GLENN ROSS

W H E N D O WE D O W H A T WE D O ?

(Received 3 December, 1976) When did Booth kill Lincoln?" (Hint: Booth shot Lincoln on April 14, 1865, but Lincoln lingered through the night, dying on the fifteenth.) Still not sure? Suppose we adopt a Davidsonian analysis of actions whereby: (1)

If A does x by doing y , then A ' s doing x is the same as A ' s doingyJ

Since Booth killed Lincoln by shooting him, we would conclude that the action which was a killing is identical to the action which was a shooting. But if identical events happen at the same time, then the act o f killing occurred when the act o f shooting did (i.e., on April" 14). So Booth killed Lincoln before Lincoln died - a result which many have found counter-intuitive. John Vollrath in 'When Actions are Causes '2 argues that this paradoxical result can be avoided, for the acceptance o f (1) (labeled as 'the explanatory identity thesis') does not commit us to: (3)

If A does x by doing y , then A does x at the same time that A doesy.

for the missing premise which completes the enthymeme, viz., (2)

If A ' s doing x is the same as A ' s doing y , then A does x at the same time that A does y .

can plausibly be denied. It is m y thesis that even should we accept Vollrath's analysis of action descriptions, we have no reason to deny (2). Thus, the explanatory identity thesis will still commit us to (3), although (hopefully) it is less paradoxical than it seems at first blush. Vollrath claims that: ... if an event could begin to exemplify different action types at different times, we could hold that there is something that is both a shooting and a killing, but it was a shooting before it was a killing ... it may not have been a case of 'causinga death' until Philosophical Studies 32 (1977) 419-423. All Rights Reserved Copyright 9 1977 by D. Reidel Publishing Company, Dordrecht-Holland

420

GLENN ROSS

the day after it was a case of 'causing a gun to fire'. In this sense, the shooting was not at the same time as the killing2

Vollrath's proposal allows for the event that was to be the shooting to be identical with the event that was to be the killing, but the event undergoes a 'Cambridge' change 4 when Lincoln dies (analogous to the type of change Xantippe undergoes when Socrates dies - viz., becoming a widow). Such changes are to be distinguished from 'real' changes: a distinction which seems intuitive but as yet has not been made rigorous by anyone. It has been persuasively argued elsewhere s that such changes are not themselves events, and thus Vollrath's account does not force us to speak of events happening to events, which would lack clear sense. It is not obvious how such an account should lead us to reject (2). Vollrath's formulation of (2) was intended to capture one of Davidson's principles for individuating events: "... if events are identical they consume identical stretches of time", 6 or, as Vollrath paraphrases it, that identical events happen at the same time. Suppose that we adopt Vollrath's account and, in an irenic temperament, allow that the event which is the trigger-pulling later becomes a shooting (as the bullet leaves the gun and enters Lincoln's body), and even later becomes a killing (when Lincoln dies). In this loose sense, Vollrath claims, the killing is later than the shooting. Notice that there is a shift to the use of singular terms here. Now it is a crucial feature of Davidson's analysis of action sentences that they are existential and general there is no singular term referring to an event in either: (4)

Booth shot Lincoln.

(5)

Booth killed Lincoln.

or

Instead, they are analyzed as: (4')

(Ex) Shot

(Booth, Lincoln, x)

and

(5')

(Ey) Killed

(Booth, Lincoln,y) 7

If we wish to use a singular term such as 'the killing' it must pick out the unique (for the context) object in the domain of discourse which accounts

WHEN

DO WE DO WHAT

WE DO?

421

for the truth of the existentially generalized sentence, (5'). Vollrath concedes that the event so picked out can also account for the truth of (4'). Yet, if we are to ask when the event picked out by both singular terms occurred, the most plausible candidate seems to be April 14. Thus, although it may be the case that the shooting on April 14 does not become a killing until April 15, nevertheless, since Lincoln died, the shooting is the killing. We can then interchangeably substitute these singular terms in extensional contexts. A denial that the question 'When did the killing occur?' has the same answer as 'When did the shooting occur?' is a flagrant violation of Leibniz's Law, once we have conceded their identity. To make the point more intuitive, we can even use Vollrath's own example. Suppose that an individual (Beth) becomes the queen only after reaching adulthood. Once she is coronated, however, the singular term, 'the queen of England' picks out the same individual as 'the daughter of George VI'. So we can (after she becomes the queen) meaningfully ask when the queen was born, even though she was not born a queen. Similarly we can now maintain that Booth's killing occurred, a day before it became a case of a killing. Booth performed an act on the fourteenth, and this remains the case no matter how we describe that act (e.g., in terms of its various effects: Lincoln's death, the entering of a bullet into Lincoln's body, or the mourning of a nation). Thus, we can maintain Davidson's principle that identical events happen at the same time, while allowing them to undergo Cambridge changes after they have expired. 8 More formally, we can express the identity of a shooting with a killing in these terms:

(6)

(Ex) [Occurs (x) & Killed (Booth, Lincoln, x) & Shot (Booth, Lincoln, x]).

The time at which an event falls under the predicates may vary between the predicates, but if we are to ascertain when any event which makes (6) true occurs, we need only know the time at which it belongs to the extension of the predicate 'occurs'. 9 Thus, a distinction arises between when an event under any description occurred, and when it becomes a case of an event so described. ~~ We can thereby combine Vollrath's account with a Davidsonian analysis, and avoid denying (2). Indeed, if we are to take seriously an ontology of events such as Davidson suggests, then it does not seem any more paradoxical that such entities can undergo Cambridge changes

422

G L E N N ROSS

t h a n t h a t t h i n g s like X a n t i p p e can. W h a t w o u l d b e p a r a d o x i c a l , h o w e v e r , is t o claim t h a t i d e n t i c a l events c a n h a p p e n a t d i f f e r e n t times.

University o f Arizona

BIBLIOGRAPHY [1] Davidson, Donald: 'Agency', in R. Binkley, R. Bronaugh, and A. Maras (eds.), Agent, Action, and Reason, University of Toronto Press, (Toronto, 1971), pp. 3 - 2 5 . [2] Davidson, Donald: q?he Individuation of Events', in N. Rescher, et aL (eds.), Essays in Honor o f Carl G. Hempel (D. Reidel Publ. Co., Dordrecht, 1969), pp. 216-234. [3] Davidson, Donald: 'The Logical Form of Action Sentences', in N. Rescher (ed.), The Logic o f Decision and Action (University of Pittsburgh Press, Pittsburgh, 1967), pp. 8 1 - 9 5 . [4] Helm, Paul: 'Are "Cambridge" Changes Non-Events?', Analysis 35 (1975), 1 4 0 144. [5 ] Lewis, David: 'The Paradoxes of Time Travel', American Philosophical Quarterly 13 (1976), 145-152. [6] Vollrath, John: 'When Actions Are Causes', Philosophical Studies 27 (1975), 329-339.

NOTES i See [1], pp. 1 8 - 2 3 . 2 [6], pp. 3 3 4 - 3 3 5 .

3 Ibid. 4 Vollrath does not use this term, but from his discussion in [6] pp. 336, 337, it is clear that it applies. s See [4]. 6 [ 2 ] , p . 229. See [31,P. 92. 8 Since the trigger-pulling does not become a case of a killing (causing Lincoln's death) until after it occurred, it might appear that such changes are changes in the causal relations between events. But, on Davidson's central criteria for individuating events (viz., identifying their place in the causal nexus), this would alter the identity of the event(s?). However, this is not so. The causal relations between events remain unaffected after an event undergoes a Cambridge change, but we can now truthfully express the relations the event has to other events using a new singular term to refer to that event. As David Lewis remarks in [5] "Not just any old reversal in truth value of a time-sensitive sentence about something makes a change in the thing itself". (p. 146) 9 The predicate 'occurs' serves a dual role here: (i) it avoids sortal quantification (in this case, the employment of event-variables), since it would not seem that anything but an event can occur in the relevant sense; and (ii) it allows us to specify the time of the event, for the time at which the event belongs to this predicate's extension is identical with the time stretch which an event consumes.

W H E N D O WE DO W H A T W E D O ?

423

10 This distinction is in accordance with ordinary discourse. Consider the following discussions: A: I just roughed h i m up a bit. I d i d n ' t kill nobody! B: You have now - y o u r 'friend' j u s t died. [Later] B: He killed h i m all right. C: When did he do it? B: A t midnight last night, in his b a t h r o o m , with a knife.