As the article stresses from the start, cognition has to do with knowledge and the processes that support its creation. However, the issue that becomes apparent in current literature on embodied cognition, extended cognition, enactive cognition and so on, is that not every process that is causally in support of cognition is a cognitive process. Lest we forget that, we all know that circulation supports cognition. One cannot think without it. But that hardly shows that circulation is a cognitive process. So the main issue separating out causal support for cognition from processes that constitute cognition is how to draw the line, the line that separates mere causation from constitution.

While this is the main issue that needs addressing, it is not one that will be easy to solve. Consider Susan Goldin-Meadow’s work with gesture (2003, 2004). She has shown that getting learners to gesture in physical proximity to where a calculation result should go in solving a math problem helps the learner to solve the math problem. Yet I once asked her in person “do you think the gestures are constitutive of learning or mere causal support?” Her answer was informative. First she said she thought of gestures as causal support for learning. Second she said “I don’t know how one would empirically pull apart the matter of what is causal support from what is constitutive of cognition.” Her point is that from the outside (the perspective of the viewer, the researcher) they would look the same behaviorally.

She is exactly right. One cannot tell from the behavioral data alone (nor fMRI data) whether some process in the perceptual system or motor system is mere causal support for cognition or is cognition itself. To discern that difference requires a theory—a theory of cognition. Up to now in the literature on these matters, this issue has been ignored. Any time a researcher finds a nice correlation between processes in perceptual or motor areas and solving of cognitive tasks, researchers are giving in to the temptation to say the processing in these areas must be cognitive processing. The argument is that it must be cognitive processing because it causally contributes to solving the cognitive task and the subject could not solve the cognitive task without this processing. But remember that the same is true of circulation. It contributes to the ability to do cognitive processing and the subject could not solve the cognitive task at hand without it.

A comparison that I have made elsewhere is this. When the Identity Theory of mind and brain was originally proposed, it was realized early on that just because we find empirical correlations between being in mental state M and being in brain state B, that alone does not show that \(\text{ M}=\text{ B}\). As dualist long proposed, even if the mind was not the brain, we would expect to find such empirical correlations as long as the two types of states are metaphysically or causally related in some way (as they are on even dualistic theories).

The same is the case for cognitive processing. We would expect there to be some causal relation between what happens in the world outside the brain and what happens during cognition if we are thinking about the world. We would expect there to be some causal contribution between processing in perceptual or motor areas when we think about the perceived world or the world as acted upon by us. We expect there to be causal processing tightly bound up with cognitive processing. But those close causal connections alone do not show that the processing in these regions constitutes cognitive processing any more than empirical correlations of M and B show \(\text{ M} = \text{ B}\).

So what do we do? How do we decide what is constitutive of cognition and what is mere causal support? My answer is that only when we have a good understanding of what cognitive processing is will we be able to answer this question. This foundation article goes a long way towards the answer. It does not go far enough, of course, because it does not equip us with the means to answer the question raised about the difference between causal support and constitution. However, it does far more than most because it is asking the right question and looking in the right places for the right answer.

As the reader will detect, this article offers too many suggestions of what cognition is and there is no single thread that pulls them all together. As for as I can tell, that is the only thing that remains to be done. Elsewhere I have made some remarks along the lines of providing a way to do this (Adams and Aizawa 2008; Adams 2010; Adams and Garrison 2012), so I won’t repeat them here.