A growing field in Human-Robot Interaction aims at social assistance for users on specific tasks. These applications allow for insights regarding the acceptance of the robot’s presence and interaction-related performance effects. We present a scenario in which a socially assistive robot assists users on a cognitive task. Furthermore, we quantitatively evaluate the effects of two distinct interaction strategies on performance of the user and acceptance of the robot’s presence. In one strategy, the robot acts as a structuring guide and in the other, the robot tries to individually enhance the performance of the user. Results show that users benefit from a suited interaction strategy in terms of test performance and that the robot’s presence is regarded as acceptable and also desirable.