The unbearable hardness of inferring being: Comment on "preliminaries to artificial consciousness" by Evers et al.

A commentary argues that a proposed multidimensional heuristic for structuring artificial consciousness research cannot determine whether phenomenal consciousness is nomologically possible in machines. Behavioral-cognitive profiles lack a justified principle linking function to experience, and the awareness case study shows that externally specified goals can produce as-if control rather than original intentionality. The heuristic overlooks that substrate similarity to the adult human brain is currently indispensable for justifiably inferring consciousness. Thus the framework provides a blueprint for a more sophisticated philosophical zombie but cannot tell whether anyone is there.

Study at a glance

Abstract

This commentary commends Evers et al.'s multidimensional heuristic for structuring artificial consciousness research while arguing it cannot, as stated, adjudicate the nomological possibility of phenomenal consciousness, which is at stake in current debates. Behavioral-cognitive "profiles" lack a justified principle linking function to experience, and the awareness case study illustrates how externally specified goals can just as well underwrite as-if (pseudo-intentional) control rather than original intentionality. Moreover, the proposed heuristic overlooks that substrate similarity is currently indispensable for justifiably inferring the presence of consciousness beyond the validated case of the adult human brain. Given all this, the framework seems to provide a blueprint for building a more sophisticated philosophical zombie; it does not-and cannot-tell us whether anyone is there.