تغییر بین قواعد انتزاعی نشان دهنده شدت بیماری و نه وضعیت دوپامینرژیک در بیماری پارکینسون

This study sought to disambiguate the impact of Parkinson's disease (PD) on cognitive control as indexed by task set switching, by addressing discrepancies in the literature pertaining to disease severity and paradigm heterogeneity. A task set is governed by a rule that determines how relevant stimuli (stimulus set) map onto specific responses (response set). Task set switching may entail reconfiguration in either or both of these components. Although previous studies have shown that PD patients are impaired at switching between stimuli, in the present study not all patients were impaired at switching entire task sets, that is, both stimulus and response sets: compared with controls, patients with unilateral signs (Hoehn & Yahr Stage I) demonstrated intact switching, even following withdrawal from dopaminergic medication, while bilaterally affected Stage II patients were impaired. The parametric measure of Unified Parkinson's Disease Rating Scale (UPDRS) score predicted increasing switch costs within the patient group. These findings suggest that switching entire task sets may be a function of extrastriatal, possibly non-dopaminergic pathology which increases as the disease progresses.

Parkinson's disease (PD) is a neurodegenerative disorder with a complex neurochemical profile in multiple brain regions and neurotransmitter systems. At its earliest stages, pathology is limited to degeneration of dopamine (DA) neurons in the ventrolateral tier of the substantia nigra pars compacta, which project to the putamen and rostrodorsal caudate nucleus in the dorsal striatum (Agid, Ruberg, Dubois, & Pillon, 1987; Kish, Shannak, & Hornykiewicz, 1988). In later stages, ventral striatum including the nucleus accumbens becomes DA-depleted and a parallel mesocortical DA deficit develops, affecting the prefrontal cortex (Dubois & Pillon, 1995), limbic system and hypothalamus. While a major emphasis has been placed on DA neurotransmission, especially in the context of cognitive deficits and medication ‘overdose’ (Cools, Barker, Sahakian, & Robbins, 2003; Swainson et al., 2000), gradual degeneration of the locus coeruleus, dorsal raphé and cholinergic brainstem nuclei progressively compromise the noradrenergic, serotoninergic and cholinergic systems (Braak et al., 2006; Brooks & Piccini, 2006). This complex neurodegenerative profile is associated with increasingly severe motor symptoms of tremor, muscular rigidity, bradykinesia and akinesia. Pronounced cognitive deficits are also seen on tasks of executive function sensitive to frontostriatal deficits, such as the Wisconsin Card Sorting Test (WCST), intra and extra-dimensional (ID/ED) shifting, Tower of London (TOL), Odd-Man-Out task and their variants (Bowen, Kamienny, Burns, & Yahr, 1975; Canavan et al., 1990; Channon, Jones, & Stephenson, 1993; Cools, 1984 and Downes et al., 1989; Gotham, Brown, & Marsden, 1988; Morris et al., 1988, Owen et al., 1992 and Owen et al., 1993; Richards, Cote, & Stern, 1993; Robbins, James, Owen, Lange, et al., 1994; Taylor, Saint-Cyr, & Lang, 1986). However, these tasks comprise multiple cognitive components, including concept formation, hypothesis testing, working memory, and stimulus selection, and deficits reflect impaired functioning on any, or more than one, cognitive process. In order to elucidate the nature of the parkinsonian cognitive deficit, task set switching investigations have focused on the shifting component of executive function (e.g., Rogers & Monsell, 1995), but have not converged on a robust deficit. Impairments have been reported in terms of inflated switch reaction times (RT) and error rate (Cools et al., 2001a and Cools et al., 2001b; Cools et al., 2003; Hayes, Davidson, Keele, & Rafal, 1998; Pollux, 2004 and Witt et al., 2006), or switch error rate but not RT (Brown & Marsden, 1988; Pollux & Robertson, 2002). For example, Cools et al. hold that PD switching deficits are a function of ‘cross-talk’ interference from irrelevant stimuli, and reflect DA dysfunction in dorsal corticostriatal loops, since performance is ameliorated by dopaminergic medication (Cools et al., 2001a and Cools et al., 2003). Other studies however fail to find a switching deficit (Fales, Vanek, & Knowlton, 2006; Rogers et al., 1998; Woodward, Bub, & Hunter, 2002). As such, consensus on whether PD causes executive deficits as measured by task set switching, and an accurate characterisation of the role of the basal ganglia, the associated corticostriatal loops and DA within these regions, in executive control, have yet to be realised. It is proposed here that these discrepancies may stem from (i) paradigm heterogeneity and (ii) the effects of disease severity. In order to compare task set switching paradigms, we address the two major elements of a task set: the stimulus set, which is the mental representation of target stimuli, and the response set, the representation of available responses (Meiran, 2000). The task rule signifies a particular cognitive operation and determines the mapping between stimulus and response set (e.g., the numerical parity rule determines that stimuli ‘2, 4, 6, 8’ map to the ‘even’ response, ‘1, 3, 7, 9’ map to ‘odd’). The reconfiguration in stimulus–response (S–R) mappings after adopting an alternative rule or cognitive operation (e.g., judge whether the number is greater or less than 5) is central to task switching, and may dictate that the same stimuli be associated with different responses (same stimulus set, different response set: e.g., ‘1’ now maps to ‘less than 5’ instead of ‘odd’), or that different stimuli be associated with different responses (different stimulus set, different response set: e.g., ‘X’ maps to ‘consonant’). Hence, the complexity of S–R reconfiguration determines not only the magnitude of the switch cost, but also its cognitive significance and neural basis. Neuroimaging evidence implicates lateral and posterior prefrontal cortical as well as parietal regions in the process of remapping stimuli and responses (Braver, Reynolds, & Donaldson, 2003; Dreher & Berman, 2002; Dreher, Koechlin, Ali, & Grafman, 2002; Forstmann, Brass, Koch, & von Cramon, 2006; Rushworth, Hadland, Paus, & Sipila, 2002; Wylie, Javitt, & Foxe, 2004; Yeung, Nystrom, Aronson, & Cohen, 2006). These findings are also consistent with neuropsychological evidence of switching deficits in frontal lesion patients (Aron, Monsell, Sahakian, & Robbins, 2004; Mayr, Diedrichsen, Ivry, & Keele, 2006). However, incorporating the PD findings into this framework is hampered by differences in the degree to which switching engenders a switch in cognitive operation and a reconfiguration of both stimulus and response sets. The studies in which switching entails reconfiguration in both stimulus and response sets report intact switching in PD. Combining the Stroop and task switching paradigms, Woodward et al. (2002) found abnormal PD switch costs only in the colour naming (Stroop) condition, which was attentionally the most demanding, but not in the word reading (reverse Stroop) condition, indicative of depleted attentional resources rather than deficient internal (task) control. Importantly, switches in this study entailed changes in both stimulus and response sets, as subjects attended to different aspects of the stimulus and gave a different response on a switch of task. Fales et al. (2006) addressed switching as a function of the recency with which a task set had previously been performed, using letter and digit classification tasks that relied on different cognitive operations and necessitated S–R reconfiguration on a switch. They found no overall PD switching deficits, but, instead, increased error rate limited to those switch trials where the current task had more recently been performed. This finding was interpreted as a specific deficit related to backward inhibition (automatic inhibition of the previously abandoned task set) but not task switching. Notably, that PD group also displayed intact performance on other tasks of executive function such as the WCST and TOL. In contrast, the paradigms that highlight PD deficits (Cools et al., 2001a, Cools et al., 2001b, Cools et al., 2003 and Witt et al., 2006) were adapted from an earlier study by Rogers et al. (1998), who employed letter and digit naming tasks. Switching in this design required the reconfiguration of stimulus sets only: once the task-relevant stimulus, number or letter, had been selected from the digit-letter compound, the superordinate task set after a switch was still a simple speeded vocalisation of the target's identity; the mappings between stimuli and responses remained unchanged. The PD switching deficit was isolated to the cross-talk condition: the task-relevant stimulus was presented along with a distracter associated with the alternative task set (e.g., ‘7G’). Compared with the no cross-talk condition (e.g., ‘7&’), where the distracter was a non-alphanumeric character not associated with either task set (hence easily ignored), the cross-talk manipulation increases the difficulty of switching task sets by increasing the difficulty of selecting the currently appropriate stimulus in the face of interference from the irrelevant character; attentional selection is required to overcome this interference. Pollux (2004) also utilised a paradigm where switching applied to the stimulus only and also found deficits as a function of ‘attentional conflict’. These studies suggest that DA neurotransmission in frontostriatal circuits may only affect stimulus set switching, which is primarily mediated by selective attention, but it remains unclear to what extent striatal DA affects the ability to reconfigure entire task sets, i.e., both stimulus and response sets, which has been associated with frontoparietal function. Hence, we sought to clarify the impact of PD and corticostriatal DA on S–R reconfiguration in a paradigm of switching between tasks governed by abstract rules. Despite its presumed striatal-cortical progression, which renders PD an informative disease model for studying the roles of different brain regions in executive control, the second issue of disease severity is noteworthy because studies of task switching have grouped together patients ranging widely in disease severity without considering, or taking advantage of, the neuropathological differences between patients at varying stages of disease progression and disability. Disease rating scales take into account the patient's functional status as well as overt motor signs. The Unified Parkinson's Disease Rating Scale (UPDRS) offers a continuous measure of disease severity. In this composite scale, where each item is rated 0 (normal) to 4 (severely affected), the primary focus is on parts II (13-item interview on activities of daily living) and III (14-item motor exam). Conversely, the categorical Hoehn & Yahr staging system (Hoehn & Yahr, 1967) offers a broader classification of patients on the basis of two main criteria: (i) unilateral versus bilateral signs and (ii) balance and gait difficulties. We argue here that disease severity is particularly relevant to investigations into the cognitive impact of a progressive neurodegenerative disease such as PD. As the disease progresses, it not only affects regions like the striatum to an increasing extent, but also encroaches on cortex, particularly in prefrontal and parietal areas. For example, at the earliest disease stage, pathology is generally limited to the substantia nigra and dorsal striatum: a [18F]-6-fluoro-l-dopa PET study showed that in a group of unilaterally affected Stage I patients, dopaminergic underactivity was relatively confined to putamen while caudate DA neurotransmission was normal (Nahmias, Garnett, Firnau, & Lang, 1985). In contrast, the more severe signs later on in the disease, which usually become bilateral and include postural and gait disturbance, reflect more diffuse pathology with greater striatal DA loss (Morrish, Sawle, & Brooks, 1996) as well as probable prefrontal cortical dysfunction (for review, see Brooks & Piccini, 2006), parietal cortical abnormalities ( Sabatini et al., 2000 and Samuel et al., 1997) and serotoninergic and noradrenergic neuron degeneration ( Wolters & Braak, 2006). Therefore, disease progression is a critical factor determining the cognitive profile of any given PD patient. Thus, the present study directly addressed (i) the impact of disease severity and increasing cortical dysfunction on task set switching when this entails switching between abstract rules and S–R reconfiguration, and (ii) the role of striatal DA neurotransmission, or the effects of dopaminergic medication on S–R reconfiguration. The effects of PD severity on task set switching with S–R reconfiguration were systematically investigated in two ways: first, using the categorical measure of Hoehn & Yahr stage, and specifically focusing on one of the two primary classification parameters, the transition from unilateral to bilateral signs, by comparing the performance of Stage I and Stage II–III patients. Second, by analysing the impact of the parametric and arguably more sensitive measure of total UPDRS score, the summed total of parts II and III (activities of daily living and motor score), obtained on the day of testing, on switching. At the earliest stages, the effects of PD on cognitive control may be said to represent the effects of a relatively limited asymmetric dorsal striatal DA lesion which is most pronounced on the contralateral side of the motor signs; any cortical DA deficits as a function of subcortical DA neurotransmission may be limited to the motor cortex, which the compromised putamen projects to. Frontal and prefrontal DA neurotransmission however can be assumed to be relatively normal, given that these cortical areas are reciprocally interconnected to the mostly intact caudate. At later stages, DA dysfunction is primarily a function of two factors: (1) the striatal DA deficit which encompasses the caudate and more ventral regions of the basal ganglia, and which reduces DA neurotransmission in striatocortical loops, and (2) a parallel mesocortical DA deficit. As discussed, noradrenergic and serotoninergic deficits may also become apparent. Thus, while it is impossible to unequivocally rule out frontal pathology as a function of subcortical DA deficits at the earliest disease stage, particularly given patient heterogeneity, it may be possible to assume that this is relatively limited compared to the deficits seen in more severely affected patients, which reflect more direct frontal or other cortical DA abnormality, as well as non-dopaminergic pathology. Since S–R reconfiguration requires intact frontal functioning, more severely affected patients were predicted to exhibit S–R reconfiguration deficits. We also investigated the role of DA in task set switching. Whilst it has previously been shown that dopaminergic medication ameliorates cross-talk deficits, which we have argued arise from the need to switch stimulus sets, it is unclear whether it would also enhance switching between abstract rules and reconfiguring S–R mappings. The effects of dopaminergic withdrawal in Stage I patients will help clarify the basis of any disease severity findings: (i) if, similarly to switching between stimuli (or stimulus sets), switching between abstract rules (which entails switching both stimulus and response sets) relies on DA neurotransmission in striatal-PFC loops, withdrawal should inflate Stage I switch costs. (ii) If S–R reconfiguration relies on DA neurotransmission at the level of cortex, namely in frontal and parietal areas, then medication should in fact ‘overdose’ the theoretically intact Stage I cortical DA systems (e.g., Swainson et al., 2000), and reduce switch costs, i.e., improve switching, in the ‘off’ state. Such a finding would also suggest that Stage II S–R reconfiguration deficits reflect frontal DA dysfunction. (iii) If S–R reconfiguration depends on frontoparietal circuitry but not on DA, then the manipulation should have no effects on this type of switching. This would also suggest that deficits seen in more progressed patients likely originate in the non-dopaminergic pathology that emerges as the disease progresses. In addition to investigating the association between disease severity, striatal DA neurotransmission and switching deficits, an attentional manipulation was undertaken to test a prediction that follows from the cognitive significance assigned to cross-talk deficits. As discussed previously, cross-talk refers to interference between task sets stemming from the nature of the presented stimuli, typically a task-relevant character and a task-irrelevant distracter. Switching to the currently relevant task set requires overcoming interference from the alternative task set which the irrelevant character effectively primes. The mechanism by which this interference is said to be overcome is attentional selection, which presumably operates on the compound of task-relevant and task-irrelevant characters, boosting the representation of the relevant one and inhibiting the other. It has been argued that increased attentional selection load is critical in exposing switching deficits in cross-talk studies; in other words, that PD is associated with task set switching deficits when the additional process of attentional selection is required (e.g., Cools et al., 2001a, Cools et al., 2001b and Witt et al., 2006). We tested this hypothesis directly. Switching in a compound stimulus condition (e.g., ‘7G’), where the compound of target and distracter is associated with both tasks and hence requires both attentional selection and S–R reconfiguration, was compared to switching with a single, unitary stimulus (e.g., ‘7’) associated with both tasks which requires S–R reconfiguration but not attentional selection. If attentional selection is indeed a critical cognitive process which necessarily interacts with switching to expose the PD deficit, then switch costs with compound as opposed to unitary stimuli should be greater in PD compared with controls, particularly following dopaminergic withdrawal.