The present study researched the systematic biases in working memory and how items interact with each other in working memory. The first goal of the study was to assess whether working memory representations of one another or systematically interact. This was tested by the repulsion bias in the representations. The second goal was to test whether the interaction is modulated by attentional priority. Attended items exhibited a weaker repulsion bias indicating that attention helped to protect the representation from the impact of the un-attended item.The average mean error for the unattended item was 3.68º while for the attended item it was 2.19º. These results are consistent with the hypothesis that items in working memory systematically interact with each other and further suggests that the main theories in working memory that do not assume interactions need to be updated.