Does the cost function of human motor control depend on the internal metabolic state?

Optimal feedback control theory (OFCT) has been verysuccessful in explaining human motor coordination in aprincipled manner [1,2]. OFCT derives motor controlpolicies from the minimization of a cost function, whichpredicts a large variety of movement data [3]. However,very little is known about the nature of this cost func-tion. From the many proposed costs (which take thesame mathematical form) two stand out as being moti-vated in a principled manner: noise [4,5] and“effort”(incl. muscle fatigue and metabolic demand [6]). Thesetwo cost are evolutionarily sensible as they maximize anorganism’s fitness: increasing task-relevant precision andenergy efficiency. It was recently suggested [6] thatnoise and effort are weighed against each other whendetermining motor coordination. However, the neuronalimplementation or representation of such costs in thebrain remains unclear [7]. We test the hypothesis thatthis trade-off may be directly affected by our internalmetabolic state (e.g. blood glucose level). We test thehypothesis if a subject’s internal metabolic state has animpact on the strategy for motor control. Specifically,low metabolic levels could bias motor coordinationacross redundant muscle groups from larger, metaboli-cally more costly muscles towards smaller muscles toincrease efficiency.We performed preliminary experiments by conductingreaching experiments under a dietary regime. 5 right-handed participants 21-27 years old performed reachingmovements in a virtual reality arm movement-trackingrig (visual stimuli were projected via a mirror systemonto the plane of hand movement). Air sleds ona surrounding table supported the participant’sarmtoallow frictionless movement. The task began whenparticipants moved their hand to a visual workspacecentre; then a 1.5cm radius target sphere appeared15cm away in one of 8 directions (total of 800 trials,randomly ordered directions). Subjects chose when tostart reaching towards the target, having to come to astop within the target sphere within movement dura-tions of 75 to 125ms. Feedback was given in the form ofa score that increased for successful trials and decreasedconversely. Each experiment involved two morning ses-sions on separate days. 3 subjects followed their normaleating/drinking routine for the first session then fastedfrom 8.00pm the evening before the second session andvice versa for the 2 other subjects. Blood glucose mea-surements were taken before and after each sessionusing a personal blood glucose monitoring system.Our preliminary data suggests a systematic shift in thecoordination of muscle groups acting on each jointbased on available energy levels. We found statisticallysignificant changes in shoulder and elbow joint utilisa-tion (integrated absolute change in joint angle) in all 5subjects and in up to 5 out of 8 target directionsdepending on the metabolic state. Moreover, we canmodel the subject’s reaching trajectories under bothmetabolic conditions using OFCT by a change in theweight of the metabolic cost term as a function of inter-nal metabolic state. Our preliminary findings are impor-tant as we link for the first time metabolic state toneuronal computations underlying motor control.