The comanipulation paradigm, in which a user and a robot simultaneously hold a tool, allows for gesture guidance. In particular, virtual fixtures, which are geometrical constraints imposed to the tool by the robot, have received great interest in the domain of surgical applications. So far, this concept has been implemented in the context of open surgery. This paper explores the application of virtual fixtures for minimally invasive surgery, in which the tool is inserted in the patient through a fulcrum. Here, a key issue is to return to the surgeon forces that are virtually applied at the instrument distal tip, while the robot is physically attached to the instrument proximal handle. To this aim, two approaches are investigated. A first approach consists of applying a full wrench at the proximal end of the instrument that is equal to the wrench constituted by a pure force applied to the instrument's distal tip. A second approach consists of applying a pure force to the instrument proximal end, thanks to a lever model about the fulcrum. The two approaches are compared through experiments, during which naive subjects blindly perform virtual object palpa-tion and robot-guided movements. During experiments, indicators involving motion and force analysis are computed. The user capacity to distinguish between several virtual objects is evaluated as well. Although drastically different, the two approaches provide assistance with a similar level of efficiency. Index Terms-Force feedback, human-robot interaction, medical robotics.