Having a Disposition and Making a Contribution

Abstract

Dispositional accounts of various phenomena have claimed that dispositions can be intrinsically masked. In cases of intrinsic masking, something has a disposition while also having an intrinsic property that would prevent that disposition from manifesting in the face of its stimulus. This paper develops a theory of disposition ascriptions capable of recognizing such dispositions. The theory is modeled on the view that dispositions are powers. I propose that having a disposition is a matter of exerting a corresponding kind of influence. Unlike powers theories, however, the account largely falls silent on questions of fundamental metaphysics. It does not build dispositions into fundamental ontology, posit necessary connections between properties, or otherwise appeal to sui generis modality.

Dispositions are often masked. We protect fragile glasses with packing material so they do not manifest their disposition to break when struck. A dose of antidote can prevent a poison from manifesting its disposition to cause harm when ingested. These dispositions are masked in that something safeguards against their manifestation without threatening the disposition itself. The mask would prevent the disposition from manifesting if its stimulus condition obtained and it would do so without causing the disposition bearer to lose the disposition. The packing material would prevent the glass from breaking. It does not make the glass any less fragile.

Masking in these examples is an extrinsic matter. In each case, the disposition is masked by something external to the object that bears the disposition. This is part of what makes these cases such clear examples of masking. While the glass may not break when struck or the poison cause harm when ingested, most of us do not balk at recognizing their dispositions to do just that. These dispositions are intrinsic properties. So while outside factors may conspire to prevent the glass from breaking or the poison from causing harm, this does not itself tell against their respective dispositions. Having those dispositions is a matter of what an object is like intrinsically. If they do not manifest in one environment, we have no trouble imagining others in which they would.

Whether masks must be extrinsic, however, is controversial.¹ To those who believe that masks are necessarily extrinsic, there is a marked difference between safeguards that are intrinsic to an object and those that are not. An object cannot be disposed to manifest an outcome under some conditions and also possess an intrinsic safeguard that would prevent that disposition from manifesting in those same conditions. In contrast, those of us who believe that intrinsic masking is possible claim that something may have a disposition while also possessing an intrinsic safeguard that would prevent that disposition from manifesting. Lauren Ashwell (2010) offers the example of a poisonous berry with indigestible skin. Being poisonous, the berry is disposed to cause harm when it is ingested. But thanks to its indigestible skin, the berry can be consumed safely as long as that skin remains intact. The berry’s skin masks its poisonous disposition. Removing the skin would not render the berry poisonous. Rather, it would reveal the disposition that the skin had masked.

This debate is not idle. What dispositions we recognize in cases like these has broad implications for other philosophical projects. Intrinsically masked dispositions have been invoked to support dispositional theories of desire (Ashwell 2014, 2017) and rule-following (Martin & Heil 1998). They are a critical component of dispositional accounts of free will and the ability to do otherwise (e.g., Clarke 2010; Fara 2008; Haji 2008; Vihvelin 2013). They may help defend dispositional theories of perception and character traits. Their possibility invites us to reconsider alleged counterexamples to dispositional theories of phenomena more broadly. Dispositions that were alleged to be missing may instead have been masked.

If intrinsic masking is possible, it brings out a tension between competing theories of disposition ascriptions. Disposition ascriptions have traditionally been understood in terms of conditionals. Something has a given disposition just in case certain subjunctive conditionals are true. While conditional theories may recognize dispositions that have been masked by extrinsic factors, they have generally struggled to recognize dispositions that are masked intrinsically. In contrast, intrinsically masked dispositions are readily accommodated by the chief alternative to the conditional approach: theories of dispositions situated within a metaphysics of powers. Powers theories hold that certain dispositions or disposition-like properties are an irreducible part of fundamental ontology, not to be explained by non-dispositional properties or laws. While powers theories allow for the possibility of intrinsic masking, they are a significant metaphysical commitment, particularly if our goal is merely to recognize certain intrinsically masked dispositions.

My aim in this paper is to offer an alternative. This paper outlines an account of disposition ascriptions modeled on theories of powers. It casts having a disposition as a matter of exerting a kind of variable modal influence. But unlike powers theories, it falls largely silent on questions concerning fundamental metaphysics. And, in this respect, it shares much of the spirit of conditional theories of disposition ascriptions. It is compatible with a Humean view of fundamental properties and laws. It does not commit us to necessary connections between properties, build dispositions into fundamental ontology, or otherwise appeal to sui generis modality. The paper proceeds as follows: Sect. 1 introduces the problem that intrinsically masked dispositions pose for conditional theories of disposition ascriptions and outlines a solution to that problem afforded by powers theories. Sections 2 and 3 develop a theory of disposition ascriptions that adapts this solution. I then apply it to the problem of masking in Sect. 4 and to opposing dispositions in Sect. 5. Section 6 closes by considering an objection to this view.

I want to acknowledge two limitations of this proposal upfront. First, pace Jenkins and Nolan (2012), I assume here that a disposition’s stimulus and manifestation conditions must be of a type that possibly obtains or occurs. No one, for example, is disposed to prove that four is prime when it is provable that four is prime.² Second, I will assume that dispositions are intrinsic properties. Given the laws, any two things that have the same intrinsic properties also have the same dispositions. It may very well be the case that some dispositions are not intrinsic in this sense (McKitrick, 2003; Yablo, 1999). However, I take dispositions to be at least paradigmatically intrinsic, and the account here is concerned only with dispositions as such. Should some dispositions turn out to be extrinsic, the account will need to be extended to accommodate them. Though I see no reason why that could not be done, I will not attempt to do so here.

1 The Problem of Intrinsic Masking

Masking has presented a longstanding problem for conditional theories of disposition ascriptions. The challenge raised by intrinsic masks comes out most clearly in the context of this more general problem.

Conditional theories tell us that the truth conditions for disposition ascriptions can be stated as subjunctive conditionals. The account proceeds in two steps. In the first step, the dispositional predicates of ordinary language are translated into a standard, explicitly dispositional form. To have a conventional disposition is to be disposed to manifest some condition M in stimulus condition C.³ In the second step, the account supplies truth conditions for these new predicates by slotting their stimulus and manifestation conditions into the antecedent and consequent positions of a corresponding conditional. The simplest form of the conditional account (SCA) holds that an object o has a disposition to manifest M in stimulus condition C iff o would M if C. For example, if fragility is the disposition to break when struck, SCA tells us that something is fragile just in case it would break if it was struck. Masking calls this into doubt (Bird, 1998; Johnston, 1992). Objects retain their dispositions in the presence of a mask but may fail to manifest them, rendering the simple conditionals associated with them false. A wine glass that has been wrapped in packing material may be fragile despite the fact that it would not break if it was struck.

Conditional theories have generally sought to address this problem in one of two ways. One strategy has been to identify sanitized testing conditions that are free of masks and other forms of interference and then take a thing’s disposition to be a matter of how it would behave there. In practice, this has meant revising the account so that the antecedent condition of the relevant conditional will not be satisfied when masks are present. For example, following Lewis (1997), Choi (2008) incorporates these sanitizing elements directly into the stimulus conditions associated with conventional dispositional predicates. So understood, it is no mark against the vase’s fragility that it would not break if struck. For fragility is not the disposition to break when struck, but rather the disposition to break when struck under some more narrowly specified circumstances where (the thought goes) packing material and other sources of interference are not present. Other proposals introduce these sanitizing elements by instead attaching a more general rider to the conditional in SCA. Contra SCA, they hold that something has a disposition to M when C just in case it would be M if C in ideal conditions (Mumford, 1998), normal conditions (Malzkorn, 2000), or ceteris paribus (Steinberg, 2010). Others pair SCA with novel semantics for subjunctive conditionals that mimic the results of those general riders without explicitly qualifying the form of the simple conditional account itself (Bonevac et al., 2006; Gundersen, 2002).

The second strategy has been to make the truth of disposition ascriptions a function of how something would behave across a diverse range of testing conditions. Manley and Wasserman (2008) propose that something has a disposition to M when C iff it would M in a suitable proportion of C-cases, where each C-case is a possible scenario in which C obtains. Seen this way, the truth of a disposition ascription does not hang on any one conditional. It depends on a range of conditionals considered in aggregate and the pattern of would-be behavior that they reveal. Something may have a disposition that fails to manifest in some C-cases provided it manifests in a sufficient proportion of the C-cases overall. The vase can be fragile even though it would not break if it was struck in those cases where it was wrapped in packing material or some other form of interference was present. What matters is that there are enough cases without such protections where it would break if it was struck.

While these revised accounts may accommodate extrinsic masking, they struggle in cases where masks are intrinsic.⁴ The relevant conditionals will often be false. The poisonous nature of Ashwell’s (2010) berry is masked by its indigestible skin. That skin is presumably present under normal, ideal, or ceteris paribus conditions. It will be present in many, even most, possible C-cases. And when it is present, it will generally be false that the berry would cause harm if ingested. This is no accident (Everett, 2009). Choi’s (2008) sanitized stimulus conditions and Manley and Wasserman’s (2008) C-cases are explicitly defined in terms of factors that are extrinsic to the putative disposition bearer. They consider how an object would behave in a different environment, not how it would behave if it were intrinsically different. Dispositions that are masked by intrinsic properties are apt to remain masked in a new environment. Moreover, all of these accounts trade on the notion that masks are aberrations, that they will not be present in normal or ideal conditions, or so common that they are present in a sufficiently large proportion of cases. Some intrinsic masks may indeed be aberrations. But as Ashwell’s (2010) example illustrates, they need not all be. It is perfectly normal, in all senses of the word, that the poisonous berry has its skin.

Conditional theories are not unique in this respect. Masks pose similar problems for related views. Vetter’s (2014, 2015) possibility account forgoes conditionals but adopts a proportionality measure similar to that of Manley and Wasserman (2008). According to Vetter (2014, 2015), something has a disposition just in case there is a sufficiently strong possibility that it manifests that disposition’s characteristic behavior. Something is disposed to M iff it will M in a suitable proportion of possible cases. Like Manley and Wasserman (2008), this account recognizes extrinsically masked dispositions but does not reliably recognize dispositions that are masked intrinsically. Since things retain their intrinsic properties across possible cases, intrinsically masked dispositions like the poisonousness of Ashwell’s (2010) berry will manifest in only a small proportion of possible cases and so will often go unrecognized.

Theories employing causal models fare somewhat better (Friend & Kimpton-Nye, 2023; Gebharter & Fischer, 2021). Causal models provide a framework for identifying sanitized testing conditions of the sort favored by some conditional accounts. For example, Gebharter and Fischer (2021) propose an account of dispositions ascriptions in terms of causal Bayes nets: triples $\langle V,E,P\rangle$ where $V$ is a set of random variables, $E$ is a binary relation on those variables, and $P$ is a probability distribution. By taking different values, variables in $V$ represent the presence or absence of a disposition’s stimulus, manifestation, masks, and other factors. Say that a context is an assignment of values to variables other than those that represent the stimulus and manifestation. Very roughly, Gebharter and Fischer (2021) claim that something has a disposition just in case there is an appropriate model and context such that the probability of the manifestation given the stimulus is greater than the probability of the manifestation not given the stimulus. Something may have a disposition then even if there is no chance that it manifests in the presence of an intrinsic mask, provided there is an appropriate chance that it manifests given that the stimulus occurs in a context that represents the mask as absent.

Testing for dispositions in different contexts allows Gebharter and Fischer (2021) to recognize intrinsically masked dispositions. When the skin of Ashwell’s (2010) berry is modeled as missing or broken, the probability that the berry harms someone given that they ingest it is plausibly greater than the probability that it does so not given their ingesting it. However, this comes at the cost of their account of masked dispositions more generally. There is no special status afforded to contexts that remove masks. The model sees no difference between a context that represents a mask as absent and one that represents circumstances that would enable something to act in ways that it is not disposed to do. Having a disposition only requires that there is some context, however niche, in which something satisfies the relevant criteria. Just as we can model the likely behavior of a fragile vase in conditions where it is not wrapped in packing material, we can model the behavior of a rubber ball in conditions where it is subject to extremely low temperatures. In those contexts, striking significantly raises the probability that both will break. But while the vase is fragile, the ball seemingly is not.

In contrast, intrinsic and extrinsic masking can be easily accommodated by a powers metaphysics.⁵ Powers are properties that are dispositional in nature (Bird, 2016). They bear certain causal, nomic, or dispositional relations to other properties necessarily, such that to have power is to be necessarily disposed towards certain ends. For example, conceived as a power, mass can be characterized by the disposition to resist acceleration and attract other objects with mass. To have mass is to necessarily be disposed towards those ends. A powers metaphysics holds that at least some fundamental properties are powers. Their dispositional nature is part of the ontological bedrock. It is not something to be explained or analyzed, but a point from which explanation and analysis can begin. This view of powers as sources of dispositionality opens the door to intrinsic masking. If having a power is sufficient for having a disposition, then at least some disposition ascriptions are not contingent on the truth of any corresponding conditionals concerning how the disposition bearer would respond under various circumstances.⁶ Something with mass, for instance, may be disposed to attract other objects with mass even if its other properties would systematically prevent it from doing so. So long as it has the appropriate power, it stands in disposition-like modal relationships to relevant states of affairs and can in turn be said to have the relevant disposition.

For example, Mumford and Anjum (2011) claim that all properties are powers. To be fragile, for instance, is to have a power that disposes towards breaking. Something that is very fragile may exercise that power with a strong intensity and something that is less fragile may exercise it with a weaker intensity. They model the exercise of such powers as vectors through a quality space. Where power is directed towards a type of effect with some intensity, and its vector is directed towards the corresponding quality with a corresponding magnitude. Inhibitory factors that might prevent the power from bringing about that effect are cast as opposing vectors. The result of that contest is determined by vector arithmetic, much like component forces. The disposition’s effect occurs if the total resultant vector of all the relevant powers crosses a given threshold. In masking cases, the resultant vector falls short of that threshold due to the exercise of an opposing power. The vase’s fragility, for instance, may dispose of it towards breaking, but it may be countered by an opposing power exercised by the protective wrap**. Intrinsic and extrinsic masking differ only in the source of that opposing power. In cases of extrinsic masking, the mask is a power of something other than the disposition bearer. In cases of intrinsic masking, the mask is the power of the disposition bearer itself.

For all their virtues, powers carry significant metaphysical commitments. Admitting powers into our ontology commits us to positions concerning the nature of fundamental properties, primitive modality, and the contingency of the laws. Seen merely as a means of recognizing intrinsically masked dispositions, this is a considerable cost. One virtue of conditional accounts and their relatives is that they carry comparatively few metaphysical commitments. To say that there are true biconditionals relating disposition ascriptions and subjunctive conditionals is not itself to endorse a reduction from one to the other. Conditional theories are compatible with a broadly Humean analysis of dispositions in terms of categorical properties and laws. But they do not demand it. One could follow Jacobs (2010), for instance, and take the order of explanation to go in the other direction such that the truth conditions of counterfactual conditionals are themselves given in terms of dispositions.

I want to suggest another option. Powers theories provide an attractive model for recognizing both intrinsically and extrinsically masked dispositions. I believe that we can adopt that model without subscribing to an ontology of primitive powers.

2 Contributions

Powers theories can account for intrinsic masking by framing dispositional properties as independent sources of modal influence. If we are to adopt that model, we will need to account for such influence without building it into the ontological bedrock.

 Events have a similar kind of influence (Johansen 2014, 2015). One way in which events have an influence on the world is by constraining how the state of the world evolves. Assume that the laws are deterministic. Given the laws, the state of the world at one time determines the state of the world at all other times. In that case, the obtaining of a world state imposes a kind of constraint on what the future may hold. It nomically necessitates that some sequence of future world states will obtain and that others will not. Ordinary events impose this same kind of constraint on future states of affairs, albeit to a lesser degree. The occurrence of an event limits the space of nomically possible futures. Any sequence of world states that obtains in the wake of an event must be consistent both with the laws and with the fact that the event occurs. And not all state sequences that are consistent with the former are also consistent with the latter. As a result, the event permits the obtaining of some state sequences and precludes the obtaining of others. In doing so, it has an influence on future states of affairs that is contingent on the laws, but independent of its environment. Following Johansen (2014), I will call that influence an event’s contribution and formally characterize contributions as sets of worlds. The contribution of an event e is the set of nomically possible worlds in which e occurs. Each world in that set marks one way in which the state of the world could evolve following e. State sequences that follow e in one of those worlds may obtain in its wake. Those that do not are precluded.

It will be helpful for us to refine this account in two ways. First, I will understand events as property exemplifications (Kim, 1976; Lewis, 1986a). Events are the having of a property by an object at a time and place. Second, in light of that, we will need to take some care in how we identify events across worlds in order to ensure that contributions properly reflect the significance of that constitutive property instance. When determining an event’s contribution, we should disregard the event’s constitutive object for the purpose of identifying its counterparts in other worlds. I will take the event to occur in any world where something in the constitutive object’s actual location has the constitutive property at the constitutive time. Call that thing the object’s event counterpart in that world. For example, suppose that you set a 200 g apple on a scale. The presence of a 200 g object on the scale at that time is an event. Its contribution is the set of nomically possible worlds in which something with a mass of 200 g is on the scale at that time. In some worlds that will be the apple. In others, it will not. But in every world in the contribution, there will be an event counterpart of the apple with the right property, at the right place, at the right time.

Like causation, contributions can be attributed to objects as well as events. Causation proper is commonly regarded as a relation between events. Nonetheless, we routinely extend causal credit from events to the objects that participate in them. If a window breaks because it was struck by a baseball, we do not hesitate to say that the baseball broke the window. Objects may be causes in a derivative fashion in virtue of participating in events that are causes proper. We can attribute contributions to objects in the same way. If events are the having of a property by an object at a time and place, objects participate in the events they constitute and, by extension, can be said to make contributions to those events when they occur.

Note that this is a very different use of ‘contribution’ than we find in the powers metaphysics of Molnar (2003), Mumford (2004), and others. On the Molnar/Mumford view, contributions are reified entities that mediate the relation between powers and the events that occur as a result of the power being exercised. Seen this way, a power manifests its contribution, and that contribution, often in conjunction with the contributions of other powers, results in the occurrence of some event. The power of fragility, for instance, would manifest a contribution towards its bearer breaking, which may or may not result in its breaking. A power may thereby make the same contribution in different circumstances but yield different effects. To posit such contributions is to introduce a third kind of entity that exists in the world in addition to powers and the events they produce. As McKitrick (2018) notes, it is not entirely clear what that third kind of thing is.⁷

I do not propose to reify contributions in that fashion. I will understand contribution as a species of influence reduced to patterns of nomic covariation, much like counterfactual and nomic sufficiency theories conceive of causation. The contribution of an event in this sense is a logical constraint on what future states of the world may obtain in light of the laws. Contributions of this sort ask very little of our ontology. They are in principle compatible with a powers metaphysics.⁸ If the laws supervene on fundamental powers, the occurrence of an event still constrains future states of affairs in the fashion I have described here. More importantly for our purposes, this type of contribution sits easily within a broadly Humean, categoricalist metaphysics. We can suppose that the fundamental properties are categorical and the laws supervene on their distribution per Lewis’s ( 1986c) principle of Humean supervenience. Alternatively, we might reject a Humean view of laws in favor of a realist view. The contribution of an event will supervene on that event and the laws in either case.

Contributions may not seem especially relevant to a theory of disposition ascriptions. As nomic constraints, they are defined by modally stark relations to world states. They render it impossible for some sequences of world states to obtain and leave others possible, without privileging one possible state sequence over another. In contrast, dispositional modalities typically fall between mere possibility and necessity. Something that is fragile, for instance, does not necessarily break when struck, but it does not just possibly break when struck either. Even in the absence of masks and other forms of interference, fragile things do not always break when they are struck.⁹ Things that are not fragile sometimes do. Fragility’s modal profile lies somewhere between the extremes of mere possibility and necessity. And where it falls in that range may vary. Dispositions are gradable. In addition to asking whether something is fragile, we may also ask how fragile it is and whether it is more or less fragile than something else.¹⁰

Contributions reveal a more disposition-like character in their relationship to ordinary events. Constraints on how the state of the world can evolve bear directly on what events may occur as a part of that evolution. Suppose that an event makes a contribution in which it is followed in every member world by an event e. That contribution would necessitate that e occurs. For while it might permit the state of the world to evolve in any number of ways, each one leads to e. The contributions of ordinary events will not necessitate future events like this. A future event e will occur in some worlds in a contribution and not in others. It will be distributed across a contribution’s worlds in a manner that reflects its lawful relationship to the contributing event. That distribution provides a means of grading modal relations between mere possibility and necessity. An event may occur in a larger or smaller proportion of a contribution’s world. The larger the proportion of worlds in a contribution where an event occurs, the greater the possibility that it occurs given that contribution.

In making a contribution then, an event has an influence on future states of the world that leaves subsequent events more or less apt to occur. Say that a contribution favors an event e to the extent that e follows the contributing event in some proportion of its member worlds. For a contribution to favor an event, in this sense, is for it to limit the range of open futures such that it at least minimally allows for the possibility that the event occurs. The larger the proportion of those worlds in which e follows the contributing event, the more that contribution favors e—the greater the possibility that the event occurs given the contribution. Similarly, say that a contribution favors an event e in conditions C to the extent that e follows the contributing event in some proportion of its member worlds where C. Again, the larger the proportion of those worlds in which e occurs, the more that contribution favors e in C.

Proportional measures like this have been invoked as both a mark of dispositionality and a means of grading a disposition’s strength (Manley & Wasserman, 2008; Vetter, 2014, 2015; Vihvelin, 2013). Such accounts claim that something has a disposition just in case it manifests that disposition’s characteristic outcome in a suitably large proportion of relevant worlds or world-like entities.¹¹ The larger that proportion, the more disposed the thing is. Such proportions serve as a measure of modal strength. They allow us to recognize gradable dispositional modalities that fall between possibility and necessity. They similarly allow us to recognize that contributions may favor events with an equivalent degree of modal strength.

To be sure, there are difficult questions to be asked of any view that adopts this kind of proportional measure. The nomically possible worlds are non-denumerably infinite and the subsets of those worlds in which certain events occur presumably are as well. The same goes for the other sets of entities that have been invoked in these proportionality measures. All of these views presuppose that there is some sense in which sets of the same infinite cardinality can occupy different proportions of a set of that same infinite cardinality. Spelling out that sense of proportion is clearly no small task. Manley and Wasserman (2008) have suggested that there may be a further measure on those sets that would not involve comparisons of cardinality. Vetter (2014) has proposed that we take finite stand-ins of these infinite sets and apply standard measures of proportion to those. I will not defend a particular account of proportion here, but I will assume that such an account is possible. If any account of proportion succeeds, some version of it is apt to apply to all of these views.

3 Disposition Ascriptions

This suggests an account of disposition ascriptions. Powers metaphysics can equate having a disposition with having a corresponding power. We can frame a similar account in terms of contributions. To have a disposition is to make a corresponding type of contribution—one that favors the behavior associated with that disposition to an appropriate degree. Something is fragile because it has an influence on the world that favors it acting in the ways we associate with fragile things to a degree that is commensurate with the modal profile of fragility.

What this amounts to will depend on how we individuate disposition types. While disposition types have traditionally been individuated by their characteristic stimulus and manifestation conditions, at least some dispositions do not appear to fit this form well (Fara, 2005; Manley & Wasserman, 2008; Vetter, 2014). For example, a loquacious person is disposed to speak, but it is not at all obvious that there are any particular stimulus conditions in which they must be so disposed. A very loquacious person may be disposed to speak at any time, regardless of their circumstances. Taken at face value, dispositions like this are characterized solely by their manifestation conditions. The loquacious person may simply be disposed to talk. In the interest of accommodating dispositions of either form, I will take all disposition types to be individuated by characteristic manifestation conditions and some to be further individuated by characteristic stimulus conditions. I will recognize “is disposed to M” and “is disposed to M when C” as canonical forms for dispositional predicates and assume that the conventional dispositional predicates of ordinary language can be translated into those forms.

What property we ascribe with those predicates may depend on context. We may recognize something as fragile, malleable, or poisonous in one context that we would not recognize as such in another context. For example, metalworkers may correctly describe an iron rod as malleable. Teachers preparing a children’s art project may equally well say that modeling clay is malleable but the rod is not. In making those claims, the metalworkers and teachers do not contradict one another. It would clearly be a mistake to infer that they disagree about what properties the iron rod actually has. Rather, because their claims are made in different contexts, they refer to different properties with the same predicate. The metalworkers claim that the iron rod has one property. The teachers deny that it has another.

Given that dispositions are gradable, it is appealing to see property differences of this sort as a matter of degree. Seen as a difference of degree, the iron rod and modeling clay are malleable to different degrees and the teachers and metalworkers adopt different standards concerning how malleable something must be to qualify as malleable simpliciter. The iron rod is malleable enough to count as malleable in the context of metalworking, but not in the context of a children’s art project. More generally, we can say that to be disposed to M (when C) simpliciter is to be disposed to M (when C) to at least the degree required in the context in which that disposition is ascribed. So understood, the degree to which something is disposed is prior to it having that disposition unqualifiedly in a particular context. In ascribing a disposition to something, we recognize that it is disposed towards certain behavior to at least some minimum degree. Context determines how fragile something must be to qualify as fragile, how malleable it must be to qualify as malleable, and so forth.¹²

Proportional measures of dispositionality provide a mechanism for this type of gradability and context sensitivity (Manley & Wasserman, 2008; Vetter, 2015). The proportion measures the degree to which something is disposed to act in a given fashion. Context then determines the minimum degree to which something must be so disposed in order to qualify as having that particular disposition simpliciter. For example, Manley and Wasserman (2007, 2008) understand the degree to which something is disposed in terms of a proportional account of dispositional comparatives: one thing is more disposed than another thing to M when C just in case the former would M in a larger proportion of C-cases than the latter. That comparative establishes an ordering over objects by the degree to which they are disposed to M when C. The larger the proportion of C-cases in which something would M, the more disposed it is to M when C. For something to be disposed to M when C simpliciter is for it to sit above a contextually determined point on that scale.¹³ That threshold marks the minimum proportion of C-cases in which something must M in order for it to qualify as being disposed to M when C simpliciter in a given context.

We can understand disposition ascriptions in terms of contributions in much the same fashion:

(CONT) An object o has a disposition to M (when C) iff o has an intrinsic property P such that the event constituted by o’s having P makes a contribution in which o’s event counterparts M in a sufficient proportion of worlds (where C).

Recall that a contribution favors an event e to the extent that e follows the contributing event in some proportion of the worlds in that contribution. Similarly, say that something makes a contribution that favors it M-ing (in C) in so far as it makes a contribution in which its event counterparts M in some proportion of worlds (where C). The larger that proportion, the more the contribution favors its maker M-ing (in C). CONT claims that something has a disposition just in case it has an intrinsic property in virtue of which it makes a contribution that favors the behavior associated with that disposition to a sufficient degree. If that is a disposition to M, the contribution sufficiently favors its maker M-ing. If that is a disposition to M when C, the contribution sufficiently favors its maker M-ing in those conditions where C.

What counts as a sufficient degree—and by extension a sufficient proportion—will depend on the property being ascribed. CONT suggests a simple account of dispositional comparatives: an object x is more disposed to M (when C) than an object y just in case, on account of their intrinsic properties, x makes a contribution that favors it M-ing (in C) more than any such contribution made by y. As on other proportional measures, that comparative establishes a rank ordering of objects by how disposed they are to M (when C). Something is then disposed to M (when C) simpliciter just in case it sits above a threshold in that ordering determined by context. That threshold marks the minimum degree to which a contribution must favor its maker M-ing (in C) for it to be disposed to M (when C) simpliciter. It determines the minimum proportion of worlds (where C) in a contribution in which the event counterparts of the contributing object must M in order for it to qualify as being disposed to M (when C).

To illustrate, compare Granny’s fragile, thin porcelain vase with one of her sturdy, green rain boots. Though both objects make contributions that favor their breaking when struck, the degree to which they favor that behavior differs significantly. The vase makes a contribution by virtue of being made of thin porcelain. That contribution consists of the set of nomically possible worlds where something made of thin porcelain is located at the same time and place as Granny’s vase. Those event counterparts are all made of thin porcelain and subject to the actual laws, but they and their surroundings otherwise differ in every possible way. They will break in a significant proportion of the worlds where they are struck. Compare that with the contributions made by the boot. For example, the boot makes a contribution in virtue of being green. There are worlds in that contribution where the boot’s event counterparts are struck and subsequently break. (Some of those counterparts are made of thin porcelain after all.) But overall the worlds in which the boot’s event counterparts break are a much smaller proportion of the worlds in that contribution where they are struck. For unlike the property of being made of thin porcelain, there is no meaningful lawful relationship between being green and breaking when struck. Other contributions that the boot makes on account of its intrinsic properties fair similarly. As a result, CONT recognizes that the vase is considerably more fragile than the boot. In regarding the vase as fragile and the boots as not fragile, we adopt a standard for fragility that lies between the two. The vase makes a contribution that favors its breaking when struck to a degree that is sufficient for it to qualify as fragile. The boot does not.

4 Masked Dispositions

CONT has no trouble recognizing masked dispositions. Masks may prevent an object from manifesting a disposition. But this does not prevent the object from making a contribution that favors that manifestation. Contributions may strongly favor events that do not occur. Provided something makes a contribution that sufficiently favors the behavior associated with a disposition and it does so in virtue of having an intrinsic property, CONT recognizes it as having that disposition. The presence of a mask does not change that.

For example, suppose we wrap Granny’s vase in packing material that masks its fragility. The vase is again disposed to break when struck, but now thanks to the packing material it would not break if it was struck. CONT sees the vase as fragile. Wrap** the vase in packing material does not change the fact that it is made of thin porcelain nor does it change the contribution that it makes in virtue of being so constituted. As before, the vase’s event counterparts in that contribution are all made of thin porcelain. Like the vase, some of those counterparts are wrapped in packing material that will prevent them from breaking when struck. But those worlds will be a small proportion of the overall contribution and they are included in the contribution regardless of whether the vase is actually wrapped in packing material. If the proportion of worlds where the vase’s counterparts broke when struck was sufficient to render the vase fragile when masks were absent, it is also sufficient to render the vase fragile when a mask is present.

CONT recognizes intrinsically masked dispositions on these same grounds. Masks that are intrinsic to an object are not relevantly different from those that are extrinsic. Consider Ashwell’s (2010) poisonous berry once more. The berry is disposed to cause harm when ingested, but that disposition is masked by the berry’s indigestible skin. Under CONT, the berry is poisonous because it has a property in virtue of which it makes a contribution that sufficiently favors it acting in a manner associated with poisonous things. Suppose that is the property of having flesh with a certain chemical composition. The berry’s event counterparts in that contribution all share that property, but otherwise vary in all the ways permitted by the laws, both intrinsically and extrinsically. Many of those counterparts will share the berry’s indigestible skin. Many others will not. And the latter will reliably cause harm when ingested. So even if none of the berry’s counterparts with indigestible skin cause harm when they are ingested, it is enough that a suitable contingent of those without it do. Should that be the case, the berry makes a contribution that significantly favors causing harm when ingested and so plausibly is poisonous under CONT.

This result holds even if having flesh with this particular chemical composition is often, or even always, the result of processes that reliably lead objects with that flesh to have indigestible skin. For even in common cause scenarios like this, it is nomically possible for both effects to occur on their own (Paul & Hall 2013). Suppose that events of type A always occur alongside events of type B on account of having a common cause. It is nonetheless nomically possible for an A-event to occur without a B-event. Provided A-events and B-events are otherwise independent of one another, nomically possible worlds may simply begin in states where A-events occur in the absence of any B-events and then evolve under the actual laws. Worlds like this are part of an event’s contribution. So regardless of the causal history that leads the berry to have its particular kind of flesh and an indigestible skin, the contribution that the berry makes in virtue of having that kind of flesh will include a significant range of worlds in which its event counterparts lack an indigestible skin as part of the initial conditions. That contribution may still favor causing harm when ingested to such a degree that the berry qualifies as poisonous.

CONT’s treatment of masks applies equally well to finks. Like masks, finks prevent dispositions from manifesting in the face of their stimulus. Unlike masks, they do so by removing those dispositions from their bearers. In response to a disposition’s characteristic stimulus, a fink changes an object such that it no longer has the disposition in question. For example, Martin’s (1994) classic example of finked conductivity imagines a device that would render a conductive wire non-conductive when touched by an electrical charge. CONT will recognize the loss of those dispositions. By changing what properties an object has, a fink will also change what contributions that object makes. Under CONT, finks remove dispositions from objects by removing the properties in virtue of which they make the relevant contributions. However, since the fink only causes that change in response to the stimulus, CONT will correctly recognize a disposition in the mere presence of a fink. So long as the object makes a contribution that confers a disposition, CONT will recognize that the object has that disposition.¹⁴

5 Opposing Dispositions

This account sheds light on an important class of intrinsically masked dispositions: cases in which an intrinsic mask gives rise to an opposing disposition. Something has opposing dispositions when it is disposed both to M (when C) and to N (when C), where M and N are mutually incompatible outcomes (Choi, 2013). For example, imagine that John has a long-standing disposition to ramble when he is nervous. He has recently resolved to stop and with significant, sustained effort has been successful. When his resolve holds, he is actually less prone to ramble than the average person. But his rambling will begin again if his resolve ever falters. John plausibly has two opposing dispositions. His continued resolve is required because he remains disposed to ramble when he is nervous. His resolve masks that disposition and confers the opposing disposition to not ramble when he is nervous.

Some critics of intrinsic masking find such claims to be puzzling and possibly even incoherent (Choi, 2012, 2013). Their concern, it seems, is that to ascribe opposing dispositions to something is to assert an apparent contradiction. CONT can help to dispel that appearance. It accounts for having opposing dispositions individually, but also having them together.

Contributions interact with one another (Johansen, 2015). Just as one event makes a contribution individually, multiple events make a contribution jointly. The joint contribution of multiple events is the constraint they collectively impose on how the state of the world can evolve. It is the set of nomically possible worlds in which they all occur or, equivalently, the intersection of their individual contributions. For example, if an event a makes the contribution {w₁, w₂, w₃} and an event b makes the contribution {w₂, w₃, w₄}, the joint contribution of a and b is {w₂, w₃}. So understood, contributions build on one another in a principled way. The contribution made by a permits the state of the world to evolve as it does in world w₁. The contribution made by b does not. b’s contribution is an additive constraint on a’s contribution. It closes off possible futures that a’s contribution leaves open.

Contributions may favor events to a greater or lesser degree collectively than they do individually. Suppose that a contribution C favors an event e. The proportion of e-worlds in C may be very different from the proportion of e-worlds in the intersection of contributions C and D. Should e occur in a greater proportion of worlds in the intersection of C and D than it does in C, we can say that D supports C’s favoring e. The two contributions favor e collectively to a greater degree than C individually. Should e instead occur in a smaller proportion of worlds in the intersection of C and D than it does in C, there is a very real sense in which D has undermined C’s favoring e. The contributions favor e to a lesser degree collectively than C does individually.

We can take dispositions to interact in this same fashion. Under CONT, having a disposition consists of making a contribution that sufficiently favors its associated behavior. To support or undermine a contribution in favoring that behavior is to similarly support or undermine the disposition that it confers. Suppose that a contribution C confers a disposition. If that disposition is characterized solely by its manifestation condition, a contribution that supports or undermines C in favoring that manifestation also supports or undermines that disposition. If that disposition is instead characterized by stimulus and manifestation conditions, it is supported or undermined when a contribution supports or undermines C’s favoring that manifestation in those conditions where the stimulus is obtained. Should the contribution that plays this supporting or undermining role also confer a disposition, we can attribute that supporting or undermining role to the conferred disposition as well.

Consider John and his disposition to ramble and not ramble when he is nervous. CONT holds that John has those dispositions because he makes a pair of corresponding contributions. One contribution sufficiently favors his rambling when he is nervous. The other sufficiently favors his not rambling when he is nervous. Call the former contribution pro and the latter contribution con. How those contributions interact will shape the interactions of the dispositions they confer. Assuming that John’s resolve would reliably prevent him from rambling when he is nervous, con plausibly undermines pro’s favoring John rambling when he is nervous to a significant degree. John’s event counterparts in pro all share the property that confers his disposition to ramble when he is nervous. In contrast, his event counterparts in the intersection of pro and con share both that property and his resolve. As such, the counterparts in the intersection of those contributions plausibly ramble in a much smaller proportion of the worlds where they are nervous than do the counterparts in all such worlds in pro. If so, John’s disposition to not ramble when he is nervous not only opposes but undermines his disposition to ramble when he is nervous. The two dispositions interact in a way that leaves John significantly less apt to ramble when he is nervous, all things considered.¹⁵

Opposing dispositions on this view exhibit a genuine tension. The manifestation of one is incompatible with the manifestation of the other. But this tension is perfectly coherent. To ascribe opposing dispositions to something under CONT is to acknowledge its competing influences. Opposing dispositions arise from contributions that sufficiently favor incompatible events. While those events are incompatible, the contributions that favor them are not. So understood, there is no contradiction in ascribing opposing dispositions to one and the same thing. Each disposition corresponds to an influence that sufficiently favors its manifestation and, in doing so, weighs on the other. So while it remains true that John is disposed both to ramble when he is nervous and not ramble when he is nervous, to leave it at that is to under-describe the case. Those opposing dispositions interact such that the latter disposition significantly undermines and masks the former. Undermining a disposition in this sense does not remove it. John is still disposed to ramble when he is nervous. But that disposition has been countered by an opposing disposition such that there is no mystery in the fact that it reliably fails to manifest in the face of its stimulus.

6 Liberalism

To close, I want to address an objection that is likely waiting. CONT is a liberal theory of disposition ascriptions. It ascribes dispositions that other theories would not. We might reasonably worry that CONT is too liberal.

Consider an example from Choi (2017). Suppose that food scientists discover that every grain of rice contains both a lethal chemical compound and an antidote to that compound that prevents its harmful effects. The two elements are mixed together such that they cannot be separated under normal circumstances. For CONT and similarly liberal theories, this would amount to the surprising discovery that ordinary rice is in fact poisonous. Every grain of rice has a property in virtue of which it makes a contribution that favors it harming someone when they ingest it. Its poisonous nature is systematically masked by an intrinsic antidote. Choi (2017) objects that recognizing dispositions like this undermines the practical importance of disposition ascriptions.¹⁶ Dispositional concepts, he suggests, primarily serve our practical interests. We ascribe dispositions to things in order to predict and explain their behavior so that we can better navigate the world. We classify some things as poisonous and others as not poisonous, for instance, in order to know what is and is not harmful to ingest. But if dispositions can be intrinsically masked in cases like Choi’s (2017), knowing whether something has a particular disposition may require us to know a great deal about its internal structure. We often will not know much about that structure in the everyday contexts in which we rely on and seem to effectively use disposition talk.

Similar challenges arise for other liberal metaphysical views. For example, mereological universalism holds that for any two material objects, there is a third object that they compose. If universalism is correct, there are countless objects that we have no practical interest in and do not ordinarily recognize in our thoughts and talk. Even so, universalists generally view their theory as compatible with our ordinary judgments about what exists. Following Lewis (1986b), they hold that ordinary thought and talk restrict our domain of quantification to the objects of common sense. Just as I might say that there are no more bananas but mean only that there are no more bananas in the kitchen, I might say that nothing is composed of the moon and my left foot and mean only that there is no such thing among the objects of common sense.

Liberal theories of disposition ascriptions can adopt a similar response. Our practical interest in dispositions typically concerns their manifestations. We care whether something is poisonous, for instance, because we care about the harm it may do and the conditions in which such harm is apt to occur. As a practical matter then, there is often no benefit in distinguishing objects that strictly lack a disposition from those that have a disposition that is intrinsically masked such that it would not manifest under ordinary circumstances. Their behavior in ordinary circumstances is the same. Under CONT then, it often is not relevant to our practical interests whether something has a disposition full stop. Rather, our concern in such cases lies with those dispositions that are free from intrinsic interference that would prevent them from manifesting under ordinary conditions. Ashwell (2010) calls these superficial dispositions. Unlike dispositions broadly, we can recognize superficial dispositions without knowing much at all about a thing’s internal structure. CONT suggests a simple account of superficial dispositions: An object o has a superficial disposition to M (when C) iff the event constituted by o’s having all of its intrinsic properties makes a contribution in which o’s event counterparts M in a sufficient proportion of worlds (where C).

Under a liberal theory of disposition ascriptions, it is reasonable to suppose that in everyday thought and talk we use dispositional predicates to ascribe superficial dispositions when it better serves our practical interests. In the world of Choi’s (2017) rice example, when we tell our dinner guests that the rice we have served is not poisonous, what we mean is that the rice is safe to eat—that it does not have a superficially poisonous disposition. Using disposition-talk in this way when it better serves our practical interests is consistent with regarding liberal theories like CONT as strictly true. Should a guest point out that there is a lethal chemical compound in rice, we may very reasonably acknowledge that rice is indeed poisonous, simply not in a way that must concern us.

Regardless, practical considerations are not decisive. Even if intrinsically masked dispositions served no practical purpose, we may have very good reason to recognize them. CONT’s liberalism is motivated. My argument for CONT has primarily proceeded from the bottom up. I have not argued for CONT on the grounds that it successfully navigates a battery of pre-theoretic intuitions about cases. Rather, I have argued that CONT recognizes a kind of disposition-like influence that we in fact find in objects. I have assumed here that events make contributions in the manner outlined by Johansen (2014) and shown that those same contributions can be attributed to objects. I then argued that contributions may favor outcomes in a manner that reflects the gradable modalities that characterize dispositional relationships. If that is correct, then whether or not we accept CONT as a theory of disposition ascriptions, we are left with the conclusion that objects have an influence on the world that shares the modal character of dispositions and that they have that influence even in the presence of intrinsic factors that may prevent them from manifesting the associated outcomes. Just as something having the power to M when C is considerable reason to regard it as disposed to M when C, something having a disposition-like influence is considerable reason to regard it as having a corresponding disposition. The alternative is that something may have a disposition-like influence but lack the corresponding disposition. If that is the case, then the question of whether dispositions can be intrinsically masked becomes largely terminological. It becomes a question about how we ought to use dispositional predicates rather than what modal qualities something actually has.

I take these considerations to weigh in favor of CONT’s liberalism. At the same time, they also suggest that CONT may do valuable work for someone who would follow Choi (2017) in rejecting liberal theories of disposition ascriptions on practical grounds. Rather than framing CONT as a theory of disposition ascriptions, they could recast it as a theory of disposition-like influence. Anyone concerned with intrinsically masked dispositions on account of their application in dispositional theories of various phenomena would be free to reframe their preferred theory in terms of contributions. Instead of asking if something has a given disposition, they could ask if it makes a corresponding contribution and if that contribution served the needs of their preferred theory.