How and to what end may consciousness contribute to action? Attributing properties of consciousness to an embodied, minimally cognitive artificial neural network

摘要: An artificial neural network called reaCog is described which based on a decentralized, reactive and embodied architecture developed to control non-trivial hexapod walking in an unpredictable environment (Walknet) while using insect-like navigation (Navinet). In reaCog, these basic networks are extended such way that the complete system, adopts capability of inventing new behaviors – via internal simulation planning ahead. This cognitive expansion enables system be enriched with additional procedures. Here, we focus question what extent properties phenomena characterized different level description as for example consciousness can found this minimally system. Adopting monist view, argue phenomenal aspect mental neglected when discussing function Under condition, discussed equipped bottom-up top-down attention, intentions, volition, some aspects Access Consciousness. These have not been explicitly implemented but emerge from cooperation between elements network. The Consciousness concern above mentioned ability plan ahead invent guide (new) actions. Furthermore, global accessibility memory elements, another characterizing realized by allows both reactive/automatic (access-) conscious behavior. We discuss examples interactions domain domain. Metacognition or Reflexive property reaCog. Possible expansions allow further Consciousness, verbal report states, Metacognition. summary, already simple if leaving aside.