#基于意识的神经形态计算
Zenodo (CERN European Organization for Nuclear Research) September 3, 2026 DOI: 10.5281/zenodo.22273676 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Theoretical or philosophical paper Peer reviewed |
|---|---|
| Key points | Proposes that integrating consciousness as a fundamental design principle in neural architectures, through mechanisms like predictive coding and integrated information theory, could lead to computational systems exhibiting behaviors suggestive of understanding and nascent self-awareness. |
Abstract
This paper proposes a novel approach to neural computing, termed "Conscious-Based Neural Computing," which seeks to build computational systems exhibiting rudimentary forms of understanding and self-awareness by directly mimicking the underlying mechanisms of human consciousness. The core idea is to move beyond treating consciousness as a purely theoretical construct, and instead, integrate it as a fundamental design principle within neural architectures. This work outlines the key components of this approach, including a complex neural network structure designed to simulate core aspects of consciousness, such as predictive coding, integrated information theory, and contextual awareness. The system aims to generate activity through these mechanisms, ultimately leading to a computational system capable of exhibiting behaviors suggestive of "understanding" and a nascent form of "self-awareness." This approach represents a significant departure from traditional neural computing paradigms and offers a potential pathway towards truly intelligent systems. ---