punishing information (e.g. whether a person reciprocates or violates trust), they do not appear to apply this knowledge to modulate their behaviour (Wong et al., 2017). Of relevance here, failure to modify behaviour within specific social contexts has been proposed to contribute to social cognitive deficits in bvFTD (Ibanez and Manes, 2012). Whether this extends to encompass a failure to modify behaviour in relation to social rewards, however, remains to be established. As such, bvFTD patients may fail to modulate behaviour in order to avoid negative social consequences, or may not be motivated by rewarding social outcomes—an area which warrants further investigation.
Collectively, altered sensitivity to primary and/or secondary rewards, or deficits in the ability to modify behaviour in accordance with changes in reward value, may contribute to reduced motivation in bvFTD. Although apathy is one of the most prevalent and disabling non-cognitive symptoms in bvFTD (Merrilees et al., 2013), limited research has systematically explored this symptom in light of current theories of goal-directed behaviour. One study used the Philadelphia Apathy Computerised Test (PACT), a computerized reaction time task, to examine initiation, planning and motivation (Massimo, 2015). Impaired performance across all three components of the task was associated with widespread OFC, lateral prefrontal cortex and cingulate atrophy. However, the PACT does not allow assessment of the ‘belief plus desire’ criterion for goal-directed behaviour. Similarly, although a recent study by Perry et al. (2017) demonstrated reduced subjective and physiological responses to aversive olfactory stimuli, their behavioural paradigm did not distinguish between pursuit of rewards in valued versus devalued conditions. Thus, the relationship between goal-directed behaviour and symptoms of apathy requires further elucidation.
While many symptoms of bvFTD may be underpinned by deficiencies in goal-directed behaviour, we propose that a concomitant increase in habitual behaviour may underlie symptoms of perseverative, stereotyped or compulsive/ritualistic behaviours (Rascovsky et al., 2011). Examples of such behaviour include preoccupation with counting and/or clock watching, restricted interests in leisure activities or hobbies and rigid adherence to routines or food preferences (Nyatsanza et al., 2003). While it has been suggested that stereotypical behaviour, overeating and disinhibition are underpinned by changes to the same OFC-insular-amygdala-striatal network, (Ames et al., 1994; Nyatsanza et al., 2003), this requires confirmation through structural and functional neuroimaging studies that can distinguish between goal-directed and habitual behaviour. Specifically, insensitivity to the ‘belief plus desire’ criteria may give rise to habits that continue in circumstances where such behaviours are not optimal. While this is yet to be examined in bvFTD, evidence in autism spectrum disorder, which is also characterized by stereotypical or restricted patterns of behaviour, supports this hypothesis (Alvares et al., 2016). In this study, individuals with autism spectrum disorder showed impaired performance on an outcome devaluation test, such that they continued to perform the action associated with the devalued outcome. The authors suggested that symptoms of rigidity and stereotypic behaviour may be underpinned by reduced flexibility in integrating outcome values with actions, thereby rendering actions more habitual. In the same vein, it is possible that reduced flexibility in updating actions in accordance with value-related information may also contribute to symptoms of perseverative, stereotyped or compulsive/ritualistic behaviour in bvFTD, a hypothesis that warrants future investigation.
Challenges and future directions
A well established body of evidence from the animal literature has developed targeted experimental paradigms and identified the neural circuitry that mediates specific components of goal-directed behaviour. The translation of this approach to investigating the component processes and neural bases of symptoms of abnormal behaviour in bvFTD offers a new perspective on these symptoms (Fig. 3), yet a number of key issues are yet to be systematically explored.
Whether bvFTD patients are sensitive to changes in the relationship between actions and their outcomes is yet to be established. Current evidence indicates that bvFTD patients perform poorly on tests of reversal learning, where two previously learned action-outcome contingencies are reversed, such that the previously rewarded action is now punished, and vice versa (Rahman et al., 1999; Bertoux et al., 2013). Nonetheless, reversal learning paradigms do not distinguish between inflexibility in modulating actions in relation to the ‘value’ of their outcomes versus general cognitive inflexibility. As such, it is unclear whether similar deficits would be observed if only one of two previously learned action-outcome contingencies were selectively degraded, rather than reversed. Adapting a contingency degradation paradigm would, therefore, clarify whether bvFTD patients are able to selectively reduce performance of an action when they learn that it is no longer causal. Similarly, the extent that bvFTD patients are sensitive to changes in the value of outcomes remains to be established through using an outcome devaluation paradigm.
Whether changes in goal-directed behaviour in bvFTD vary according to sensitivity to different types of reward (e.g. primary versus secondary) is also only beginning to be explored. Specifically, the question of why patients may become hypersensitive to some rewards, but hyposensitive to others, warrants investigation. For example, while the majority of patients show symptoms of overeating, this may not necessarily occur in 100% of cases, nor does increased sex drive or increased propensity to make financial decisions. Indeed, some patients show more rigid eating behaviour (Shinagawa et al., 2009; Ahmed et al., 2016), hyposexuality (Ahmed et al., 2015) or minimal deficits in financial decision-making (Chiong et al., 2014). Such changes are likely to be relative, rather than an absolute preference for primary versus secondary rewards, and may reflect the clinical and pathological heterogeneity of bvFTD, or an exacerbation of pre-existing individual differences in sensitivity to different types of reward. Assessing the relative value of different reward types, and their impact on goal-directed behaviour in these patients, via translation of animal paradigms, will help inform the development of therapeutic interventions that target reward-related symptoms in bvFTD, and give important insights into how primary versus secondary rewards are represented at a neurobiological level. From a theoretical standpoint, future studies that contrast goal-directed and habitual behaviour across different neurodegenerative and neuropsychiatric disorders associated with frontostriatal deficits may also help identify transdiagnostic intervention approaches.
The evidence reviewed above provides support for a new conceptualization of the apparently diverse symptoms of bvFTD as a failure in a common underlying mechanism subserving the capacity for goal-directed behaviour. This conceptualization represents a new opportunity for developing treatments to address these challenging symptoms and for informing appropriate paradigms for animal disease models of bvFTD. First, by understanding the common mechanisms that underlie different symptoms, pharmacological interventions may be trialled and in turn appropriately evaluated. Of relevance here, modulation of striatal dopaminergic pathways has proved effective in Parkinson’s disease (for a review, see Chaudhuri et al., 2006). In bvFTD, striatal atrophy (Halabi et al., 2013) and dopamine deficiencies (Rinne et al., 2002) have been reported. Preliminary results suggest that dopamine modulation may benefit symptoms of apathy, disinhibition and risk-taking behaviour (Rahman et al., 2006; Huey et al., 2008; Lin et al., 2016). Nonetheless, larger randomized, placebo-controlled studies will be necessary to determine the efficacy of dopamine therapy for treatment of behavioural symptoms in bvFTD (for a review, see Tsai and Boxer, 2014). Moreover, behavioural interventions that apply principles of goal-directed behaviour and differential reinforcement, such as the Antecedent-Behaviour-Consequence (ABC) model of behaviour management (Logsdon et al., 2007), have been proposed for management of behavioural symptoms in bvFTD (Merrilees, 2007), but require validation through large-scale controlled studies. Secondly, whereas behavioural tests for animal models of Alzheimer’s disease have used hippocampally-dependent learning and memory tests, such tests are unlikely to be suitable for animal models of bvFTD (Roberson, 2012; Ahmed et al., 2017). The incorporation of behavioural tests of goal-directed behaviour, such as the contingency degradation and outcome devaluation tests, to clarify whether bvFTD patients are sensitive to the two criteria for goal-directed action [i.e. (i) sensitivity to action-outcome contingency; and (ii) sensitivity to changes in outcome value] may therefore prove fruitful in evaluating the efficacy of disease-modifying pharmacological treatments. This has important implications for animal drug trials, given the need for appropriate behavioural tests in enabling direct comparisons between animal models and clinical investigations of this disease. The translational approach outlined here, therefore represents an innovative research avenue to develop novel therapeutic interventions addressing the common factor of goal-directed behaviour and to assess the efficacy of such treatments simultaneously, from animal models through to clinical trials in humans.
Acknowledgements
The authors wish to thank Prof John R. Hodges, who provided helpful comments on a previous version of the manuscript, and Heidi Cartwright, who helped create the figures.
Funding
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 (NHMRC) (APP1037746) and the Australian Research Council (ARC) Centre of Excellence in Cognition and its Disorders Memory Program (CE11000102). In addition, this study was supported by NHMRC Project Grant (APP1121791). F.K. is supported by an NHMRC-ARC Dementia Research Development Fellowship (APP1097026). B.W.B. is a NHMRC Senior Principal Research Fellow (GNT1079561).
