A summary of my article ‘Abilities to do otherwise’

[ This is a summary of ‘Abilities to do otherwise’, the topic of which I have recently been working on again. ]

Abilities not defined solely by stimulus and manifestation

The first point I argue for (section 2) is that intrinsic dispositions and abilities are not to be defined solely in terms of stimulus conditions and manifestations. They must also be defined by what I call a set of definitional circumstances. By way of example, the idea is that:

(1) This lump of salt is disposed to dissolve when placed in water

has to be understood as something like:

(2) This lump of salt is disposed to dissolve when placed in water at standard temperature and pressure

if it is to succeed in picking out a property.

But the definitional circumstances can’t be just any old set of circumstances (3018). For example, if we say someone is able to walk, we say nothing about what the agent can do in circumstances where some basic preconditions for walking are not met. The ability isn’t finked or masked in such cases; rather, it “fails to apply” (3019).

I call the minimal set of preconditions that need to be met for the performance of some action type A, the action-realisation conditions for A. So for example, the ability to walk has no application when someone is floating in outer space – the action-realisation conditions for walking simply aren’t met.

The set of circumstances which (combined with the stimulus and manifestation) defines the ability can be any subset of the action-realisation conditions. I illustrate this with the example of Ann and her elderly father Julian. When Ann asks if Julian can walk, she’s not asking whether Julian has the “average” walking abilities of a human adult (if there even is such a thing), she’s simply asking whether he can make it the 30 or so metres across a flat driveway to the car. But when Ann asks this, she’s still asking about one of Julian’s intrinsic abilities. She’s not asking about whether Julian would have (i.e. gain) the ability to walk to the car, were he to approach one. No, the ability she’s asking about can be specified as something like the ability to walk very short distances across flat surfaces (3020).

I suggest that it is important to see that definitional circumstances are different to opportunities. The former define the ability. But no claim is made yet about whether the person is in such circumstances, which is one way of understanding an opportunity (3020).

I follow Alexander Bird and Ann Whittle and use hyphens to indicate when circumstances form part of the definition of an ability, as opposed to being circumstances in which the ability is possessed. For example, in the scenario above, Ann is asking about Julian’s ability-to-walk-very-short-distances-on-flat-surfaces.

On p. 3022 I argue against privileging one set of definitional circumstances and saying that the ability to A should be tied to that set of definitional circumstances. For example, the ability to walk should not be tied irrevocably to circumstances which specify earth-like gravity conditions. Rather, we have to recognise that even conditions such as these are supplied by context with the result that the phrase ‘is able to walk’ could, given the right context, be referring to an ability to walk in moon- or Mars-like gravitational conditions.

Vihvelin’s account of abilities

Kadri Vihvelin recognises three senses of ‘able’ (3023-3024):

  1. The skill sense
  2. Narrow ability (someone possesses the skills and the psychological and physical capacity needed for A-ing)
  3. Wide ability (someone possesses the narrow ability plus the opportunity to exercise it)

Vihvelin thinks that the key to understanding the disagreement between compatibilists and incompatibilists is “understanding the difference between narrow and wide abilities” (Vihvelin, cited on p. 3024). Vihvelin thinks it is wide abilities that are relevant to free will, although not always wide abilities to do the things we think – sometimes the wide ability to try is enough.

Vihvelin’s account of abilities employs the notion of a test-case. Roughly, an agent has a narrow ability if she has an intrinsic property which is retained throughout a period of time, and if she were to try while in a test-case, she would succeed in performing the action in a suitable proportion of cases (3024).

The notion of test-case is crucial. Vihvelin employs it to address the problem of masks faced by conditional accounts of dispositions and abilities (see p. 3025). But it has the result of tying any given ability property to a single set of definitional circumstances. On Vihvelin’s account, there is just one ability to walk, and we assess whether a person has that ability by considering whether she succeeds at walking in the test-cases: in the walking test cases, for there is only one set of test-cases for walking. But what this means is that Vihvelin “precludes from view a whole set of ability properties”: the set which defines an ability not by considering the agent’s performance in standard or normal conditions (which is essentially what the test-cases are).

Abilities to do otherwise

I begin section 4 by distinguishing two ways in which an ability might be general. First, abilities might be general in the sense that they make no claim about whether the definitional circumstances obtain (3029). So for example, someone might have the following ability:

(A1) the ability-to-walk-up-4 %-inclines

This is an intrinsic ability, and someone might possess it regardless of whether they are currently facing a 4% incline. To say someone is able to walk in the above sense (to attribute to them ability A1) is to say nothing about the person’s current circumstances – it is to say nothing about whether someone has the opportunity to exercise that ability. This sense of generality is in contrast to the particularities of the person’s current circumstances.

Second, there is generality with respect to the defnitional circumstances. For example, the following two abilities vary in how much detail they include in their definitional circumstances (3030):

(A1) the ability-to-walk-up-4%-inclines

(A2) the-ability-to-walk-6-mph-up-4%-inclines-against-a-head-wind-of-12-mph

The first of these is more general in the sense that it “applies to” more circumstances. To assess whether someone has A1, we would have to assess their performance in all the 4% incline scenarios. By contrast, to assess whether someone has A2, we only need assess their performance in the 4% incline scenarios where there is a 12mph head wind – but in addition we have a further specification on the type of action required: the walking must be at 6mph.

This first distinction could be thought of as a general / particular distinction.

The second distinction could be thought of as a general / specific distinction, and it is important to see that it is a matter of degree: abilities can be more or less general in this sense.

With this distinction in hand I present the following principle, P1, which begins to identify which sense of ‘able’ is relevant to free will:

(P1) Intrinsic abilities (narrow abilities included) are relevant to what the agent can do in the particular circumstances he finds himself in only if the circumstances which define the ability match those circumstances which obtain.

What this principle says, in effect, is that the only senses of ‘able’ relevant to free will will be those whose definitional circumstances “match” the circumstances that the agent is currently in. Put otherwise: the agent must have the opportunity to exercise an intrinsic ability, if it is to be relevant to what she can do in her current circumstances.

P1 must be qualified with two further principles (3031). P2 is stated as follows:

(P2) If an agent possesses the intrinsic ability-to-A-in-C it does not follow that he possesses the intrinsic ability-to-A-in-C’, where C’ is a subset of C.

Intuitively, this says that the possession of a general (in the second sense) ability to A does not entail the possession of every ability to A which is more specific than it. The reason for this is as follows. An agent can have an ability-to-A-in-C even if there are some C-type situations where they fail to perform the action: to possess an ability you need a good measure of success, but not infallibility. But suppose there is a portion of modal space (a subset of the C-type situations) where the person always fails to perform the action. If a description of that portion of modal space is added to the specification of the ability in question we arrive at a more specific ability which the agent does not possess. I gave the following example (3032):

(Running-1) the ability-to-run-on-surfaces-with-inclines-between-0-and-5%-and-in-heats-of-between-0-and-32 degrees C

(Running-2) the ability-to-run-up-inclines-of-5%-in-heats-of-between-0-and-32 degrees C

It might be that someone, e.g., Rachel, possesses Running-1 but lacks Running-2. Rachel can run in enough of the Running-1 scenarios to allow us to attribute to her the ability. She fails in some cases. And many of those failure cases turn are Running-2 cases, so many, in fact, that Rachel doesn’t count as having Running-2.

Now suppose that Rachel is standing at the foot of a 5% incline. We want to know whether she can run up the incline (maybe because we want to know whether she can do so freely). Should we say ‘Yes’, because she has Running-1, or should we say ‘No’, because she lacks Running-2? I suggest that we should say ‘No’ (3033). This leads to the following principle:

(P3) If two intrinsic abilities (narrow abilities included) have definitional circumstances which match the circumstances that the agent is currently in, then the one with more circumstantial detail in its definition takes precedence.

I suggest that this is an important conclusion with respect to determining which sense of ‘can’ or ‘able’ is relevant to free will. It will be the sense which satisfies (P1) through (P3). It also shows that Vihvelin’s account of the abilities relevant to free will fails: because she doesn’t recognise that the definitional circumstances vary, in effect she only ever asks about the-ability-to-A-in-ordinary-circumstances (suitably nuanced in an attempt to avoid the problem of masks). But someone might possess the-ability-to-A-in-ordinary-circumstances and be in “ordinary circumstances” and yet still be unable to A because there is a more specific sense of ‘able’ which makes refernce to just some subset of the ordinary circumstances and in that subset the person fails to succeed, so doesn’t have the ability.


Kittle, Simon. “Abilities to do otherwise.” Philosophical Studies 172, no. 11 (2015): 3017–35. doi:10.1007/s11098-015-0455-8.