Çevrimiçi çalışma belleği eğitiminde transfer etkileri

Thus far, working memory training studies were always conducted in confined laboratory environments. This study aimed to carry out a working memory training procedure in a naturalistic online environment. After extensive piloting, online tests and a training task were developed using recent technologies such as HTML5 and KineticJS with cross-browser compatibility. Participants were randomly assigned either to the experimental group that received an adaptive training schedule or the control group that received a non-adaptive training schedule. To test the effects of the online training task and to measure near and far transfer, we developed the online versions of eight tasks forward digit span, backward digit span, letter span, alphabet span, Stroop inhibition task, block design, Raven's Standard Progressive Matrices (RSPM), and verbal arithmetic task. Participants took a pretests and a posttest in two sessions. The suggested training regimen was 15 minutes in each session, four sessions a week during 6 weeks (46 days in total). Our initial sample consisted of 2,002 adults aged between 18 and 58 (M = 28.28). In the training task, a number of images were shown on a 3 × 3 grid and participants were asked to choose the correct images among nine images and drag them into the correct initial position. In the group with rotation, the grid rotated 90 degrees to the left or right 50% of the time, which required a visual transformation of information. Our analyses showed that memory tasks, the arithmetic task, and the block design task had satisfactory test-retest reliabilities. Moreover, all the tests in this study showed appropriate clustering in the principal component analysis. The first component consisted of RTs in memory tasks, arithmetic and block design task. The second component had only accuracy measures in memory tasks. The third component had high component loadings from the Stroop task. The fourth component received the highest loadings from the RSPM and arithmetic accuracies. As for the training, in both groups we observed stabilization in maximum level scores after the fifth session. The training task had moderate correlations with backward digit span accuracy that measures storage and manipulation in working memory and arithmetic RT, which is a measure of fastness of response in verbal problem solving. Manipulation of information rather than passive storage was a better predictor of problem solving skills. At the end of the study, there was no difference between the experimental and control groups while we detected a significant difference between the contact (10 or more training sessions) and no-contact groups (less than 10 training sessions) in digit span differences and a marginally significant difference in letter span differences. Both of them were in the expected and desired direction. Although we had a low compliance level in our training, the training task has also a promising future as a measure of fluid intelligence, manipulation of information, and problem solving. We also studied the nature of attrition and compliance in our data including predictors of attrition. We provide future directions and suggestions about this fertile but untouched area of online testing and training of working memory. Keywords: Online working memory training, working memory testing, adaptive training, online cognitive testing, online cognitive training, practical applications in cognitive psychology.