More generally, we can now see that there are in fact two differentdimensions along which one's notion of a constraint might be broaderor narrower. A first dimension is that of the source of aconstraint — in other words, what it is that brings about aconstraint on freedom. We have seen, for example, that some theoristsinclude as constraints on freedom only obstacles brought about byhuman action, whereas others also include obstacles with a naturalorigin. A second dimension is that of the type of constraintinvolved, where constraint-types include the types of internalconstraint just mentioned, but also various types of constraintlocated outside the agent, such as physical barriers that render anaction impossible, obstacles that render the performance of an actionmore or less difficult, and costs attached to the performance of a(more or less difficult) action. The two dimensions of type and sourceare logically independent of one another. Given this independence, itis theoretically possible to combine a narrow view of what counts as asource of a constraint with a broad view of what types of obstaclecount as unfreedom-generating constraints, or vice versa. Asa result, it is not clear that theorists who are normally placed inthe ‘negative’ camp need deny the existence of internalconstraints on freedom (see Kramer 2003; Garnett 2007).


Satisfied customers are saying