BACKGRAUND: Violation of gait and body balance in patients with diabetic polyneuropathy actualizes the development of new effective measures of motor rehabilitation. In this regard, an observation was carried out in which biofeedback training was used for the reference reaction in the conditions of variable feedback parameters.AIMS: To investigate the effect of a stepwise change in the depth of feedback (sensitivity) in a motor-cognitive task with visual feedback on the reference response to the result of its execution.METHODS: Single-sample observational observation with controlled conditions. 27 patients with diabetic polyneuropathy. All patients were treated according to modern standards. The observation was performed in one series at the beginning of the course of inpatient treatment. Estimation of body balance on a power platform — stabiloplatform) — stabilometry. The procedure included 5 stages of management, in which the patient followed the instructions (training) in which there was a stepwise increase in the depth of feedback («sensitivity») at each stage from 10 to 50% of the conditional «normal», with a step of 10%. A quantitative assessment of the external result (execution of instructions) and parameters of regulation of the vertical posture was carried out.RESULTS: All patients included in the follow-up showed that the implementation of instructions with increasing depth of feedback decreased sharply when the conditionally «normal» value was exceeded by 30–35%. At the same time, the initial parameters of vertical posture control in the sample patients did not affect the dynamics of results when performing the task with visual feedback. Factor analysis indicates the presence of the only significant factor associated with the effectiveness of performing a motor-cognitive task here — changes in the depth of feedback. The accepted significance level α = 0.05.CONCLUSIONS: The depth of feedback in motor-cognitive tasks with visual feedback is a key characteristic that affects performance. The ability to increase the depth of feedback in the range of about 35; from conditionally «normal», can be useful for ensuring differentiation of loads during motor rehabilitation of patients in order to increase the effectiveness of training. In addition, in theoretical terms, the proposed model of the motor-cognitive problem can be relevant in the study of «dual» problems, where the relationship of motor and cognitive components is studied.