This talk focuses on the indispensable role of representation in
constituting intelligence. Central to the inquiry into intelligence
is understanding the relationship between mind, world, and action. I
argue, first, that even basic embodied intelligence, often cited as
evidence against representation, relies fundamentally on contentful
states. Drawing on analyses of intentionality and procedural memory,
skills are shown to possess directive content (a world-to-mind
direction of fit) defined by satisfaction conditions. This
representational layer is crucial for explaining the normativity,
learning, and potential failures (e.g., apraxia) inherent in skilled
action. Second, I argue that directive content alone is
insufficient. Genuine intelligence requires the capacity for
descriptive representation—states with a mind-to-world direction of
fit, capable of being true or false in a way that corresponds to
reality. This commitment to representational realism and truth is
not merely a feature of high-level cognition but a foundational
requirement for systems that can learn about, understand, and
flexibly adapt to the complexities of the world beyond immediate
interaction loops. Intelligence, therefore, is inextricably linked
to the dual representational capacities to both shape the world
according to goals and accurately reflect its state.