Sobana,
Thanks for the clarifications!
It seems to me that you have three explanatory variables: 1) Group, 2) Task, and 3) Time. Among the three, Group is a between-subjects (or subject-grouping) variable, while the other two are within-subject variables. This leads to a 2 x 2 x 2 mixed ANOVA.
It seems to me that factor 'Day' would be essentially the same thing as Group. The word 'covariate' is very popular in the neuroimaging community with the meaning of a variable that is of no interest to the investigator. However, that tends to obscure the modeling strategy. An explanatory variable is a variable in the model regardless the interest to the investigator. So, to make the situation more straightforward, I usually avoid using the word 'covariate' except when referring to a quantitative variable.
Lastly, I assume that you have two reaction time values (as well as accuracy) per subject (one for each task), and that's what I meant in my previous message. In that case, you would use 3dLME to handle such within-subject quantitative variables (covariates).
Gang