Constrained chaos in three-module neural network enables to execute multiple tasks simultaneously

•Computer experiments of simultaneous multiple-tasks execution using three-module recurrent neural networks are reported.•Chaos is introduced by reducing connectivity between inter- and intra-module neurons.•Sensitive responses to simultaneous independent multi-inputs applied to fragments of sensory-neurons are observed.•Outputs from motor module drive multi-actuators simultaneously corresponding to multi-inputs.•Constrained (weak) chaos have superposing-like multi-functional activities despite consisting of a large number of highly nonlinear neural elements.