Interference and Prevention in Powers-Based Causation
I have so far argued that the standard view can deal with counterexamples where it is assumed that individual external objects (simple prevention and pre-emption), or individual factors, or powers, intrinsic to the objects involved (finks and antidotes), interfere and/or prevent the efficacy of the actions of individual Agents or powers. However, Mumford & Anjum have also addressed the question of whether the total set of powers involved in an instance of causation could necessitate an outcome (2011: Ch. 3). At first blush, their argument may seem to address a total cause conception of the kind I am promoting. But this is not the case. They characterise the total cause merely in terms of the obtaining of a set of powers, in a way that ignores the manner in which the powers are exerted in an interaction between powerful particulars. That makes quite a difference.
Mumford & Anjum argue that if a given effect E is assumed to be the necessary result of a combination of a set of powers, say [pp p2, pt, p4], then—for it to pass the test of necessity—whenever powers [p-p4] obtain we should always get E no matter what other factor > obtains with them. I acknowledge that, as a matter of empirical fact, this is not the case. If obtains we get E, but if [p-p4 +] obtains, we get E *. Their argument is logically valid. I just find that it doesn’t address the standard view, and as far as I know, the view they address has never been defended.
Admittedly, Hobbes writes that an entire cause is the ‘aggregate’ of all the powers of the interacting objects, which could be interpreted as a characterisation of causes merely as a certain combination of powers. However, one will then have to neglect Hobbes’ repeated claims that it is the body of the Agent that acts upon the body of the Patient, albeit in virtue of possessing various powers (Hobbes 1656: Ch. IX, §§1-4, 8). Obviously, what Hobbes has or hasn’t said isn’t really paramount here. We can just appeal to the empirical fact that the waypowers are distributed between interacting entities is important for the production of an outcome, not just the collected presence of powers at a place and time.
The standard view states that causes are interactions between powerful particulars; not the mere obtaining of a set of powers. It matters for the characterisation of the cause at hand to which particulars the powers [p-p4] belong. An interaction between particulars A and B where the powers [p-p4] are distributed as A[pt,p,] vs. B[p;,p4] is a different kind of cause than an interaction where the same set of powers [p-p4] are distributed as A[pp p,] vs. B[p„ p4] (A has instead of p„ and B has p, instead of pt). Therefore, one cannot argue without further qualifications that, for powers-based causation to be necessary, whenever powers are present one must always get E. One has to consider how the powers are distributed between the particulars.
Let us go through this in detail. Consider an interaction (<») between two particulars A Sc B with each a set of powers |p(, p,...], which then produces (d) an outcome E:
According to the standard view, this is one kind of interaction and so a particular type of cause leading to a certain effect, E. And it is different from the following interaction in which the same powers are present, but distributed differently between the particulars to produce the effect £ * (A has p, instead of p„ and B has p, instead of ps):
Atp,,/?,] <*B[p„p4] dE*. (2)
So even in different cases where we have the same powers present, we may not get the same effect if they be distributed between the particulars in different ways. Furthermore, neither (1) nor (2) are instantiated in the following interaction where we have [p-p4] + an additional power >•.
The last interaction is not an instance in which either (1) or (2) occurs, but are somehow interfered with by >; it is a different interaction, and so a different cause producing a different effect. Either (/> is a part of the interaction, in which case the only thing that is prevented is the occurrence of causes of kind (1) and (2), or (/> is introduced after or during the interaction, and in either case it comes too late to prevent the manifestation of an exertion of influence that has already taken place. Remember that the standard view takes causation to be a continuous process of mutations that continues as long as the objects are exerting an influence on each other. At any point in that process an exertion of influence will
Causal Necessity 97 have an immediate result, but interfere at any given point and the process morphs into another process through the addition of a participating powerful particular; one that will necessarily evolve a certain way.
Really, if we take into account the distribution of powers among interacting particulars, then the only way to introduce an additional power (/> to the set of powers [p-p4] is either by introducing an additional external particular, C, bearing the power >, as in the case of simple prevention and pre-emption discussed earlier, or we introduce > as already belonging to the Agent and/or Patient, in which case we have finkish and antidotal cases all over again.
I can’t see a way to interfere or prevent anything on this model. The only thing that could make interactions of this kind anything less than necessary is if we already postulate that the individual powers are slightly indeterminate in their contribution to an interaction. This seems to be what Mumford & Anjum mean in calling powers tendencies. However, their argument in favour of treating powers as tendencies is the argument that I have just discussed, notably that sets of powers do not necessitate any particular outcome because something can interfere with their production of effects.
It is possible that Mumford & Anjum do not think of their view as motivated by the interference and prevention argument alone, but rather see a theoretical advantage in the idea of a tendential modality in itself. After all, they claim that even in situations where ‘there is nothing that prevents the effect from occurring, but still, it need not occur’ (2018: 18). As Stathis Psillos remarks in his review of Mumford & Anjum’s most recent book, What Tends to Be, this move ‘seems to pile a new mystery on the old’ (Psillos 2019), notably of explaining how it could be that an influence is being exerted but nothing happens even if nothing is interfering.
Psillos speculates whether they are modelling the modality of tendencies on phenomena like radioactive decay; at any moment the atom has the tendency to decay—and nothing is there to prevent it from decaying— and yet it may not decay. Psillos is worried that idea turns on a problematic view of the probabilistic nature of decay, which I think is a valid worry. Indeed, one can worry about more than the probabilistic aspect of decay. If one thinks decay is a spontaneous non-causal phenomenon, because it is not triggered by an interaction between the atom and any other powerful particular, then it isn’t causal at all. If one instead thinks that we are dealing with an atom that has the power to decay, but doesn’t because it isn’t exercising the power, then we don’t have a case where a power is being exerted without producing an effect, but a case where a power isn’t exerted.
Anyway, if we confine ourselves to interactions between powerful particulars, then Mumford & Anjum’s tendential modality seems to allow for the possibility that two billiard balls could collide and fail to exerttheir powers on each other, or, that even if they do exert them, they don’t change each other’s state of motion because they fail to exert their passive power to change state of motion. To my mind, that would be rather mysterious.
Causal Necessity Without Ceteris Paribus Clause
I think it is important to point out that the powerful particulars view really amounts to an account of unconditional necessity without appeal to ceteris paribus clauses. This is because it includes in the notion of cause the components that others take to be mere conditions; the conditions that must stay the same to secure the connection between C and E. The ceteris paribus clause is meant to cover all these allegedly non-causal conditions. I think it follows from my argument above, that we don’t have to add a ceteris paribus clause to exclude the presence of any additional item (p that could interfere or prevent, because the presence of any such additional item would only constitute a failure of the obtaining of a certain kind of cause C rather than the prevention of an effect, and constitute the obtaining of a certain other kind of cause C*, which inevitably gives E*.
In this chapter I have argued (i) no causal realist has argued that causal connections are logically necessary, (ii) the problem of action at a temporal distance addresses a view that has never been defended, but was merely Russell’s mistaken conception of causal realism, and that the powerful particulars view is immune to the problem, (iii) that the problem of interference and prevention similarly addresses only relational realism while the powerful particulars view is immune to it, and (iv) that Mumford & Anjurn’s version of the problem of interference and prevention addresses a view of causes as sets of powers that has never been used to defend causal necessity, and that the powerful particulars view is immune to their arguments too. The conclusion must be that CASE does not amount to a serious challenge to the necessity of causal connections.