Working Memory Capacity Influences Backward Inhibition: Analysis Of Spatiotemporal Patterns Of Event-related Potentials

BackgroundAfter an intermediate task,Mayr and Keele(2000) observed larger reaction times (RT) when returning to a task than when finishing other tasks in a row.For instance if there exist three tasks A,B,and C,the RT for the same third task(A) will be slower in the sequence of A-B-A than in the other case of C-B-A.It was referred as backward inhibition(BI).It may reflect active inhibition to the old task set to reduce its interference to the new.However,research on causal relationship between working memory capacity(WMC) and attentional inhibition prompted that certain link between them:higher WMC usually correspond to stronger attentional inhibition. Controversy exists about the occurrence of BI in the stage of whether cueing evaluation or response decision.In order to clarify when and where about BI occurrence,subjects with different WMCs completed tasks in different sequences which can produce BI effect while event-related potentials(ERP) recorded.MethodsTwenty-four college students were recruited,right-handed,without neurological and psychiatric disorders.Those without BI effect or too much artifact in ERP data would be rejected. Firstly applying a delayed match-to-sample task,eight digits were selected randomly from 0-9 digits to consist of a sample array,subjects were instructed to press left or right button to judge whether or not a probe digit later presented appearing in the previous sample array.According the performance,the Pashler-Cowan formula K=Stimulus×(Hit%- False-alarm%) was used to calculate WMC.Secondly subjects engaged three different tasks of digit characteristic judgment that is indicated by three symbols as cueing.After cueing,each subject had to judge target item - one of six Arabic numerals(2,3,4,7,8,and 9) - as prime or non-prime / odd or even / more or less than five by pressing left or right button.ABA-type of repetition sequence and CBA-type of non-repetition sequence were taken as test and control tasks correspondingly.Correction rate(%) and response time(RT) of the third task in both ABA-type and CBA-type(RTABA,RTCBA),ERP of 19 channels of 10-20 system during cueing processing(from -600 ms to 0 ms before target onset) and target processing(0-1400 ms after target onset) was analyzed,average activities during 1200-1400 ms as baseline.WMC,BI value(BI = RTABA-RTCBA) and the correct rate were compared with independent sample t-test separately using SPSS 13.0 software.Both RT and ERP were analyzed with hybrid two-way ANOVA(capacity as random factor,task sequence as repeated measurement factor).Statistical parameter mapping(SPM) of ERP was obtained after the interpolation with F-value or t-value.The significance levelαis 0.05.ResultFive subjects without BI effect and one with noisy ERP were removed from final analysis.And data from 18 subjects entered into statistical analysis.Each group of high and low WMC contains 9 subjects evenly.No significant difference of age was found between two groups except for their K-value(t(16)=7.252,P= 0.000＜0.001).No significant difference of correct rate was found(t(16)=0.738,P = 0.471＞0.05) either but the BI value(t(16)=2.345,P=0.032＜0.05),the high WMC group has larger BI effect.For RT analysis,no significant interaction effect of capacity and task sequence was found(F(1,16)=0.074,P = 0.789＞0.05).Main effect of task sequence was found(F(1,16)=14.859,P=0.001＜0.05) and none for main effect of capacity(F(1,16)= 1.467,P=0.243＞0.05).The SPM(F) of spatiotemporal pattern of ERP showed significant main effects of task sequence occurred in frontal and parietal regions during the period of-500 -400 ms(cue processing before target onset) and 450～500 ms;the main effects of capacity occurred in frontal and parietal lobes during the period of-200～150 ms and 600～950 ms;significant interaction effects of capacity and task sequence occurred during two stages with similar regions:-390～-370 ms(cue processing) and 250～290 ms(target processing) in frontal-parietal regions with different functions revealed by further SPM(t) analysis.ConclusionFronto-parietal(cingulate) network is the neural correlates of BI effect,which may overlap with supermodal regions of both working memory and attentional network.BI effect could occur in either cue evaluation and its retention or target evaluation and response preparation.High capacity individual has larger BI effect, which may result from two stages:cue evaluation(210～230 ms after cueing onset) and target evaluation(250～290 ms after target onset).