Supervisory methods in shared control allow a dynamic adjustment of the level of autonomy in a surgical robot based on the current task demands and the capabilities of the human operator. Several benchmarks tasks are available to evaluate the performance of these controllers, however the operator can struggle with the task due to inexperience or limited environmental information. In this paper, we propose an admittance control strategy based on guidance priority adaptation to enable a human operator to assume a supervisory role during one-handed peg transfer task. We implement an epsilon-greedy maximum entropy inverse reinforcement learning (EG MaxEnt IRL) algorithm to enable an agent to control the surgical tool in a virtual environment while the human supervises the procedure. We successfully implement the proposed method and observe that the supervisory method can be further improved with a cooperative control, specifically a segmented control.