Fady SK. Alnajjar∗1 and Shingo Shimoda∗1 

∗1 BSI-Toyota Collaboration Center, RIKEN, Nagoya, Japan shimoda@brain.riken.jp 

Role of input and output synergies in biological controllers 

The notion of muscle synergy, which is able to create the muscle activation signals from the limited number of simple control signals, is widely recognized as one of the most important mechanisms to overcome the problem of controlling the redundant musculoskeletal systems[1][2]. As the opposite direction of the muscle synergy signal flow, however, simpli- fication of the sensor inputs to human controllers should be also very important to understand the environment and create behavior purposes to address the inputs. Bow-tie structure described in Fig. 1 is proposed to represents the essence of biological controllers where there are great diversity of inputs and outputs while a smaller diversity of protocol is used to connect the outputs and the inputs[3][4]. In the previous studies, we have formulated the input and output synergies and clarified the features how these synergies were changed during learning the behaviors[4][5]. The input and the output synergies can be described as the lower dimensional control spaces. We experimentally showed that the synergy spaces gradually shrink when we learn the reactive motions while the synergy spaces expanded when we learning the skilled behaviors. The synergy space analysis was useful to understand the recovery from post-stroke paralysis. We have shown that the behavior adaptations of the healthy subjects to the unfamiliar environment and the recovery from the motor paralysis of the post-stroke patients were the similar process from the viewpoint of the synergy space changes[6]. In the cases of the healthy subjects, the synergy spaces were once shrank to quickly react to the unexpected inputs in the unfamiliar environment. The experimental result showed that the synergy spaces were gradually expanded to create the skillful and efficient behaviors as the subject used to the environment. The damaged network of the neurons were the unfamiliar factors for the post-stroke patients. The synergy spaces were shrank after the disease and gradually expanded as was the case with the behavior adaptations of the healthy subjects. These results suggested that our brains can allocate the limited resource to the required tasks by tuning the synergy Fig. 1. Bow Tie Structure represents the essence of biological controllers where there are great diversity of inputs and outputs while a smaller diversity of protocol is used to connect the outputs and the inputs spaces. When we need the quick responses, smaller synergies space is used at the expense of behavior varieties. On the other hand, when skilled behaviors are required, we use larger synergy spaces using larger computational resources. We need the further analysis to clarify the process how our brain detect the appropriate size the synergy space in the feedback loop including the body, CNS and the environment without any supervising signals. 

[1] Y. Shinoda, Y. Sugiuchi, T. Futami, N. Ando, and T. Kawasaki, “Input patterns and pathways from six semicircular canals to motoneurons of neck muscles. I. The multifidus muscle group,” J. Neurophysiol, vol. 72, 1994. 
[2] A. D’Avella, P. Saltiel, and E. Bizzi, “Combinations of muscle synergies in the construction of a natural motor behavior,” Nature Neuroscience, vol. 6, pp. 300–308, 2003. 
[3] M. Scete and J. Doyle, “Bow ties, metabolism and disease,” Trends Biotechnol. 22 (9), pp. 446–450, 2004. 
[4] F. SK. Alnajjar, M. Itkonen, V. Berenz, M. Tournier, and S. Shimoda, “Sensor synergy as environmental input integration,” Frontier in Neuroscience, Vol. 15, Article 436, 2015. 
[5] F. SK. Alnajjar, T. Wojtara, H. Kimura, and S. Shimoda, “Muscle Synergy Space: Learning Model to Create an Optimal Muscle Synergy,” Frontier in Computational Neuroscience, 2013. 
[6] F. SK. Alnajjar, V. Berenz, Ken ichi Ozaki, Kensuke Ohno, and S. Shimoda, “Muscle Synergy Features in Behavioral Adaptation and Recovery,” The international conference on NeuroRehabilitation, 2014.