Abstract

In this manuscript, we propose a classification of cognitive faculties based on the current macro-approach of grounded cognition. Within this framework, we situate the classic distinction between narrow and general intelligence, offering our conceptualisation of these categories. We argue that the difficulty of assessing intellectual abilities in the artificial domain cannot be properly addressed without first determining whether these abilities are grounded or ungrounded and understanding the implications of this distinction. In fact, we maintain that in the artificial sphere, ungrounded capabilities are not necessarily synonymous with a lack of intelligence. To account for this, we propose a disembodied conception of intelligence, one that is decoupled from basic cognitive abilities as well as from constructs such as subjectivity, selfhood, and mind. For these reasons, we also suggest conceiving of ungrounded intelligence as extended. Finally, we introduce what we call the Value Grounding Problem as a conceptual test for distinguishing between grounded and ungrounded cognitive capabilities in the artificial field.