We have developed a multi variate analysis of social communication behavior which allows for discriminating the emotional state of an agent interacting with other agents (Bouquet method). Domestic chick or common marmoset was reared under socially isolated conditions, later tested for its development of communication behavior with peers by Bouquet, and compared with that of animals reared under grouping conditions. We found the existence of high sensitive period for social interaction, given less experience, developing less affiliated behavior. From chick model, a couple of neuronal difference was observed between two groups, amygdale core central cell size, MRI volumetric measure in mesolimbic area, and gene expression patter including brain type tryptophan hydroxylase in nucleus accumbens. The isolated chick behavior changed better to be socially affiliated by taking SSRI/SNRI or Ubiquinol (the reduced form of CoQ10) together with social interaction experience after the sensitive period. Finally, we could discriminate the behavior of Asperger syndrome children (n = 7) from that of the typically developed siblings (n = 6) during the clinical interview by applying Bouquet method.