Forces applied to the pedals were used to generate a visual image to be used to train a group of cyclists to modify their pattern of force application. The system used included instrumented pedals mounted on a stationary bicycle frame, a minicomputer with a 12-bit analog-to-digital converter, and a graphics computer. The cyclists were instructed to modify their riding mechanics to minimize the forces applied to the pedals during a 90-degree segment of the recovery beginning at a crank angle of 225 degrees after top dead center. The cadence was 60 RPM and the average power output 112 watts. The subjects rode for 32 minutes each day for 10 days. During these training rides, the control group (n=4) was given feedback on their pedalling rate only, while the experimental group (n=4) was presented with feedback (KR) on their pattern of force application as well as pedalling rate. To prevent the experimental group from becoming dependent upon the visual feedback of the applied forces, the image was presented and withdrawn on a regular schedule. There was more feedback early in the training schedule than later. For each day of the training sessions, the mean force applied to the pedals was recorded. For the experimental group it was possible to partition the means into the intervals with and without feedback.

To assess the impact of the visual feedback on the cyclists' ability to alter their pedalling mechanics, the mean force applied to the pedals during the 90-degree segment was computed during a pre-test, a post-test immediately following the end of training (post-test 1), and a post-test seven days later (post-test 2). In addition, the pattern of force application for the whole pedalling cycle was computed. Only information on cadence was given to the groups during these tests.

The experimental group significantly (p<.05) reduced the recovery forces between the pre-test and post-test 1. There was no difference in mean pedal force between the post-test 1 and post-test 2. There were no significant between group differences. The experimental group reduced their pedal forces almost to the final values within the first day.

Three of the four subjects in the experimental group showed no loss in performance when the KR was removed. The fourth subject had considerable difficulty completing the task without the KR. The KR appeared to facilitate the acquisition of the new skill in the early stages. Because of the lack of difference between the mean forces computed in the intervals with and without feedback, it was speculated that the group used the KR initially to solve the task requirements. After that, the KR was used only as a check to ensure that there had not been a decrement in performance.

Analysis of the pattern of force application, through the complete cycle, showed that as a result of training sessions modifications of the recovery force resulted in a reduced propulsion force. This satisfied the requirements of the task that the pedal rate and power output remain constant. The reduced recovery forces implied that the subjects were not working against themselves as hard as before the training sessions began.

It was concluded that (a) performance of a complex task can be modified by feedback of a biomechanical nature, (b) visual feedback of pedal forces assisted the experimental group in achieving lower forces more rapidly than the control group, (c) the presentation and removal of feedback can assist subjects to form an internal representation of the task without including the augmented feedback, and (d) the reduction of the pedal forces in the recovery resulted in reduced pedal forces in propulsion. It was speculated that this reduction would result in an improved economy of riding.