Failures of agency

Human beings don't always do what they judge best: sometimes we judge it best to do A, but nevertheless do B. In philosophical debates such behaviour is usually referred to as akrasia, or lack of self-control. This dissertation focuses on the question: does the occurrence of such behaviour have implications for the fact that we understand each other and ourselves as rational agents? The first part elaborates on the notions of action, agency and rationality. Agency is defined as the capacity to act. Actions are characterized by goal-directedness and by control; to act is to orientate your behaviour on the basis of reasons. This shows that agency is inherently connected to rationality. This raises the question: should cases where people do something other than they judge best be seen as successful exercises of agency? It is argued that whereas certain cases of going against your judgement are characterized by goal-directedness, not all of them are. Also, certain important notions of control seem not applicable to any cases of going against your own judgement. This means that whenever we go against our own judgement, we violate certain expectations generated by our self-understanding as agents. Going against your judgement is always a failure of agency. However, the fact that we now and then fail to exercise our agential capacities does not mean that we do not have such capacities. The second half of the dissertation discusses several implications of this view on failures of agency. Firstly, the existence of failures of agency shows that our desires do not automatically track our evaluative assessments, and this forces us to explain the connection between evaluation and motivation in successful exercises of agency. Secondly: if certain kinds of control are absent in failures of agency, does this mean that we are never responsible for failures of agency? The view defended is that our practice of moral responsibility relies on the idea that only agents can be blameworthy, but that it is not so clear that actual control is also required for blameworthiness. This leads to the conclusion that we might be legitimately held responsible for behaviours that are characterized by a lack of control, as long as the person assessed can still be understood as an agent. This raises the question: when do failures of agency give us reason to conclude that we should not understand someone as an agent? It is argued that most people who suffer from failures of agency still exercise important agential capacities that warrant understanding them as agents, even in many cases where their failures can be called pathological. The dissertation ends with the claim that our attitude towards our own failures of agency is necessarily ambivalent. When we say