Development of a model for the study and measurement of consciousness in artificial cognitive systems based on the integrated information theory

Abstract

The study and measurement of consciousness in artificial cognitive systems have been subjects of great interest in the fields of artificial intelligence and neuroscience. This article introduces an innovative model based on the integrated information theory (IIT) aimed at addressing this fundamental issue. The IIT, proposed by Giulio Tononi, offers a robust theoretical approach to comprehend consciousness in terms of information and its integration. Our model focuses on identifying and quantifying key aspects of consciousness in artificial systems, including information complexity and the structure of artificial neural connections. We have developed a conceptual framework to assess the “integrity” of information within a system, enabling us to measure consciousness in terms of its capacity to integrate information coherently, setting it apart from other approaches. Through experiments and simulations conducted with a video game using the Go No Go task, we demonstrate how our model can be applied to artificial cognitive systems, allowing for the evaluation of their level of consciousness in different contexts. This approach carries significant implications for the advancement of more advanced and ethical artificial intelligence systems, as it provides a methodology for evaluating and comparing their level of consciousness. Ultimately, our work contributes to the progress in understanding and measuring consciousness in artificial systems, paving the way for future developments in artificial intelligence and computational neuroscience.