A robotic agent equipped with a global workspace theory architecture and an episodic memory component, inspired by psychology and neuroscience, demonstrated static, temporal, and context memory capabilities during real-world interactions. Consciousness, as modeled in the system, participated in forming, maintaining, and retrieving episodic memories. The work argues that integrating episodic memory within a consciousness framework holds sustainable potential for developing artificial general intelligence. It also summarizes key features for modeling episodic memory in cognitive architectures and discusses how aligning machine consciousness with other AGI research could advance the creation of cognitive machines.
A model integrating global workspace theory with an attention mechanism is proposed as a step toward machine consciousness. In simulations, the agent shifted attention among multiple stimuli, reflecting the dynamics of conscious content. It also reproduced attentional blink and lag-1 sparing, two well-known human attention effects, suggesting compatibility with human cognitive processing. The model uses separate workspace nodes to reduce computation while enabling global availability, embeds attention as a competition mechanism for conscious access, and includes a synchronization mechanism that preserves the lag-1 sparing effect while retaining the attentional blink effect. This framework provides a foundation for future work in artificial consciousness.