| Cognitive flexibility is an important feature of cognitive control.It refers to our ability to properly and promptly adjust our own thoughts and behaviors based on the changing environment and internal states(Scott 1962,Armbruster,Ueltzh?ffer et al.2012).Cognitive flexibility plays an important role in our daily life,children's cognitive development,and pathological studies of some diseases such as obsessive-compulsive disorder.Although some mechanisms of cognitive control can promote cognitive flexibility,so far we still do not know how cognitive flexibility is achieved.Braver and some colleagues proposed the dual mechanism of control mechanism,namely,proactive control and reactive control.Their research suggests that cognitive flexibility can be achieved through flexible regulation of how different cognitive control mechanisms work for changing task requirements and internal states(Braver,Paxton et al.2009).For example,in some cases,proactive control may be more conducive to behavioral responses,while in others,reactive control may be more helpful for behavioral responses.In addition,one of the most concerned issues in the field of neuroscience research on the relationship between motivation and cognition is how the brain networks interact to promote the effect of reward on cognitive control.(Botvinick and Braver 2015).Although previous studies have found that reward motivation can promote the proactive mode of cognitive control,we still have little knowledge of the neural network underlying it.From the perspective of brain network,we can better understand the synchronization of activity patterns of different areas of the brain,to understand the dynamic relationship between these different brain regions,and insight into how these brain regions interact to promote different modes of cognitive control(Papo,Buldú et al.2014).We inferred that the fronto-parietal network(FPN)and/or the salience network(SN)might be involved in the incentive enhancement of proactive control,since their particularly important role in bridging the relationship between reward and cognitive control.Therefore,in the first study,we examined whether reward regulates the relationship between brain network activity and active control.We first repeated the previous studies,using the AX-Continuous Performance Test(AX-CPT)to study if the participants employed a proactive control strategy or reactive control strategy.We found that rewards could enhance the taking of a more proactive control strategy.Then,we decomposed brain networks of f MRI data by using independent component analysis,and calculated the correlation between the time series of these brain networks and task sequences.We found that there is a significant positive correlation between the time series of the salience network(SN)and the task sequence.In addition,it shows that the activation of the SN in the task has a significant correlation with the task performance.We then explored the relationships among rewards,task engagement of the SN,and proactive control.The results showed that reward could modulate the relationship between task engagement of the SN and proactive control,that is,the activation of SN is positively correlated with the behavioral index of proactive control under the reward condition,but the two have no correlation at baseline level.Finally,we examine the relationship among rewards,different types of trials,and the activation of SN.The results show that compared with the baseline condition,reward can enhance the activation of the network to the trials requiring proactive control,rather than the activation of the trials requiring reactive control.It may be the case that this neural modulation to the trials requiring proactive control support the enhancement of behavioral proactive control after reward.Our finding suggest that reward may moderate the relationship between task engagement of the SN and proactive control.These results show that the SN plays an important role in the reward facilitation effect of proactive control.This study may help to obtain a better understanding of the reward-related promotion of proactive control and the underlying network activity.With this study,we hope to improve our knowledge about the different aspects of cognitive control.On the other hand,the cued task-switching paradigm is a classic paradigm for studying cognitive flexibility.This task requires subjects to frequently switch between two or more tasks.In general,subjects have poorer performance on switch trials(longer response times and higher error rates)than repeat trials,which is called "switch cost." Previous studies found that frontal-partial and some other brain areas is involved in the switching process of the task.However,most of the predecessors used uni-voxel analysis method,which cannot speak directly to these regions' potential roles in representing and modifying task-sets.Studies found that different cognitive processes,such as the complexity of rules or the acquisition of skills,could affect the task representations(Jimura,Cazalis et al.2014,Woolgar,Afshar et al.2015).Moreover,it remains unclear whether and how task switch and task repeat could affect task representation.We propose two completely opposite hypotheses for this problem: the first hypothesis is that task switching reduces the flexibility of neural representation of different tasks compared to task repetition.This is because the switch cost caused by task switch on behavior performance may be caused by inflexible task representation(i.e.instability of new task representation or residual activation of the previous task)which may result in inflexible task representation at the time of task switching,and then lead to poorer performance in switching task(i.e.switch cost).In contrast,the second hypothesis is that task switching enhances the flexibility of task neural representation compared to task repetition.This is because,in order to overcome the interference caused by the previous task(i.e.,proactive inhibition),or in order to establish a state of readiness for a new task,stronger cognitive control is required,and stronger cognitive control may be manifested in more flexible neural representation of the task(Waskom et al.,2014;Alexandra Woolgar,Hampshire,et al.,2011).To test these hypotheses,in the second study,we investigated whether task switching and task repetition can modulate neural representation using the cued task-switching paradigm.In the process of data analysis,we first used the multi-voxel pattern analysis of functional magnetic resonance imaging data,to identify brain regions which can accurately distinguish(correctly classified)two kinds of tasks,the results include the frontoparietal and some stimulus processing brain regions.Then,based on these regions of interest,the classification and discrimination ability(classification accuracy and distance to hyperplane)of different task neural representations by task switching trials and task repetition trials are compared.The results show that,compared with the task repetition trials,the task switching trials has lower accuracy for different task classification,and the distance from the sample to the hyperplane is also closer.These results may indicate that the representation of individuals in task switching trials is less flexible.Although previous there have been some studies on the switch cost,and some explanations have also been given,it is still unclear to us how the brain can flexibly and dynamically encode tasks in order to facilitate or support flexible task transitions.Moreover,although a handful of recent studies employing such multivariate methods have shown that the currently relevant task-set can be decoded from activity patterns in frontoparietal and visual cortex,and that task distinctiveness in these areas can be enhanced by reward and training.One major reason for this is that the vast majority of functional magnetic resonance imaging(f MRI)studies delineating neural substrates of task switching have simply contrasted mean activity levels between switch and repeat cues or trials.These work has implicated lateral inferior frontal cortex,pre-supplementary motor area(pre-SMA),superior and inferior parietal cortex,and the striatum in switch processes(Dove et al.,2000;Sohn et al.,2000;Brass and von Cramon,2002,2004;Johnston et al.,2007;Braem et al.,2013;Ruge et al.,2013).However,mean activation contrasts cannot speak directly to these regions' potential roles in representing and modifying task-sets.For that purpose,multivariate,information-based brain mapping techniques,like multivoxel pattern analysis(MVPA)(Haynes and Rees,2006;Kriegeskorte et al.,2006)are required.A handful of recent studies employing such multivariate methods have shown that the currently relevant task-set can be decoded from activity patterns in frontoparietal and visual cortex(Woolgar et al.,2011;Waskom et al.,2014;Wisniewski et al.,2015),and that task distinctiveness in these areas can be enhanced by reward(Etzel et al.,2015)and training(Garner and Dux,2015).These studies suggest a key role for frontoparietal and stimulus processing regions in representing task-sets,but they do not characterize the role of these regions in the dynamic reconfiguration of task-set representations from trial-to-trial.Thus,in the third study,we also used the cued task-switching paradigm to study the frontal-partial areas' s flexible representation of different tasks,and the relationship between these neural representations and the behavioral flexibility.In the study,we firstly calculated the similarity of the neural pattern similarities of each adjacent two trials,which represent the similarities of neural representation when two task of the adjacent trials were performed.Because in task switching,the higher similarity between adjacent trials may mean stronger interference from the previous trial to the subsequent trial,and if this interference can be overcome(lower similarity),it may imply stronger flexibility in representation of the switching trials;When the task is repeated,the higher similarity between adjacent trials may imply that more information is retained from the previous trial to the subsequent trial,while more information is retained(higher similarity)may reflect that the repeated trials are more stable in neural representations.Therefore,we believe that this index may reflect the flexibility or stability of the individual's neural representation in performing the task.By paired-samples t-test with the switching conditon,we found that the pattern similarity of adjacent trials was higher in the repeat condition.Subsequently,we compared the pattern similarities of the second-order repetition and first-order repetition,as well as a comparative analysis of second-order switch and first-order switch.The results show that the pattern similarities of the second-order repetition is higher than that of the first-order repetition,and the second-order switch is more similar than the first-order switch.Next,we calculated the correlation coefficient between the pattern similarities and reaction times in the task repeat trials and the task switch trials,respectively.We found that there was a positive correlation between the pattern similarities and the behavioral performance in the repeat condition.On the contrary,there was a negative correlation between the pattern similarities and the behavioral performance in the switch condition.These results demonstrate that frontoparietal and stimulus processing regions provide the neural substrates of “dynamic adaptive coding”,flexibly representing changing task-sets in a trial-by-trial fashion.In addition,there are two different theories proposed by previous studies to account for the switch cost.The first theory is that the switch cost is caused by the interference from executing the previous task.In their view,although the previous trial or previous task was no longer relevant,its persistence or residual activation caused disruption to the current task.They suggest that the time required to overcome the interference caused switch cost.The other model is called task-sets reconfiguration,which propose that when task switching,task sets need to be reconfigured,and switch cost reflect the time needed to reconfigure the new task set.There is no clear conclusion about these two controversy so far.In the previous study,we examined whether the flexibility(stability)of neural representations of tasks in the frontoparietal brain regions can modulate individual level behavioral flexibility.However,we still do not understand how the flexible neural representation of brain regions such as the frontoparietal is achieved when task transitions(task repetition vs.task switching)are implemented.Is it similar to the switch cost of behavioral performance: a flexible neural representation is both conditioned by the task set inference and adjusted by the task set reorganization? Therefore,based on previous theories and the above studies,we proposed several corresponding models,and according to the statistical results of the data in the study,we verified which model is more in line with the actual situation.Specifically,in the fourth study,we used the pattern dissimilarity analysis of the f MRI data and the cued task-switching paradigm to study the influence of task sets interference and task sets reconfiguration on flexible neural representation when the task is switched or repeated.In order to validate the model and hypotheses proposed in the study,we calculated the pattern dissimilarities between every two types of the trials and compared these pattern dissimilarities.In addition,we explored the transform of task sets representation within each trial by representational dissimilarities,and compared the multi-voxel response patterns of the cue and target phase.The results showed that the pattern dissimilarities among four different types of trials were in line with our model and assumptions:(1)repeat task sets might have flexible neural representation than the switch task sets;(2)flexible neural coding may be affected not only by the task sets inference from the previous trial(task sets interference theory),but also by the task sets reconfiguration process of the current task(task sets reconfiguration theory);(3)At task transition,the neural task-set representation moves in representational space,away from representing the previous-trial set(due to task-set inertia)and towards representing the reconfigured,newly relevant set.Moreover,some researchers have pointed out that proactive control and reactive control may play an important role in task switching,and they could achieve a dynamic balance when performing the task.However,up to now,we still do not know the neural mechanism of top-down proactive control and bottom-up reactive control in task switching.Moreover,whether or not both proactive and reactive control can modulate flexible behavioral performance is unknown.In addition,we have found from the above studies that the frontoparietal and several stimulus processing brain regions may play an important role in task representation and cognitive flexibility.However,the above studies have not focused on the differences between the frontoparietal and stimulus processing brain regions,and how they may interact to promote flexible behavioral performance.Therefore,in the fifth study,we used dynamic causal model to examin how top-down proactive control and bottom-up reactive control are modulated by task switching and then act on flexible behavioral performance through the effective connection between the frontoparietal and the stimulus processing brain region.In the study,we first performed a conventional general linear model analysis,and compared the different brain activation during task transition(task switch vs.task repeat).The results showed that the frontoparietal regions have stronger activation in task switch than in task repeat condition,while some stimulus processing regions(visual region)have stronger activation in task repeat condition.Then,we established a series of models of these brain regions based on the analysis method of dynamic causal model,and verified which of these models is more consistent with our data.After that,the parameters in the final "winning" models(the most probable models)under the task repeat and task switch conditions were tested by one-sample t-test to verify whether the effective connectivity in these brain regions were significantly greater or less than zero.Then,we examined whether the effective connectivity between the frontoparietal and stimulus processing brain regions were modulated by task transition(task switch vs.task repeat).The results showed that the top-down connection between the posterior parietal cortex and the visual cortex was negative and significantly decreased under task switch compared with task repeat,while the bottom-up connection between the visual cortex and the frontal eye field was positive and significantly improved.Finally,we calculated the correlation coefficient between the effective connectivity that is significantly increased or decreased when task switching compared with task repeat and the behavioral switch cost.The results showed that the effective connectivity between the frontoparietal and visual regions influenced by task switch has a correlation with the behavioral switch cost.These results indicate that flexible cognitive performance may be affected by both top-down proactive control and bottom-up reactive control.In summary,this paper studied cognitive flexibility from two aspects: adjusting different cognitive control mechanisms(proactive control vs.reactive control)to achieve flexible cognitive performance,and the dynamic adaptive and flexible neural coding when task transformation.It is found in this study that reward can enhance proactive control,while flexible cognitive control may be influenced by both top-down proactive control and bottom-up reactive control.In addition,we also found that task transition could modulate the coding of different tasks in frontoparietal and stimulus processing regions.Moreover,the fexible neural represention for different tasks in these regions may provide the neural substrates of dynamic adaptive coding,which may mediate the flexible behavioral performance.Furthermore,the fexible neural represention in these regions may influenced by both task set inference and task set reconfiguration. |
PDF Full Text Request