The data for the recall and forget treatments have been com- bined within each session. A comparison over time for recall/forget and the differences between the sessions can be seen in Fig. 5. Over the two sessions the averages for every item for both forget and recall conditions increased with the exception of the statement “the agent forgot you” and a small negative overall difference (total) from S2 and S3 for the forget condition which is contrasted with the largest positive overall difference for the remember conditions. Looking at Fig. 5, we should observe that averages over 2 indicate agreement and a positive response and averages below 2 indicate disagreement and a negative response. For all items (except forget) in the recall conditions for both sessions, the average response to each item is positive. In Session 3 when the type of forgetting was more natural, with less errors, the overall responses were (weakly) positive. Similarly, the studies concerning Eva (Kasap and Magnenat- Thalmann, 2012) found that interest in their system did not