What Should AGI Learn From AI & CogSci? Pei Wang 1 , Bas R. Steunebrink 2 , and Kristinn R. Th´ orisson 3 1 Temple University, Philadelphia PA 19122, USA. pei.wang@temple.edu 2 The Swiss AI Lab IDSIA, USI & SUPSI 3 Reykjavik University & Icelandic Institute for Intelligent Machines Abstract. While the fields of artificial intelligence (AI) and cognitive science (CogSci) both originated from a deep interest in the same phe- nomenon – intelligence – and both setting themselves high aims in their early days, each has since greatly narrowed its focus, and all but aban- doned their core subject for a more limited version of the phenomenon. The many non-obvious causes for this change over the decades are per- haps understandable, but they have significantly reduced the potential of both fields to impact our understanding of the fundamentals of intelli- gence – in the wild and in the laboratory. This position paper argues that researchers in the field of artificial general intelligence (AGI) should care- fully posit their research objectives and methodology to avoid repeating the same mistakes. 1 The Big Picture of Intelligence and Cognition Roughly speaking, artificial intelligence (AI) and cognitive science (CogSci) come from the same observation and imagination, namely that in a certain sense, the human mind and the electronic computer are – or can become – similar to each other. The similarities (and differences) have been suggested by many people, including Wiener [26], Turing [16], von Neumann [9], McCulloch and Pitt [7], though each from a different perspective. Initiated in this atmosphere, AI and CogSci can be seen as two sides of the same coin: while the former attempts to build a mind-like machine [11], the latter tries to study the mind as a machine [1]. Their relation is like that between engineering and science in general, that is, there is a strong mutual dependence. It is obvious that, to build an intelligent system, one has to have a clear idea about how intelligence works, and most of our knowledge on that topic comes from the study of the human mind. On the other hand, to evaluate the correctness of a theory of cognition, a straightforward way is to model it in an artifact to see if it produces the expected results. Given this relation, it is natural for AI to get inspiration from CogSci, as well as for CogSci to use AI models. Various theories have been proposed both to explain the phenomena observed in human cognition and to guide the design of machine intelligence (cf. [8, 10]). However, as the difficulties in this research became more and more clear, the mainstream in both fields gradually departed from the original objective to