Actualism Doesn’t Have Control Issues: A Reply to Cohen and Timmerman

Abstract

Recently, Cohen and Timmerman (Journal of Ethics & Social Philosophy, 10(3), 1–18, 2016) argue that actualism has control issues. The view should be rejected, they claim, as it recognizes a morally irrelevant distinction between counterfactuals over which agents exercise the same kind of control. Here we reply on behalf of actualism.

1 Introduction

Recently, Cohen and Timmerman (2016) have offered a novel challenge to the view known as

Actualism. At t an agent S morally ought to φ at t’ iff, at t, S can φ at t’, and what would happen if S were to φ at t’ is better than what would happen if S were to ~φ at t’. (2016: 2)

Actualism, they maintain, has control issues. It recognizes a morally irrelevant distinction between counterfactuals over which agents exercise the same kind of control.

Here we reply on behalf of actualism. In section one we reproduce Cohen and Timmerman’s central argument. In section two we show it to be unsound. It relies crucially on the false claim that an agent exercises identical control, on more than one occasion, over the performance of an act. Section three concludes.

2 Against Actualism

To make Cohen and Timmerman’s challenge vivid, consider the case they offer:

Sadistic Suzy

Suzy is a sadistic doctor who is responsible for the well-being of Sick Sid. At t₀, Suzy can do the following:

• Completely cure Sid by giving him two shots of the complete cure (one at t₁ and one at t₂).

• Partially cure Sid by giving him two shots of the partial cure (one at t₁ and one at t₂).

• Let Sid die by giving him only one shot, giving him mixed shots or giving him no shots.

If Sid is completely cured, he will live a long and happy life for n years. If Sid is partially cured, he will live for 1/2n years. Under any other act-set that Suzy can perform, Sid dies immediately thereafter at t₃. (2016: 3).

Suppose further that the following statements are true:

Early. If Suzy were to deliberate, at t₀, she would give Sid no shot at t₁.

Late. If Suzy were to give Sid a shot of the complete cure at t₁, she would give Sid no shot at t₂.

Deliberation: Suzy, at t₀, is deliberating.

Finally, the crucial stipulation:

Identical Control. “The kind of control Suzy has at t₀, over all act-sets she can perform from t₁ to t₂ is the same” (2016: 4).

The problem for actualism should now be in view. The trouble is that Early and Late – counterfactuals over which it is claimed Suzy enjoys the same kind of control – figure differently in actualism’s account of what Suzy ought to do. Suzy could freely administer the complete cure at t₂, but she won’t. For, if she were to administer the first shot of the complete cure at t₁, then, given Late, she would not administer the second shot of the complete cure at t₂. So, actualism recommends that Suzy administer the first shot of the partial cure at t₁. In this way, Late – a claim about what, given certain circumstances, Suzy would freely do in the future – makes a difference to what she ought to do now. Early, by contrast, makes no such difference. Actualism requires that Suzy administer the first shot of the partial cure at t₁ even though, given Deliberation, she will freely do no such thing. It appears, then, that actualism regards Early and Late as morally different, even though, as stipulated by Identical Control, Suzy exercises the very same kind of control over both. Cohen and Timmerman state the problem for actualism succinctly: “[I]n order to determine what, at t₀, Suzy ought to do at t₁, it doesn’t matter what Suzy can (at t₀) do at t₂. Rather, what matters is what Suzy would do at t₂ were she to do something or other at t₁” (2016: 4). Late matters in determining what Suzy ought to do; Early does not. But it seems there’s no good explanation for this differential treatment.

The problem for actualism, demonstrated by Sadistic Suzy, can be generalized. Cohen and Timmerman thus maintain that actualists are objectionably committed to

Actualism’s Consequence. For any obligation o agent S has at t₀ to φ at t₁, and for any early counterfactual,¹ and for any late counterfactual,²o is not even partly determined by the early counterfactual(s), but is at least partly determined by the late counterfactual(s). (2016: 5)

This consequence does indeed appear troubling. In what follows, however, we show this appearance to be illusory.

3 Against Identical Control

Cohen and Timmerman’s challenge hinges on their crucial stipulation: Identical Control. This is essential to generating the problem. It is only because Suzy is thought to exercise the same kind of control over both Early and Late that Actualism’s Consequence seems objectionable. But this stipulation,, we contend, cannot be true.

We can start by noting that the following is an attractive feature of any account of control:

Counterfactual Power. If S has control over φ-ing, S possesses characteristics sufficient to determine whether or not S φs.

Counterfactual Power captures the idea that having control over an act entails the ability to dictate the facts concerning the performance of the act. It is intended to be neutral among differing accounts of control. Cohen and Timmerman, for example, endorse the similar thought that, “If an agent S has the ability to do otherwise, it follows that S has a kind of control over the truth-value of certain counterfactuals” (2016: 3–4).

Two observations are in order. First, Counterfactual Power is, in part, what individuates acts. It is precisely because Suzy has this power at multiple times that she performs multiple acts. Suppose, for example, at t₀, you ingest a mind-controlling drug that causes you to type ‘t’ at t₁, ‘h’ at t₂, and ‘e’ at t₃. Here you enjoy control only of the act, at t₀, of taking the drug. The typing of the letters, because you lack Counterfactual Power over each letter press, is merely a part of the act of taking the drug at t₀. The presses that follow are not distinct acts. Second, one’s exercise of Counterfactual Power need not occur immediately prior to the performance of the act. Suppose the mind-controlling drug you ingested at t₀ is of the slow release variety. You’ll start typing at t₁₀. This lag is perfectly consistent with the supposition that you enjoyed Counterfactual Power over typing at t₀. As long as one has a characteristic sufficient to ensure that some act is performed – regardless of when its performance takes place – one has Counterfactual Power over that act.

Given this pair of observations there is nothing objectionable in Cohen and Timmerman holding that Suzy, at t₀, enjoys Counterfactual Power over both Early and Late. What is objectionable, however, is their suggestion that Suzy possesses Counterfactual Power over a single act at more than one time. More to the point, what is objectionable is the claim that she exercises Counterfactual Power over Late at both t₀ and t₁. For Counterfactual Power implies

Single Exercise. An agent can exercise control over the performance of an act at most once.

As long as one can later exercise Counterfactual Power over some act, one cannot earlier exercise that same power over that same act. Single Exercise follows from the fact that if one possesses, at t₁, characteristics sufficient to determine whether or not she φs at t₂, then one will lack, at t₀, characteristics sufficient to determine whether or not she φs t₂. One’s Counterfactual Power, at t₁, strips one of such power at any time prior.

To illuminate how Counterfactual Power implies Single Exercise, consider a similar power: consent. Suppose you’ll undergo surgery at t₂. If you grant consent, then it will be an operation. If you don’t, it will be battery. Now, suppose that your doctor will request your consent twice, at t₀ and at t₁. While, by virtue of your consent, you can determine whether the surgery is an operation or battery, you exercise this power only once, at t₁. Whatever you choose at t₀ is superseded by your choice at t₁. Even if we suppose that you consent at t₀, and further suppose, at t₁, when your consent is again requested, you again grant it, you still determine whether the surgery is an operation or battery only once, at t₁. While it’s true that you choose not to exercise your power to change your plan (by revoking your consent), you nonetheless do exercise your power to make the surgery an operation rather than a battery (by reaffirming your consent).³

That an agent has control over an act at most once, nonetheless, may be thought objectionable. It appears to be in tension with an assumption commonly accepted by participants in the debate over actualism and possibilism: Namely, that there are true counterfactuals of freedom – i.e., facts about what agents would freely do if placed in certain circumstances.⁴

Appealing to such counterfactuals, one may offer the following objection. Suppose the below statements are all true:

• i. If S were to φ at t₀, then S would λ at t₂.

• ii. If S were to ~φ at t₀, then S would ~λ at t₂.

• iii. If S were to ψ at t₁, then S would λ at t₂.

• iv. If S were to ~ψ at t₁, then S would ~λ at t₂.

What Happens: S, in fact, performs φ at t₀, ψ at t₁, and λ at t₂.

Assuming that all of the acts above are under S’s control, it would appear that S controls whether or not to λ on more than one occasion. If S, at t₀, elects to φ then, given (i), she would λ at t₂. If she elects to do otherwise, given (ii), she would ~λ. Later, if S, at t₁, elects to ψ then, given (iii), she would λ at t₂. If she elects to do otherwise, given (iv), she would ~λ. Hence, it appears that the truth of (i) - (iv) is inconsistent with Single Exercise.

Yet, this objection fails. Suppose we assume that S has Counterfactual Power over φ-ing, ψ-ing, and λ-ing at t₀, t₁, and t₂, respectively. Since Counterfactual Power implies Single Exercise, if the objection is correct, this stipulation should be inconsistent with the truth of (i) - (iv). However, given What Happens, we know that S will φ at t₀, ψ at t₁, and λ at t_2. This is how, we can imagine, she elects to use her Counterfactual Power over these acts. And this is all that’s needed to secure the truth of (i) and (iii). Because, if it is true that S φs at t₀ and it is true that S λs at t₂, then the following is true: If S were to φ at t₀, then S would λ at t₂. This follows from Conjunction Conditionalization – (A ∧ C) → (A > C) – which is an axiom of Pollock’s SS (Pollock 1976: 42–43) and a theorem endorsed by both Lewis (1986: 132) and Stalnaker (1968, 1987: chp. 7).⁵ The same reasoning allows us to secure (iii).⁶ To get (ii) and (iv), it simply needs to be that, on the closest possible world, the reverse of What Happens occurs: S elects to use her Counterfactual Power to ~φ at t₀, ~ψ at t₁, and ~λ at t₂. The assumption of Counterfactual Power over each act is consistent with (i)-(iv). And hence, the appeal to true counterfactuals of freedom thus does nothing to impugn Single Exercise.

We are now in a position to reassess Cohen and Timmerman’s challenge. Recall, actualism is thought objectionable because it illicitly “demarcates between counterfactuals that do and do not determine S’s obligations, even in cases in which, at t₀, S has the same kind of control over what S does both at t₁ and t₂, and thus has the same kind of control over the truth-value of both [Early and Late] counterfactuals” (2016: 5). But this suggests that Suzy exercises Counterfactual Power over her act at t₂ twice, both at t₁ and at t₀. This cannot be so. Here’s the argument.

• 1. If S has Counterfactual Power, at t₀, over S’s φ-ing at t₂, then S, at t₀, possesses characteristics sufficient to determine whether S, at t₂, φs or not.

  • a. Closer to Cohen and Timmerman’s words: S has the ability, at t₀, to determine the truth-value of certain counterfactuals, namely, whether or not she φs at t₂.

• 2. If, at t₀, S possesses characteristics sufficient to determine whether S, at t₂, φs or not, then nothing between t₀ and t₂ could prevent S from φ-ing at t₂.⁷

  • a. If something could, between t₀ and t₂, prevent S from φ-ing at t₂, then it would not be the case that S possesses characteristics sufficient to determine the truth-value of the relevant counterfactuals, namely, whether or not she φs.

• 3. Identical Control holds that S has Counterfactual Power, at t₀, over S’s φ-ing at t₂.

• 4. Hence, Identical Control implies that nothing could, between t₀ and t₂, prevent S from φ-ing at t₂. (from 1,2,3)

• 5. If S has Counterfactual Power, at t₁, over φ-ing at t₂, then S, at t₁, possesses characteristics sufficient to determine whether S, at t₂, φs or not.

  • a. Closer to Cohen and Timmerman’s words: S has the ability, at t_1, to determine the truth-value of certain counterfactuals, namely, whether or not she φs at t₂.

• 6. If S, at t₁, possesses characteristics sufficient to determine whether S, at t₂, φs or not, then something – namely, S at t₁ – could, between t₀ and t₂, prevent S from φ-ing at t₂.

  • a. Possessing characteristics sufficient to determine whether she, at t₂, φs or not, implies that she could prevent herself from φ-ing at t₂ at t₁.

• 7. Identical Control holds that S has Counterfactual Power, at t₁, over φ-ing at t₂.

• 8. Hence, Identical Control implies that something can, between t₀ and t₂, prevent S from φ-ing at t₂. (from 5,6,7)

• 9. Hence, Identical Control implies a contradiction. (from 4,8)

If one stipulates Identical Control, one stipulates a contradiction. And once this stipulation is dropped, Actualism’s Consequence appears unobjectionable. This is sufficient to rebuff Cohen and Timmerman’s challenge.

4 Conclusion

Cohen and Timmerman conclude their essay with a request. They write, “We want to know why, at t₀, Suzy’s rotten moral disposition described in [Late] allows her to avoid incurring moral obligations that her rotten moral disposition described in [Early] does not, given that, at t₀, Suzy has (ex hypothesi) the exact same control over the truth-value of each counterfactual” (2016: 15). We have, on behalf of actualism, offered an answer. It’s simply false that, at t₀, Suzy exercises the same control over Early and Late. One exercises Counterfactual Power over the same act at most once. Thus, there is a morally relevant difference – a difference in control – that undergirds actualism’s differential treatment of Early and Late. So, in the end, we agree that Sadistic Suzy illustrates a problem. But the problem lies with Identical Control, not Actualism’s Consequence.