Sean’s Public Notebook

Epistemic actions are taken to think, not to progress toward a goal

Kirsh and Maglio distinguish pragmatic vs. epistemic actions: A pragmatic action is an action one takes in order to accomplish a particular task, usually to bring a person closer to their goal. An epistemic action, on the other hand, is an action a person completes in order to think through something by uncovering previously hidden information, or information that would be harder to compute “mentally.” And it turns out that Experts manipulate and perform more epistemic actions than non-experts in many situations.

What’s important here is that epistemic actions are not always seen by actors as useful towards a particular goal in the moment they undertake them; they are exploratory by nature! Cognition can depend on action, rather than intelligent action always depending on cognition first. That is: Epistemic actions can often precede cognition because a person may take an action first in order to understand a problem space, or unveil information, before devoting the energy to think about what to actually do next. This is generally in line with Kirsh’s broader work on distributed cognition, in which agents offload cognition into the world rather than make mental models and depend solely on thinking “in the head.”

But this raises a question I don’t have a good answer for: Is the action truly separate from cognition at all? Both embodied and distributed cognition might question the very notion of precedence here since action and “thought” could all be wrapped up into a single cognitive act.

References

Kirsh, D. and Maglio, P. (1994) On Distinguishing Epistemic from Pragmatic Action.