There are no such things as unintentional actions! There are unintentional results or unintended consequences of certain actions but those things, those events, are not actions.
It is very common to classify actions into two main types, intentional and unintentional. For example, Anscombe states that “actions can be either intentional or unintentional”. And Davidson argues that although we would not usually call the unintentional consequence of an intentional action an ‘action’, “it should not be inferred from this that [the unintentional result] is therefore something different from [the intentional action], say just its consequence.” Both Davidson and Anscombe (and many others) maintain that there are these things called ‘unintentional actions’, which are genuine actions. I think is a confusing, misguided and useless taxonomy.
Why would “unintentional actions” be classified as actions; what interesting property do they have in common with (intentional) actions? Nothing. There is no interesting connection between the unintentional results of actions and the actions themselves, except that they are the unintentional results of an action! J.W. Meiland observed that calling an unintentional result an ‘action’ only “serves to indicate that what happened has some relation to something that someone did.” He argued that the
use of ‘intentional’ and ‘unintentional’ depends on whether or not certain results were intended or unintended; these terms are used to indicate the status of those results. They are not to be regarded as adjectives modifying ‘action’.
This is right. What intentional and unintentional actions have in common is that they both have some relation to an (intentional) action. Intentional actions stand in the relation of being identical to intentional actions, while unintentional actions stand in the relation of being caused by or the consequences of intentional actions. But then actions would be the class of things that either are intentional actions or are the consequences of intentional actions. Clearly, this is not a theoretically interesting category.
For example, say that Gilbert has just bought some coffee and is running late to seminar. In the elevator he checks his wrist-watch and in so doing accidentally spills some of his coffee. Consider the following two sentences about Gilbert.
(1) Gilbert checked his wrist-watch.
(2) Gilbert spilled his coffee.
Are both (1) and (2) are true? It seems that they are. But are they both about actions (perhaps the same action)? No. It does not follow from the truth of (2), if it is actually true, that Gilbert acted or that any an action took place. What if instead of checking his watch, Gilbert thrashed around having a seizure and in so doing spilled his coffee? Is (2) true? It seems that it is. But did Gilbert act? Did an action take place? No. Gilbert spilled his coffee by having a seizure but there was no agency on behalf of Gilbert. The seizure just happened to him. It does not follow from ‘Gilbert seized’ that Gilbert acted. Likewise it does not follow from ‘Gilbert spilled his coffee’ that Gilbert acted. (Consider ‘Gilbert sneezed’, ‘Gilbert bled’, ‘Gilbert fell’, ‘Gilbert snored’, etc.) One must be wary of inferring the existence of an action just from the fact that there is an agent in the subject position of a true sentence.
Perhaps this is just a terminological quibble. Some things that happen are done by agents, some are the consequences of things done by agents, and some are things not done or caused by agents at all (e.g. Gilbert fell down the stairs.)
Keeping actions, agent-initiated events, and non-agent-initiated events distinct is very important when invesitageing philosophical questions about the practical knowledge and abilities of agents.