Toward daydreaming machines

Abstract

This paper provides some insights related to building a working computational model of human-level mind. We propose to take a fresh look at some ideas propounded more than a century ago by William James and Sigmund Freud, which were recently reconsidered by Peter Naur and ATR Brain-Building Group, respectively. Naur proposes his Synapse-State Theory of Human Mind (SST), while the research at ATR resulted in the Machine Psychodynamic paradigm (MΨD). We argue that SST and MΨD propose complementary ideas about implementation of mental functionalities, including those related to the quest of consciousness. The 20th-century AI gave machine the ability to learn. The great challenge in the 21thcentury AI is to make a robot actually want to learn. MΨD proposes a solution based on pleasure defined as a measurable quantity to be used as a general reinforcement. SST proposes a neuroscience-inspired architecture, where the key blocks are item-nodes, attention-node, and specious-present excitation. MΨD may supplement SST with a pleasure node and related Pleasure Principle.