This paper proposes that neuropsychoanalysis offers a novel paradigm for artificial general intelligence (AGI). It argues that grounding AGI in affective neuroscience and neuropsychoanalysis, rather than solely cognitive science, would significantly advance the field. The research highlights the subcortical areas of the brain, emphasizing the centrality of emotions and affects in psychic life, unlike the current cortical-centric approach in AGI research. The paper outlines AGI system design principles based on the brain/mind model of Mark Solms and Jaak Panksepp, particularly Panksepp's seven affective systems, integrated through Judea Pearl's causal analysis. The conclusion posits that a subcortical AGI is the optimal solution for AI control.

Luca M. Possati