Here is the thread I promised on ecological psychology (tentative, since I don’t actually know enough about it).

Tl;dr: what is right about ecological psychology is compatible with a neurocomputational and representational approach to cognition.
1/

Gibson’s ecological psychology maintains that we perceive affordances, which are opportunities for action (eg the walkability of the ground, the graspability of a handle). Therefore, perception and action are inseparable and ought to be studied together ...
2/

not only in the lab but also in the environment in which information about affordances is available and actions can be performed. All of this is true and important and fits the true and important view that cognition is situated.
3/

Gibson also insisted that perception is direct and therefore we don’t perceive through internal representations. A fortiori we don’t need to study inner representations and computations (since there aren’t any).
4/

This is not only wrong but harmful because it contributed to an unnecessary divide within the mind sciences.
5/

Considerations about illusions, dreams, memory, imagination, and computational tractability, among others, not to mention the whole field of cognitive neuroscience, establish that we do not perceive the world and guide actions directly.
6/

Rather, we perceive and guide action through information-bearing internal states that need to be corrected via sensory information and are subject to computational constraints. Cognitive neuroscientists call such states neural representations.
7/

Furthermore, while Gibson is right that sometimes we perceive affordances, perceiving affordances requires processing information. And we perceive more than affordances, and often it takes us more than perception—it takes some thinking—to figure out what we can afford to do.
8/

So, pace Gibson, in order to understand and explain cognition, including how perception and action are intertwined, we need to study neural computations and representations.
9/9