intentions vs rules
Intentions and rules are both features of the imagined world that act upon the real world - i.e. heterotopias - but both do so differently.
An intention is an action together with a description of an action, or a potential action together with a description of a potential action. The thinking is that actions that you take "on purpose" are constructed as follows:
- take your image of the world as it appears to be now
- describe a specific way it could change
- select among actions that "could cause that difference".
You could say that the description of the change that generates the action describes the action itself. Hence, action under a description.1
Call the world-model taken together with the description of a change that it could undergo a theory of change.
Given an observed action, one can conjecture meaningfully about the theory of change that generated it: any action will be more congruent with some theories of change than others. One can also deduce that the agent cares about the change happening, because it moved them to act.
You would need additional information to deduce why they care; this distinction is what distinguishes Anscombe's concept of intention from her concept of motivation. She wants to restrict the former word to the psychological phenomenon for which the action itself always provides least some evidence.
People switch between theories of change often, to avail the benefits of narrowness. Theories also fail to hold a lot. When an action doesn't work as intended, an agent will update their theory of change to account for what actually did happen.
Therefore, claims about what theories of change are active in a given moment rapidly lose credibility as the scope of the theory gets broader and as the causal chain being presented for analysis gets longer. Long causal chains and broad classes of actions impose high burdens of proof.
Therefore understand intention as a local tool of analysis in both the causal and logical domains.
Commentary on larger structures of goal-oriented behaviour can instead use the term motivation.
Compare rule: a rule is also a description of actions that exhibits locality and can be inferred from an agent's behaviour, but instead of being a tool for building a predictive model of a single agent, it's a totem around which several agents can coordinate. This means that a rule can be reflexive while an intention can't.
todo this note is about the descriptive–normative agent-concept pipeline and why it happens. I'm thinking the effect occurs because we're familiar with both modes as themselves being strategic tools that synergize well in social games of ideas.
Footnotes:
The abuse of notation that absorbs the hypoethesised psychological antecedents of an action into the action itself seems to have a lot to do with fundamental attribution error.