martes, 28 de junio de 2016

Neuropsychopharmacology of Cognitive Flexibility


   Cools, R. (2015). Neuropsychopharmacology of Cognitive Flexibility. Social Cognitive Neuroscience, Cognitive Neuroscience, Clinical Brain Mapping,Academic Press,  3,  349–353, http://topics.sciencedirect.com/topics/page/Cognitive_flexibility; http://www.sciencedirect.com/science/article/pii/B9780123970251002530%20-%20s0010

Introduction


Cognitive flexibility is a broad term generally referring to our ability to adapt flexibly to our constantly changing environment. It is something that human animals are uniquely good at. We can persist with current behavioral strategies as long as these seem optimal for goal achievement, yet we can also update our strategies flexibly when the need for change has become sufficiently salient. How do our minds achieve this flexibility? This is not a straightforward issue, because only some of the changes around us are relevant and require cognitive flexibility. Most other changes are irrelevant (represent noise) and should be ignored. In the latter case, adaptive behavior depends on cognitive stability rather than cognitive flexibility. What we need is an ability to dynamically regulate the balance between cognitive flexibility and cognitive stability depending on current task demands. This trade-off between cognitive flexibility and stability has been studied in a variety of domains, ranging from reversal learning, to attentional set shifting, to task-set switching, to working memory updating.


Attentional Set Shifting


The classic neuropsychological test of attentional set shifting is the Wisconsin Card Sorting Test (WCST) (Grant & Berg, 1948). However, this paradigm has been criticized for its nonselectivity: adequate performance requires a wide range of abilities, includingworking memory, reversal learning, and task switching, and does not allow attribution of a deficit to either of those subcomponent processes. The intradimensional/extradimensional set-shifting task was employed to disentangle the subcomponent processes involved in the WCST (Downes et al., 1989; Slamecka, 1968). This visual discrimination learning task consists of several stages, where subjects move from a simple discrimination learning stage via a reversal learning stage and an intradimensional set-shifting stage to the critical extradimensional set-shifting stage. This critical extradimensional set-shifting stage measures the WCST-like ability to move attention away from one dimension (e.g., color) of a multidimensional stimulus to another dimension (e.g., shape) of that stimulus.
Adequate performance at this extradimensional stage depends on an intact dorsolateralprefrontal cortex (Dias, Robbins, & Roberts, 1996). In addition, it requires optimalcatecholamine levels in the prefrontal cortex, albeit in a surprising manner (Robbins & Roberts, 2007). Specifically, Roberts and colleagues revealed that dopamine lesions in the marmoset prefrontal cortex actually improved extradimensional set shifting (Roberts et al., 1994). In addition, they showed that this effect on extradimensional set shifting may well reflect effects on performance during the preceding set-maintenance stages of the task (Crofts et al., 2001). Specifically, a subsequent study revealed that prefrontaldopamine lesions impaired attentional set maintenance on the intradimensional set-shifting stages that preceded the critical extradimensional set-shifting stage. Thus, the effect of prefrontal dopamine depletion on set maintenance may well underlie the change measured in the subsequent extradimensional set-shifting stage of the task.
The trade-off between cognitive flexibility and stability in the domain of attentional set shifting may be related to the trade-off between exploration and exploitation (Daw, O'Doherty, Dayan, Seymour, & Dolan, 2006). Exploration generally refers to active cognitive search for new, potentially better alternatives, as is necessary during extradimensional set shifting, and has been proposed to depend on the prefrontal cortex (and its neuromodulation) (Aston-Jones & Cohen, 2005; Daw et al., 2006Frank, Doll, Oas-Terpstra, & Moreno, 2009). For example, Aston-Jones and Cohen had put forward the adaptive gain theory, according to which different (tonic and phasic) modes ofnoradrenaline transmission regulate the trade-off between exploitation and exploration. In this model, increasing tonic noradrenaline promotes behavioral disengagement, thus allowing potentially new and more rewarding behaviors to be explored. The transition from the phasic mode to the tonic mode is controlled by specific regions in the prefrontalcortex, including the orbitofrontal cortex and the anterior cingulate cortex, which in turn control the firing of noradrenaline neurons in the brain stem in a top-down manner. The hypothesis that (tonic) noradrenaline in the prefrontal cortex is particularly important for explorative modes of behavior concurs with empirical findings from work with experimental animals and humans showing that extradimensional set shifting is sensitive to manipulation of (tonic) noradrenaline transmission (Robbins & Roberts, 2007). This series of findings also raises the possibility that the effect of nonspecificcatecholamine modulation of extradimensional shifting reflects effects of prefrontalnoradrenaline rather than dopamine. Indeed, dopamine depletion in the striatum anddopaminergic medication in Parkinson's disease leave extradimensional set shifting unaltered (Collins, Wilkinson, Everitt, Robbins, & Roberts, 2000; Cools et al., 2001; Lewis, Slabosz, Robbins, Barker, & Owen, 2005).
The adaptive gain theory emphasizes the importance of noradrenaline for exploration and is complementary to a different influential proposal that tonic noradrenaline activity serves a neural interrupt or network reset function, thus enabling the revision of internal representations, based on new sensory input (Yu & Dayan, 2005). This model predictsnoradrenaline to be involved predominantly when changes in the environment are unexpected, as in the case of extradimensional set shifting. This is contrasted with expected uncertainty, which arises from known unreliability of predictive relationships within a familiar environment (Yu & Dayan, 2005). Critically, they argue that expected uncertainty is signaled by acetylcholine, a proposal that is consistent with observations that cholinergic changes are associated with attentional shifts in Posner-like attention-orienting paradigms where subjects are aware of cue invalidity (Hasselmo & Sarter, 2011). By contrast, cholinergic manipulations generally leave extradimensional shifting unaffected. Thus, according to these ideas, both increases in (tonic) noradrenaline andacetylcholine align attention with a source of sensory input. However, the signals that trigger this noradrenaline- and acetylcholine-mediated flexibility might differ. The theory is generally consistent with observed sensitivity of extradimensional shifting to cortical noradrenaline, but not acetylcholine.


Task-Set Switching


To measure cognitive flexibility outside the domain of learning, task-set switching paradigms are employed. In these paradigms, subjects switch repeatedly between two or more well-learned stimulus–response mappings or task sets. For example, subjects may be asked to name either the letter or the number of a compound letter–number stimulus depending on the color of the screen. Demands for learning and working memory are minimized because new task sets are instructed explicitly.
Unlike attentional set shifting, task-set switching is highly sensitive to dopaminergicmedication in Parkinson's disease (Cools et al., 2001) and healthy volunteers (Mehta, Goodyer, & Sahakian, 2004; Mehta, Manes, Magnolfi, Sahakian, & Robbins, 2004; Van Holstein et al., 2011). The deficit is maximal when there is high competition between task sets and when stimulus–response mappings are well established, as is the case with arrow–word (or letter/number naming) tasks. A Parkinsonian task-set switching deficit is not seen when switching to poorly established task sets (e.g., classifying digits as odd or even versus high or low) (Kehagia, Cools, Barker, & Robbins, 2009). This pattern parallels the observation that striatal dopamine depletion in marmosets impairs extradimensional set shifting back to a previously learned attentional set, but not to a novel attentional set (Collins et al., 2000). Thus, striatal dopamine might be important for cognitive flexibility, but only if demands for new learning are minimized and demands for selection are maximized.
This hypothesis concurs with the traditional view of the striatum as a selection or threshold-setting device and is in line with suggestions that dopamine signals mediate the switching of attention to unexpected, behaviorally relevant stimuli (Redgrave, Prescott, & Gurney, 1999). Furthermore, it is also consistent with empirical data showing effects of striatal dopamine manipulations on cognitive switching (Kellendonk et al., 2006).
One mechanism by which the striatum might alter task switching is by altering top-down control by prefrontal cortex. Specifically, the prefrontal cortex has been proposed to bias selective attention by regulating downstream regions in the posterior cortex (Desimone & Duncan, 1995; Miller & Cohen, 2001). The striatum might influence such top-down control via the direct go and indirect no-go basal ganglia pathways, by which it projects to the prefrontal cortex. Thus, the striatum could amplify task-relevant representations via the direct go pathway of the basal ganglia, while simultaneously inhibiting competing task-irrelevant representations via the indirect no-go pathway (Frank, 2005; Hazy, Frank, & O'reilly, 2007; Mink, 1996; Van Schouwenburg et al., 2013). Interestingly,dopamine has opposite effects on these two pathways, increasing activity in the direct pathway while suppressing activity in the indirect pathway. The net effect is a lowering of the threshold for a representation to be selected. Recent data concur with this hypothesis that dopamine might modulate task switching at the level of the striatum by modulating flow through the frontostriatal circuits (Nyberg et al., 2009; Stelzel, Fiebach, Cools, Tafazoli, & D'Esposito, 2013).


Working Memory


The trade-off between cognitive flexibility and stability in the domain of working memorycan be studied using tasks that require updating, such as n-back tasks or delayed-response tasks with intervening items. There is extensive empirical support for an important role of dopamine, in particular D1 receptor (D1R) stimulation in the prefrontalcortex in the stabilization of working memory representations (Goldman-Rakic, 1995). It should be noted that the stabilization of goal-relevant representations depends not only on dopamine but also on noradrenaline and acetylcholine transmission, possibly via modulation of attention and uncertainty signals, respectively. Effects of D1R stimulation on cognitive stabilization might reflect dopamine-induced increases in the signal-to-noise ratio of neuronal firing in the prefrontal cortex (Servan-Schreiber, Printz, & Cohen, 1990), leading to increased robustness of these representations in the face of intervening distractors (Durstewitz & Seamans, 2008; (Vijayraghavan, Wang, Birnbaum, Williams, & Arnsten, 2007).
The net effect of dopamine D1R stimulation in the prefrontal cortex is an elevation of the threshold for a new representation to be selected. Of course, this is adaptive when new input is irrelevant. However, it is maladaptive, when new input is relevant. In this case, existing goal-relevant representations need to be flexibly updated rather than protected. Accumulating evidence indicates that dopamine is also implicated in this complementary updating aspect of working memory. The proposal that the striatum is implicated in selective updating of working memory (Hazy et al., 2007) concurs with a rapidly growing body of data showing striatal involvement during updating of working memory representations (e.g., Bäckman et al., 2011). According to the biophysically realistic dual-state theory (Durstewitz & Seamans, 2008), prefrontal cortex networks can be either in a D1-dominated state, which is characterized by a high energy barrier favoring robust stabilization of representations, or in a D2-dominated state, which is characterized by a low-energy barrier favoring fast flexible shifting between representations. A role for prefrontal D2 receptors in cognitive flexibility is consistent with empirical evidence from work with rodents (Floresco et al., 2006, 2013).


Conclusion and Open Questions


The empirical data and theories reviewed in this article indicate that the balance between cognitive flexibility and stability depends critically on modulation by the major ascending neuromodulatory systems. Dopamine plays a critical role, but serotonin,noradrenaline, and acetylcholine are also important. An understanding of apparent discrepancies requires us to recognize that cognitive flexibility is not a unitary phenomenon, with distinct forms of flexibility implicating different cortical and subcortical neurochemical mechanisms (see also Klanker, Feenstra, & Denys, 2013).
It is clear that cognitive approaches to neurochemistry have revealed that dopamine,serotoninnoradrenaline, and acetylcholine likely serve rather specific functions in cognitive flexibility. This specificity arises partly from the different computations that are carried by the targeted regions, which differ in receptor distribution but likely also reflects a number of other factors. These factors include the computations carried by the brain structures that control the ascending systems in a top-down manner, the baseline dependency of the neuromodulatory effects and the (phasic versus tonic) timescale ofneurotransmitter effects.
Future work will benefit from adoption of a cognitive mechanistic approach toneurochemistry, which allows us to move beyond apparent discrepancies between theories of dopamineserotoninnoradrenaline, and acetylcholine in terms of cognitive control, attention, working memory, or learning. This is pertinent given the implication of most neuromodulators in all of these processes and will help to further define the computational nature of the flexibility–stability paradox. http://topics.sciencedirect.com/topics/page/Cognitive_flexibility

No hay comentarios:

Publicar un comentario