The concept of executive function, as top-down control of processes, originated in computer science in the 1950s. However, it has since become an important concept in a range of human sciences, particularly for its explanatory power in psychology, education, and clinical neurosciences. Nevertheless, its use has been limited by vague definitions and confusion between the related conceptualizations of executive process and intelligence. Here we explore the concept of executive control in detail, drawing on psychology, neurology, and computer science/human-machine interaction. We explore both computationalist and embodied cognition approaches. We describe the core goal-directed and resource-limited features of executive control, its fractionation into components, and partial overlap with psychometric conceptions of intelligence. We also examine its associations with neurological systems beyond those usually linked to executive function (i.e., the frontal lobes). We propose that executive functions are ‘intelligent’, and can be defined by their goal-directedness. Furthermore, executive function tasks can be classified by their task goals into one of three types: Those that involve i) convergent, or ii) divergent thinking, or iii) not responding, such as in psychomotor response inhibition. Conventional intelligence tests measure only convergent thinking. The recognition of non-convergent executive functions allows the identification of executively controlled intelligent goal-directed behavior beyond that controlled by domain-general cognitive processes. This reconceptualization may benefit research in education, clinical and cognitive sciences, as well as the quest for artificial general intelligence.