The Neural Mechanisms Underlying The Maintenance Of Visual Working Memory Contents

One of the branches of working memory is called visual working memory(v WM).Visual working memory provides an essential connection between perception and higher cognitive functions,representing the active maintenance of stimulus information that no longer in view.During past few decades,many researchers have been focusing on the studies of encoding,storage and maintenance of visual working memory due to the fact that it plays an important role in Human’s cognition.According to previous studies,we have a better understanding of visual working memory in it’s capacity,load effect and activation.The stimuli in past v WM experiments were mainly about one-feature simple object,such as coloured squires,motion orientation,grating orientation.There are some classical conclusions from past researches.Firstly,the increased number of items during encoding period resulted in heavy load effects and lower precision.Therefore,the theory of v WM capacity is limited has been widely accepted.Secondly,many cognitive neuroscience researches revealed that both the visual cortex including early visual cortex and higher brain areas including parietal lobe,temporal regions represented the maintenance of contents’ information,and these results were usually from the studies of visual working memory capacity.However,previous studies mainly focused on investigating the neural mechanisms of one-feature stimuli during v WM task,the neural mechanisms of complex stimuli such as face and house images information during maintenance period are still elusive.Furthermore,during the maintenance of v WM,whether we can predict the contents through activity amplitude and activity patterns of some specific brain areas is unclear.In addition,the load effects of v WM during maintenance period and it’s neural mechanisms are elusive too.Therefore,we used behavioral experiment and task-related function magnetic resonance imaging(f MRI)to investigate the load effect and neural mechanisms from the aspects of behavioral science,activity amplitude and activity patterns(decoding)underlying the maintenance of face and house information during v WM tasks.We hope that this study can apply some worthy evidence from the aspect of cognitive neural science for the visual working memory.Experiment 1: This is a behavioral design about visual working memory to instruct participants to memorize the face and house images information.The aims about this study are as follows.Firstly,we aim to detect the appropriate number of stimuli because the individual visual working memory capacity is limited according to the classical conclusions.However,the capacity of complex objects such as house and face images in this study is unclear.Secondly,we design the behavioral experiment to investigate whether the results are consistent with previous studies and accordance with our hypothesis.Only by doing this,it can guarantee the experiment design are available and scientific.Before the formal experiment,the experiment contained two factors and three levels,six conditions in total.Two factors included face information and house images information,three levels included load two,load four and load six.Finally,according to the result,we decided the experiment was 2*2 design.Two factors were face information and house information.Two levels were load two and load four during the encoding period.During the behavioral experiment,participants were instructed to memorize all the images information during the encoding period.After the representation of the cue,participants had to maintain the related information(face or house information)for 11 s.Next,the test probe stimulus would be represented on the screen,and volunteers had to compare the test probe with maintained information.If they were consistent,subjects needed to press ‘N’ on the keyboard by left index finger.If not,they had to press the ‘M’ on the keyboard by right index finger no more than 4.4s.The results suggested that the accuracy of high load conditions was significantly higher than high load conditions.However,the difference between face information and house information on the same load was not significant.These results were consistent with previous studies and accordance with our hypothesis(load effect).Therefore,we could continue our experiment to collect task-event f MRI data to investigate the neural mechanisms of visual working memory.Experiment 2: This experiment was to investigate the neural mechanisms underlying face/house information visual working memory.The first step was to localize the regions of interest(ROIs)including face selective regions,house selective regions and primary visual cortex.The second step was to collect f MRI data when conducting the visual working memory task.Moreover,we recorded the behavioral performance in the f MRI scanner and the results showed the same outcomes with previous study outside the scanner.The results showed that the accuracy of behavioral performances were higher in low load than high load both in face and house.We did a series of analysis about the f MRI data,and the activity analysis revealed that the activity amplitude of bilateral parahippocampal place area(PPA)and bilateral inferior parietal sculus(IPS)was significantly higher during house information maintenance than face information maintenance.However,the activity amplitude of bilateral fusiform face area(FFA),bilateral superior temporal sulcus(STS)and primary visual cortex(V1-V4)during the maintenance was not significantly different between face information and house information.The activity amplitude analysis suggested that we can predict the maintained contents by comparing the activity amplitude differences on PPA and IPS.However,we cannot predict memorized contents through the activity amplitude of FFA,STS and V1-V4.Therefore,we needed to explore the activity patterns about the latter regions to investigate whether their activity patterns could predict the memorized contents during maintenance period.The propriety of multi-voxle pattern analysis(MVPA)is to classify the spatial patterns of Bold signals from specific brain regions.So we could discriminate different cognitive status.If the classification is above 50%,then we can conclude it can classify different cognitive status.The results from the MVPA revealed that the activity patterns of FFA,STS and V4 in the high load conditions could discriminate the contents(face or house)during the maintenance period.According to the results from activity pattern analysis,although the activity amplitude was the same during face and house information maintenance on the same load conditions,the activity patterns of FFA,STS and V4 showed higher decoding accuracy on face information maintenance.On low load conditions,there were no significant decoding accuracy,and the reasons may be the increased difficulty and attention process resulted in the decoding accuracy on high load conditions.Furthermore,we continued to conduct the whole brain analysis,and the results showed that the activity amplitude of bilateral putamen was significantly negatively correlated with load effect.Under the same load conditions,the activity amplitude of putamen about face information and house information was not significantly different.The whole brain analysis results suggested that the load effect in this study was revealed by activity amplitude of bilateral putamen regardless of the stimuli.Altogether,These results suggested that the we can discriminate the contents during visual working memory maintenance period by the activity amplitude of PPA and IPS.However,we cannot discriminate the load effects by the activity amplitude of PPA and IPS.Furthermore,the areas V4,FFA and STS maintained the contents mainly through their activity patterns with the influence of the load effects.And the neural mechanisms of load effects in this study was suggested by the activity amplitude of bilateral putamen.In summary,this study provided the knowledge of brain’s functions in processing the complex contents during the maintenance of visual working memory,and revealed the different processing mechanisms of different brain regions in visual working memory.