Extended cognition is the theory in cognitive science that mental and cognitive processes are not confined within the skull or the skin, but extend into the physical tools, environment, and social structures we use to think.
According to this view, we do not just think about our tools; we think with them. To free up limited mental bandwidth, we externalize memory and computation to the physical world—a process known as cognitive offloading. If an external resource performs a function that we would recognize as cognitive if done in the head (such as calculating, structuring, or retrieving), then that resource is functionally part of our cognitive loop.
For example, a physicist solving a complex equation on a whiteboard is not computing everything in their head and then writing it down. The act of writing, looking at the symbols, and reorganizing them on the board is the process of thinking. Similarly, an AI agent’s intelligence is not restricted to the static weights of its neural network; when it reads, writes, and triages local Markdown files, that file system acts as its extended mind.
By understanding extended cognition, we stop viewing tools as passive storage and start seeing them as active extensions that transform the threshold of what a cognitive system can achieve.