Citation:
Abstract:
Eleven authors with disparate relevant backgrounds give their view on what is meant by the word "cognition".
PMID: 31287972 doi: 10.1016/j.cub.2019.05.044.
Full Text
Whisking away a pesky fly doesn’t require much introspection or deliberation; a simple sensorimotor reflex will do. A pivotal chess move, on the other hand, calls for complex information processing and access to memories and learned models of the world. These processes clearly lie at different ends of a spectrum, the principal axis of which we can call cognition.
So far so good. But ‘cognition’ also implies something categorical and well delineated, and here things get murkier. Over the past month I have been asking colleagues favoring the term to define it for me. Cognition, I have learned: “requires learning”; “isn’t a reflex”; “depends on internally generated brain dynamics”; “needs access to stored models and relationships”; “relies on spatial maps”, and so on. The lack of a clear consensus isn’t very surprising. Mental activities, after all, make up a sprawling continuum that isn’t easily parcellated and labeled. Yet, we keep using terms like cognition. Why?
Part of the reason may be historical inertia and human chauvinism. For a long time, we used ‘cognition’, ‘intelligence’ and ‘consciousness’ to contrast the human experience with those of other animals, justifying our superiority in the process. We humans, Descartes argued, are cognitive beings with thoughts and feelings. Animals, in contrast, are mere machines. Then Darwin comes and challenges this neat anthropocentric order, and suddenly we are all one big family with shared ancestry. Differences between the human and animal mind are, in his words, “one of degree and not of kind”. Today, proposals to study ‘cognition’ and ‘intelligence’ in honeybees and octopuses receive serious consideration — a clear victory for Darwin. Describing animal behavior as ‘cognitive’ certainly rights a wrong, but it also further muddles the meaning of the term.
Archaic as this terminology may be, we neuroscientists have a really hard time letting go of it. It’s imperfect, but it’s what we have to manage, categorize, and compartmentalize the immense diversity of mental processes we deal with. It’s part of the language we use to frame our research and generalize our findings. Our problem is that neural circuits do not implement ‘cognition’ or other vague concepts inherited from philosophy and psychology; they implement algorithms that need to be rigorously characterized and defined. After all, our understanding of the brain can only be as clear as the language we use to describe its underlying processes. Doing away with slippery and outdated terms like ‘cognition’ would force us to come up with a new vocabulary suited to delineate and specify what we are studying. It’s not going to be easy, but it’s a challenge we should seriously consider.