You are on page 1of 26

rusro HILPINEN

ON STATES,

ACTIONS,

OMISSIONS

AND NORMS

Philosophers, lawyers and detectives have been puzzled by the problem of locating actions in time and space. The great detective Nero Wolfe observed (Stout 193511982, 16):
The average murder, I would guess, consumes ten or fifteen seconds at the outside. In cases of slow poison and similar ingenuities death of course is lingering, but the act of murder is commonly quite brief.

Wolfe seems to think that an act of murder by poisoning takes place at the time when the murderer (for example) pours poison into the victim's drink; the result which makes that individual action an act of murder may occur much later. The time of the poisoning is the time when the murderer interacts with the external world in a way which causes the victim to die. According to Wolfe, the act of murder consists in this case in the murderer's pouring of the poison into the victim's glass.' . If the spatial location of an action is determined in an analogous way, we should say that an act of murder occurs in the place where the perpetrator acts (for example, moves his hand). and this may be quite distant from the location of the victim's death. Questions concerning the location of actions are not always merely conceptual exercises, but can be of considerable practical importance, for example, when a court has to decide whether a given act has occurred within itsjurisdiction, In 1859 a New Jersey court considered a case in which the victim died in New Jersey from blows struck by the defendant in New York, and decided that the state of New Jersey had no jurisdiction over the case, because "no 'act' of the accused took place in New Jersey" (Cook 1942, 9). Like Nero Wolfe, the New Jersey court identified the act with the agent's bodily movements. However, other courts have adopted a different view on this matter. In a case in which the accused stood on the North Carolina side of the state border, and shot and killed a person who was in Tennessee, a North Carolina court held that "in legal contemplation" the act of killing took place in Tennessee, because the action "became effectual" in Tennessee (ibid., p. 12). This view seems reasonable insofar as the effect (or the result) required for the characterization of the act as a murder was actualized in Tennessee, even though the perpetrator pulled the trigger in North Carolina. "In legal contemplation", the act was not completed until the victim was dead. According to W.W. Cook (1942,9), there seems to have been a time when, according to

83
G. Holmstrom-Hinlikka IJ1Id R. Tuomeia (eds.), COfItemporary Action Theory. Vol. I, 83 -107. @) 1m Kluwer Academic PubUshers. Printed in the Netherlands.

84

RISTO HILPINBN

the English law, a murderer could not be punished unless both the blow and the death occurred in the county where the defendant was prosecuted. However, Cook notes that "the accuracy of [this} statement is not entirely clear", and from the moral and legal standpoint this requirement would have been a highly unsatisfactory way of avoiding the problems of localizing actions. In judicial contexts it is reasonable to adopt the view that the location of the act can be determined either by the actor's bodily movements or by the legally relevant effects of such bodily action (Cook 1942, 10-12).2 According to this view, the location of an act includes both the location of the actor's movements and the place where the action "takes effect", that is, where the result of the act is actualized. In the recent philosophy of action, many critics of the Wolfean view have adopted this conception of acts and argued that the: location of an act includes the location of the event or events on the basis of which the act is described and identified (Thomson 1971, 116-119; 1977, 47-60; Thalberg 1977, 101-111)?
II

The problems mentioned above have reappeared in the recent formal theories of action. In his book Norm and Action (1963) Georg Henrik von Wright observes that actions (or acts) usually involve changes in the world:
Many acts may ... be described as the bringing about or effecting ('at will') of a change. To act is, in a sense, to interfere with the 'course of nature' (p. 36).

von Wright notes: "To every the world", and adds:

act •..

there corresponds a change or an event in

the terms 'change' and 'event' must then be understood in a broad, generalized sense, which covers both changes (events) and not-changes (not-events). (!bid., p. 39~)

By this he means that the change associated with an act need not be an actual change, but a merely possible change, that is, a change which would have occurred if the agent had not been active. Von Wright analyzes actions in terms of three world-states or occasions: (i) the initial state which the agent changes or which would have changed if the agent had not been active (had not interfered with the course of nature), (ii) the end-state or the result-state which results from the action (von Wright 1963, 28), and (iii) the counter-state which would have resulted from the initial state without the agent's interference, in other words, the state which would have resulted from the agent's passivity. The counter-state is needed for expressing the "counter ... factual element" of action (von Wright 1968, 43-44). The characterization of acts by means of three states or occasions makes it possible to distinguish 23 = 8 different modes of action with respect to a single proposition or state of affairs p. These modes of action may be defined as

d(u» = V(p.u)'. Sustaining the state that p.~. 4 In this simplified terminology.ON STATES.e(u» and V(P. u d(u) e(u) Mode of action Bringing it about that p.I::::'~'" ~_.u)=l' (where '1' means the value true) will be abbreviated 'u F p'. if u F =p. .) Letting il become the case that p. d(u) Fp and e(u) F =p.. and let e be a function which assigns to each u E W the corresponding counter-state. Let d be a function which assigns to each u E Wa state which results from the agent's activity at u... For the sake of brevity. otherwise the agent may be said to be passive with respect to p. (Destroying the state that p.e(u».f_. ::. '. we can say that Actl is an act of producing p. ~_. Letting it become the case that not-po Bringing it about that not-p. and say (for example) that the agent brings about or produces the state p. Sustaining the state that not-p.~~ is"'!. but not conversely. ACTIONS. (Producing the state that p.• .I. if p is false at d(u) under otherwise similar circumstances. For example.d(u» -::/:V(P. the action in question is a productive or a I _.) Letting it remain the case Ihat p.} '. . in this'case:'-tlie -agent-iS-aCtive with respect to p. The truth-value of p at u is denoted by 'V(P.J. OMISSIONS AND NORMS 85 follows: Let W = {u. Table 1. If V(P. } be a set of possible world-states or occasions. The action-types in which V(P. and (as usual) 'V(P. the truth-value ofp depends on the agent's activity..•~ . we can say that the agent omits to bring it about that p. thus an omission entails non-performance. •. v.II act of destroying p. .. "" . If(V(p. In this case p becomes true as a result of the agent's activity: without the agent's action it would have remained false thatp.. .) Letting it remain the case that not-p. (Suppressing the state that p.'. Act3 is an act of prese~1!£'P!_ __ ~~.~.u). . and let us assume that the agent can be in a given state either active or passive.i .. I shall sometimes speak about states of affairs as if they were objects.. The falsity of p at the initial state and at the counter-state constitute an opportunity for the agent to bring it about that p.~ ~: 'I. '~. In this way we obtain the action possibilities presented in Table 1. An omission should be distinguished from the non-performance of an act: an agent can omit an act only in a situation in which he has an opportunity to perform the act in question.. '-' ~'._... we can say that the agent brings it about that p or produces the state of affairs that p.d(u» -::/: V(P.t_ ! -".d(u» -::/:V(P. I I '. ~ S .. instead of saying that an agent brings it about that p." are I f. If ~~ .e(u») omissions (abbreviated 'om'). The Main Action-Types according to von Wright. On the other hand. Action-logical expression Act1 Act2 Act3 Act4 ActS Act6 Act7 ActS "'p "'p p p p """'P Bp omBp Sp omSp P P "p p "'p "'p P "p P "'p """'P "p P P P B""']J omB""']J S"'p omS"p "p """'P P . . . W.

'omBp' and 'omSp' cannot be regarded as true or false at u .. According to this view. e Definition (ABvW) is not plausible if 'Bp' is read 'the agent brings it about that p'. that is. An agent can bring it about that P only if P is false on the initial occasion. are relative to occasions or world-states (1963.B) Bp ..) This suggests that the initial occasion should be regarded as the point of evaluation.e(u» and V(P.P is not a logical truth on the ground that Bp . In (1983) von Wright argues that the sentence (l. and (iii) u F =p. but if V(p. the action is an act of sustaining or preserving some state of affairs. whereas Bp -+ op is logically true.u). but strictly speaking. The interpretation of an action sentence involves three occasions. we should define (for example) the truth of 'Bp' (the agent brings it about that p) as follows:s (ABvW) u According (2) F Bp if and only if (i) d(u) Fp.86 RISTO HILPINBN destructive act. 'Sp'.B) is logically false... and the state or 'world' u is understood here as the situation which either is or is not changed by the agent's action. It should be clear that 'Bp' cannot be regarded as part of the description of the very situation which is changed (or kept unchanged) by the action in question. Should an action sentence be evaluated as true or false relative to the initial state or the result state . 195-196. (1983. But this is logically false. two of which (the initial state and the result state) are actualized in the course of the action.d(u» = V(P.p would say that if a state is produced on some occasion then it is (already) there on this occasion. :. if 'Bp' is regarded as a genuine action proposition which states that the agent does something. III -. It is convenient to say that the agent chooses to perform an action "at" the initial state (or situation) u: u is the state from which the action 'originates'.~ . the truth-values of sentences. to (ABvW). 23). (ll) e(u) F "'P. an action changes a situation or a state in some respect or keeps it unchanged. .does the action "take place" in the initial situation or the result situation? This is the action-logical counterpart of the philosophical problem of the time and the location of an action discussed in section I above.d(u» =/=. sentence (I. the sentences 'Bp'.if they are understood as genuine action sentences. including those of action sentences. but the third one is a merely possible (counterfactual) occasion. V(P.~ According to von Wright.". According to von Wright.

Ch)) coincide with h up to the instant t-1. the truth of D aP at (h.D) DaP . 1992. in his pioneering work on the modal logic of imperatives and action statements Brian Chellas (1969) defines the truth-conditions of action sentences for history-time pairs.h')..Ch) h. the t-alternatives to a history h must (according to (DR.----- (AD. the "instigative altemativeness relation" between possible worlds. ACITONS. Thus they can be . that is. He represents actions by means of an "instigative operator" Da which may be read "a sees to it that .t Fp for every history h' such that Rlh. sequences of world-states. but can diverge from h after that point.t F DaP if and only if h'._w. if possible worlds are understood as histories. thus CheUas's semantics validates the action-logical counterpart of the principle T of modal logic. we can say that an agent performs a certain action in a possible world. For example. OMISSIONS AND NORMS IV 87 One potential source of confusion here is the possibility of understanding the expression 'possible world' in two different ways: it can mean either temporal world-states or world-histories. In von Wright's approach. the t-alternatives of a given history h must have the same past as h. If we let h(t) be the state of world (history) h at t. Chellas calls the relation R. According to Chellas.Ch) Rlh. If we assume that time is discrete and points of time are indexed by integers. Intuitively speaking..ON STATES. R. a 'possible world' is understood in the former way: it means a possible state of the world at a given moment. is assumed to be a reflexive relation. If events are regarded as changes (or world state transformations) and an action is regarded as the bringing about of a change. as paths in a tree whose nodes are world-states. but may differ from hat t (Che1las 1969. and presents the following truth-definition for D-sentences: _n . a world-state.p. Thus he regards the concept of seeing to it that p as logically analogous to the necessity operator of modal logic: the truth of DaP at a given time t in a history h amounts to the truth of the proposition p at t in all "instigative t-alternatives" of h. 82." . On the other hand. (1.t) means that what the agent does in h constrains the course of history in such a way that p is bound to be the case at time t. Chellas's analysis of action sentences does not contain a counterfactual condition.490-491). Chellas's condition can be expressed by (DR.h') only if h(/') = h'(t') for every I' < t. we obviously cannot assume that action propositions are interpreted as sets of possible worlds: actions do not take_p~"Yitp_j!l__l!Qssible..Qrlds.

CheUas's theory makes a distinction between the initial state and the end-state (or result-state) of an action. w. If h. 373) has raised such questions in his comments on Chellas's analysis: Does the agent 'do' anything at t-l [the immediate predecessor of the time of evaluation] to define 'a certain tone . etc.t) represents the state of a world-history at t. what has been denoted by u.does action consist in choosing or somehow committing oneself to a cone"? Otherwise. (Cf.t F DaP. given the assumption of the discreteness of time. Krister Segerberg (l992a. Thus it is misleading to read Chellas's Dsentences as action sentences of the form 'a sees to it that p'. Segerberg 1992a. In Chellas's semantics. we may regard h(t) as the end-state and h(t-l) as the initial state a's seeing to it thatp. the case only if such and such is the case. or what? Since the cone of alternative histories (the t-alternatives to a given history h) represents a choice which is open to the agent in the situation h(t-l). statements to the effect that a certain result p is due to an agent a. Consequently Che11as's theory is subject to the same interpretational questions as von Wright's theory. 372. Thus we have to distinguish here between (present tense) action sentences and statements about agency. it is correct to say that one can be held responsible for it being the case that p only if it is the case that p: statements about (causal) responsibility can be evaluated only at the end-states of actions.. A person is an agent of a certain result only if he has done something which has caused the result.. and action sentences are evaluated at the end-state.) In von Wright's terminology. It would be more plausible to say that the action takes place in the interval [t-l. we obviously cannot say that the agent's action 'takes place' in the later situation h(t). the action-logical variant of the principle T of modal logic (1969. and that the pair (h(t-l). But von Wright's rejection is the case that .e. at t. D). the D-formulas should instead be regarded as statements about agency. we may say that. 88 RISTO HILPINEN pictured by a cone of histories which has its apex at h(t-l).h(t» exemplifies the action which ties the agent to the result.. This point is illustrated by Chellas's justification of the piiriCipie 'D~p. where does action come from"? And when does it take place .p' (1. at the interval [t-l.tl.7 On the other hand. that is. 373) has observed that CheUas's action semantics provides no picture of action itself and suggested that this failure may be related to the Validity of the T-principle mentioned above. It is obviously false or misleading to say that one can see to it that p only if it p: as von Wright has pointed out. 66): One can see to it that such-and-such is.. Krister Segerberg (1992a. a person can bring it about thatp only if it is not the case thatp. at time t.t]. in our earlier discussion of von Wright's theory.at t-l. a pair (h. i. that is. or be responsible for such-and-such's being. and bringing it about tbatp may be a case of seeing to it that p.

27 -28):9 When we say that an individual event happens on a certain occasion. we may regard this happening of the event as constituted by two successive occasions for the obtaining of a certain state of affairs. but von Wright's choice of the initial states as the circumstances of evaluation of action sentences excludes such an interpretation. This suggests that von Wright's <occasions' include not only world-states. For example.f If CheUas's theory is understood in this way. we may regard this occasion for the doing of the act as constinited by the two successive occasions for the corresponding individual event. on the contrary. Similarly. Von Wright seems to be partly aware of this difficulty. and that action sentences should be evaluated at such ordered pairs. his formulation of the counterfactuaI aspect of action seems unsatisfactory.l1 One of the :first philosophers who adopted this way of analyzing action statements seems to have been l. any theory of action in which action sentences are evaluated at moments or instants of time is likely to have some counter-intuitive consequences. when we say that an individual act is done on a certain occasion.. VI The analysis of actions as world state transformations makes it possible to distinguish between several interesting modes of action and agency.12 This approach is a formal counterpart of the philosophical view that the (temporal or spatial) location of an action includes the location where the action "takes effect" and the result which identifies the action is actualized.D).ennart Aqvist. He interpreted generic actions as 2-place relations between possible situations. the lack of a counterfactual condition seems to be a weakness. he notes (1963. According to von Wright. but also ordered pairs of world-states. ACfIONS. 1992a. above). the agent's passivity at any given world-state u would lead to a single world- . but such a condition can of course be added to his analysis (Hilpinen 1997b). as we have seen. w} (1974. that is. but von Wright's formulation of this view has certain shortcomings (in addition to the interpretational problems discussed. 77). v lfactions are temporally indeterminate (to some degree) or extend over a stretch of time. Czelak:owski 1997). OMISSIONS AND NORMS 89 of the T-principle does not make his theory superior to Chellas's theory in this respect.ON STATES. action sentences and unmodalized sentences cannot be combined by means of truth-functional connectives to form 'mixed' complex sentences such as (1. Chellas's theory can be given a reasonable (re)interpretation as a theory of agency statements.I? According to this view. Aqvist's view was developed further in the dynamic logics of action (Segerberg 1980. as sets of pairs {u. for example.

90 RISTO HILPINEN state (counter-state) e(u): the functions d and e have as their values worlds. Thus the agent's passivity at u should be represented by means of a function which has as its value the set of those world-states which could result from the agent's passivity. depending on how u might change without the agent's interference.u) for some situation u.. We can enrich von Wright's analysis by assuming that the agent can perform in a given situation various actions A. it would have been the case that p. The concept of action is used in this analysis as a primitive notion. Such a representation agrees with the analysis of counterfactuals based.A. above. it would have been the case that nor·p. thus von Wright's theory is in effect based On a logic of conditionals in which the principle of conditional excluded middle is valid.u).u) only if u<w. It would be more natural and realistic to assume that the agent can change u in different ways by undertaking different actions or by performing some action in different ways. in other words. wEj{A. This principle is not generally plausible. it can no longer be regarded as an analysis of the concept of action. If wEfi. and let z(u) = fi. we . not sets of worlds. According to von Wright's analysis. z(u) is the set of world-states that might result from u by the agent's passivity.P and is not that in the present case: it is more plausible to assume that the agent's passivity at u might lead to various alternative world-states. VII If von Wright's analysis is enriched in the way suggested. for any u. concepts of agent causation. This means that the counterfactuals used in von Wright's analysis satisfy the following principle: (CEM) Either: if the agent bad been passive. but simply as an analysis of the phrases 'a brings it about that P'.• each of which is represented by means of a function f which assigns to each situation u the set of world-states to which the action might lead the agent from u. 'a prevents it from being the case that P'.. on set selection functions rather than world selection functions (Lewis 1973. v and w in W. then u<v or v<u or u=v. Another weakness in von Wright's analysis is that the representation of the agent's activity by means of the d-functionseems too restricted and uninformative. 79). (CEM) is an instance of a principle called the principle of Conditional Excluded Middle (Lewis 1973. the initial state u offers the agent two action possibilities with respect to any state of affairs p: to be active or to be passive. Let Z represent passivity (the <zero action'). One of the action possibilities open to the agent is passivity. or: if the agent had been passive. 57-58).Z.14 It is assumed that the world-states uE W form a treelike partial ordering < of temporal precedence.Al' . if u < w and v-c w. and related.B. moreover.

(i. expressed by 'JYp': 16 (u.93) and Belnap (1991. A familiar and often used notion of agent causation is expressed by the sine qua non analysis. Thus we obtain the following version of the sine qua non analysis of the concept of bringing it about that p: say that the pair (u. that the agent arrives from u to w by means of A without doing anything else.b) the positive condition of agent causation and (ii) (the counterfactual condition) the negative condition. we may call the conjunction of conditions (i. (ii) is the counterfactual condition: it says that if the agent had been passive. Oii_ ~:::ag~t'-s_ac091j:"FoTIowmg Aqvist . w) is a (minimal) performance of A. (i. This analysis corresponds or=-~"~~~_-of_a __~ -of~~ to the simple sine qua non analysis of agent causation and Mullock (1989. and (ii) z(u) SO. I -. 37. according to which an agent causes a certain state of affairs (or event) only if that state had not actualized (or the event had not occurred) if the agent had not acted they way he did.p F I.792). that is. (Hart and Honore 1959. (ii) z(u) ~ I-'p I . OMISSIONS AND NORMS 91 viated below '(u. p would have remained false. and (iii) u =p. AcTIONS.ON STATES. w) FA'. If condition (ADI) (iii) is simply deleted.b) W F P. Most analyses of agent causation follow this pattern and consist of a positive condition and a negative (counterfactual) condition. 104-108. The semantical apparatus outlined above enables us to define several concepts of 'bringing it about that' and other concepts of agent causation. this will be abbre- (ABl) (u. F F F Condition (i. in other words: a state of affairs P is due to (caused by) an agent only if there is some action A (performed by the agent) such that it would not have been the case that p if the agent had not performed A (if the agent had omitted A).u).w) DIp if and only if there is an action A such that (La) w E flA.a) says that the pair (u.w) exemplifies A or is an instance of A.b) w p.a) and (i.j5 According to von Wright's analysis.u). we obtain analysis of the concept of producing or preserving (the state of affairs that) p. The corresponding analysis of sustaining action is obtained from (AB1) by replacing condition (iii) by the condition (iii') u F p. w) Bp if and only if there is an action A such that (La) w E J{A. this counterfactual should be formulated in terms of the agent's passivity.

whereas DO expresses the necessary condition aspect (or sine qua non aspect) of action: 'DOaP' may also be read as 'but for a's action it would not have been the case thatp' (Porn 1974.w). Kanger presents for the Do-sentences and Do-sentences possible worlds analyses analogous to Chellaa's condition (AD. 121): (ADO) and (ADo) U u F DoaP if and only if w F P for every w such that Riu. and (ii) DoaP = P is sufficient for something a does. In the interpretation and application of Kanger's semantics of action sentences. However. can be regarded as a conjunction (Kanger 1972. we can say that the purpose of the S-relation is to enable us to refer to worlds in which the agent does not do any of the things he does in u. according to Kanger.. 109 and where 'DoaP' and 'DoaP' are understood 121): (i) DoaP = P is necessary for something a does.92 RISTO HILPINEN vm (AB1) and (ADI) are not the only possible generalizations of von Wright's theory. side of (i) can also read as "something a does is sufficient for The right-hand p".ib) f{A. We obtain a stronger concept of agent causation by replacing condition (i.11 According to Kanger.u) !. the .Ch) (Kanger 1972. In the light of von Wright's analysis. because it is not clear how we should understand the 'opposite' (or the negation) of an action. 121). an action sentence of the form 'a sees to it that p'. thus Kanger's D6-operator expresses the sufficient condition aspect of action and agency. This characterization is not intuitively clear.b) by the condition (AD2.ib). According to Kanger.w) Sa<u. 109): (CDKa) DoaP e DoaP & DOaP. 95). then P would be false in all such worlds. we may take the S-altematives of u to be worlds in which the agent is passive: if in u the agent sees to it that p. which can be taken to mean that the action A necessitates the result (that) p or is sufficient for p. F DOaP if and only if w F -'p for every w such that The relation Ra is analogous to Chellas's "instigative" altemativeness relation and Sa is another altemativeness relation between possible worlds. 'Sauw' means that "the opposite of everything a does in u is the case in w" (Kanger 1972. as follows (Kanger 1972. Stig Kanger has defined the concept of seeing to it that p by means of a condition analogous to (AD2. Ip I. abbreviated 'DoaP'.

abbreviated here 'Coap': (ACe) u 1= Coap if and only if W F -'P for some W such that Sa(u.&------~-. In the state transition semantics outlined above. this analysis assumes the form: . 7).-. It means that it is not unavoidable for a that p. etc. Perloff 1991). according to which the agent might have avoided the result p by making a different choice at the initial state of the action (Belnap 1991. should be regarded as world-histories and not as temporary world-states. Belnap and Perloff 1990. tIl~ bringing It about ~_~~ms ~~XP~~~~_!!~s~~ cond. (i. the negative condition should be formulated as a might-statement or a might-conditional. . not as a wouldconditional. even though philosophers have sometimes treated them as alternative readings of the same agency operator.P As von Wright bas pointed out. In their recent analyses of the concept of action and agency Nuel Belnap and his associates have followed Porn's approach and defined the concept of seeing to it that P by means of conditions analogous to (AD6) and (ACe). Lennart Aqvist (1974.-~---~---~-. Condition (ii) is again based on the assumption that the counterfactual condition refers to possible circumstances in which the agent is passive. 1992. The concept of seeing to._!'.DO a ""P' .P:~~ c:_C?~(~CPL~f __ . OMISSIONS AND NORMS 93 possible worlds u. and (ii) z(u) n 1 -'p 1 is nonempty. ACTIONS. viz.j~J. This condition can be read: but for a's action it might not have been the case that P (Porn 1974. 86) has defended a similar weak form of the counterfactual condition . According to Porn and Aqvist. 96. 'p is due to (the agency of) a' means.fQg_4i_t.b) f(A. asnffi_1. a situation offers an opportunity to bring it about that p only if P is not true in that situation (on the ~~_~asion of~tion).19 The locution '~ agf::~tsees to it that p' sugs_~s_~!g~y _~.~_ causation: 1tieie is a clear intuitive difference in meaning between the two eXluisslons. w) 1= IYp if and only if there is an action A such that (i. 1977) has argued that we should accept instead of Kanger's DO-condition only a weaker requirement.i:!i_gn a~~"<?!_!!&~I). by means of a strong positive condition (a 'necessitating' condition) and a weak negative condition. rougbly speaking.a) W E f(A. Horty and Belnap 1995. '-. 1977.!~that is not subject to this restriction: it covers ~~--..productive and su~ both ----_ -~.u).ON STATES.. W.w). (AD3) (u.u) S lpj.---~~. Ingmar POrn (1974.18 IX According to the definitions proposed above. that is. that a has done something (performed an action) which has caused P or has been sufficient for p.iQ!J.

whereas bringing it about that p involves no such guarantees.. that is. we have to assume that the action descriptions A." (Segerberg 1985. These definitions are based on the following ontology of action: an agent has in a given situation at his disposal a number of action routines A . there are no routines for bringing it about that there are no wars in the world. in definitions (ABI) and (ADl)-(AD3) identify actiontypes which are reliably performable by the agent or which the agent can choose to perform in a given choice situation u. The expression <see~to it thatp' usually characterizes deliberate. we can say that he <sees to it' that a certain result will be actualized: in such cases his actions necessitate the result. To deliberate is to search for a routine. that is. including unexpected and improbable results.A I. Following Krister Segerberg. for any state p.ihave-suCb -a ·~nliOtatiori"3iid can applied to the unintended as well as intended results of one's actions. be - assun£ x If an action is regarded as an interference with a natural course of events (a course of events unaffected by agents).• .22 and that all instances of agent causation can therefore be regarded as instances of necessitating agency.or by any agent. turning the ignition key of a car is a routine whereby one can (normally) bring about the state that the engine of the car is running. an action description A such that the performance of an action of kind A is sufficient for p. Another difference between the two concepts is that seeing to it that p normally entails ensuring that the result p will obtain.?' It might be suggested that it is always possible to construct. B. 'to cause (it to be the case that) p' or 'to bring it about that p'. by means of which be can change the situation or keep it unchanged. intentional action. To be able to distinguish different forms of agent causation from each other. For example.94 RISTO HILPINEN (preserving) agency. When a person does something deliberately and with care. the actions <to bring it about that p' or <to see to it that p' are impossible actions. . but 'bringtng it abOO:ithat p. Actions are regarded here as analogous to tools which the agent . for example. not performable by a given agent .D. I shall call such action types routines: "To be able to do something is to have a routine available.B.-does· ru. and (ABl) and (ADl)-(AD3) seem reasonable analyses of the concept of agent causation. 188J3 The action routines available to an agent can be regarded as the tools which the ag~t-~_ise J9r-c~gl. then the concept of passivity or 'zero action' is meaningful. ] iliiif the routines can be characterized independently ofthi-objectiveS or results which an agent tries to reach by means of his actions. be passive. or he can refrain from applying any of these routines.gjlje __ ot:l~Lor-~lij---it ~ unchanged. For example. even though we might think that the United Nations ought to bring about such a state of affairs. For many states p. etc.

and . ACTIONS.~~c alternative called 'passivitY':2'l If we assume that the agenCcan perform simultaneously several actions and the choices available to the agent are analyzed as action complexes. for example: _" I r' AB = A and B together (simultaneously). if we letflA. It is clear that the performance of an action A together with another action B can change the world quite differently from the performance of A or B alone. The plausibility of treating the omission of an action as passivity depends also on the assumption.u) be the set of situations into which the performance of A by itself . For example.of ~y_~. but to situations in which he brakes and stops in time.?ch~ic:~_ of a~. All omissions of A share the feature that they do not exemplify A. flAB. A+B = A or B. Here I shall consider only action complexes of type AB and A +B. and it is 00 longer plausible to formulate the sine qua non condition in terms of the passivity of the agent. it is natural to assume that the sine qua non condition of causal dependence refers to hypothetical circumstances in which the agent is passive.1l4.!9_ ~ and JJl!. Thus. as in (ADl)-(AD3). abbreviated 'omA' (where A is an arbitrary action description). If action siruations are understood in this way. no accident would have occurred'. that an agent can perform in a given situation u only one of a number of alternative actions open to him.B = A followed by B. we may say that the driver of the car caused the accident by his failure to brake and stop the car before the crossing. In such cases the counterfactual aspect of action (or an omission) cannot be analyzed by referring to situations in which the agent is passive. he satisfies the sine qua non condition 'if x had not driven into the crossing. I shall also consider the omissions of acts. In this case the omission of an action A means either the ~oi~ .let 'AB' represent the joint performances of A and B. In this case the antecedent of the sine qua non schema does not refer to situations in which the driver "is passive". made tacitly above. on the assumption that the situation u is represented as set of mutually exclusive action alternatives.. There are situations in which there is no clear alternative of passivity and no alternative can be regarded as the 'natural course of events'. otherwise the character of omA depends on the complex action from which A has been omitted. and A. in other words.~y~ _~ "4__Qr_~-. neither interpretation of the concept of omission seems correct. this interpretation is not applicable to all action situations. if someone drives his car into a railway crossing and is knocked over by a passing train..ON STATES. However.u) _i_~ nQ!__ _I?Y.u) be the set of worlds to which the performance of A alone might lead from u. Let g(A.!_ul I shall assume here that the actioii-'desCriptions A and B can be joined together to form complex action descriptions. OMISSIONS AND NORMS 95 - can use to work on a siruation.

) If he does only A.. (fl:A.u) S g(A. by means of an action AB. that is. thus we may call AB the 'conjunction' satisfy the following conditions: (4) itA. q = George is alive and well.96 RISTO HILPINEN or together with other actions may transform formance of A alone. f and g should XI The application of this semantica1 apparatus to the analysis of action situations can be illustrated by the following simple example. zeu) ~ I""p & q I. We may assume that he can do this in some routine manner. (It is of course essential here that he does these two things simultaneously. He wants to open the safe without hurting himself.u) ~ I""p & q I.u) ~ Ip & ""ql.u) and g(AB. Ip & ql· Suppose that George hides behind a protective shield while detonating the u E I""p&ql.u). The states of affairs in which George is interested and the relevant acts available to him are described by the following (atomic) sentences and action descriptions: (5) p = The safe is open.) Thus the situation u and the results of the action possibilities open to George in a can be described as follows: (6) ftA. A = (the action of) detonating the explosives. George can do this by detonating the explosives while protecting himself.u) rt g(B. of A and B. for example. (In this example a clear alternative of passivity is available.25 Moreover. Let us assume that an oldfashioned safe-cracker called George wants to open a safe by means of suitably placed explosives.) It is clear that (3) g(AB. .u) s. he might be hurt by the explosion. and B alone as well as the zero action would preserve the initial state of the world. B = (the action of) protecting oneself. fiB. because p & q is false in the initial situation. thus he has to protect himself while detonating the explosives.u) = u. by using a protective shield or by hiding behind a desk. This would be a case of bringing it about that p & q. !CAB. ""P & q is true and George wants to transform this into a situation in whichp & q holds.a) ~ g(A.u)._) represents the per- g(A. In the initial situation u.

OMISSIONS AND NORMS 97 explosives (in other words.ON STATES. that is. Lewis 1973. The similarity between individual courses of action can be represented in the same way as the similarity between possible worlds. and (u. p would not be the case. If we assume that the similarity between complex actions depends only on the number of simple (atomic) actions shared by them. (Cf. the sfunction should in this example have the following properties: . ACTIONS. courses of events in which George does not protect himself. 1. but which are otherwise maximally similar to George's actual behavior at {u. but by considering omissions of the act B. The counterfactual (10) is not evaluated by considering courses of events in which George does nothing. {u. that is. '0' may represent either a positive action or an omission. But if George had been passive.w). Above. The following counterfactuals are true at w: (7) (8) If George had not detonated the explosives. those world-states which could have resulted from the initial state by a course of action which satisfies the action description A and is otherwise maximally similar to the actual course of action (u. w})!=: F q.w}. In other words: (9) (10) If George had omitted A.) The counterfactuals needed here for the analysis of the concept agency are defined by the following truth-condition: (CCA) w If G (had been chosen by the agent).w) exemplifies AB. If George had not protected himself. he would not be (or might not be) alive and well (without injury). if and only if Iq I . represented by a pair of world-states consisting of the initial state u and the result state w. If George had omitted B. This way of evaluating conditionals with action descriptions as antecedents is based on the general principle that conditionals should be evaluated by considering courses of events which satisfy the antecedent but are otherwise minimally different from the actual course of events. q would still be the case: George would be unhurt. q would not be (or might not be) the case. then there is a situation u < w such that s( G. thus he arrives at a situation w in which P & q is true. the safe would not be open. because there would not have been any explosion. by means of a selection function s which selects for each action description A and each individual course of action. The antecedents of (9) and (10) refer to the past of the world-state w. he does AB) and manages to open the safe without getting hurt by the debris.

<yv e may also include in the set of alternative actions other possibilities which are completely irrelevant. w» = .w) DSp if and only if there is an action A such that (i) w E g(A.u) and g(A . and s(om(A+B). w) (i) (ii) F D"P if and only if there is an action w E g(A . Band Z as the alternative actions or choices open to George in the situation u. The corresponding concept of 'strong' p) can be defined as follows: (ADS) or necessitating action (seeing to it that (u. We have to use the g-function and not the J-function in these definitions.w})!:. s(omB. A.u).u) n !p!. 26 XII If George's action situation is analyzed simply as a choice between a number of alternative (mutually exclusive) action possibilities. depending on how the example is understood.w» = j{B. and 'seeing to it that' can be interpreted by (ADS) or (AD6). According to (eCA). we obtain the A such that (u.p' can be understood in two ways: (i) (ii) Some action alternative to A would (or might) have led to 'p.4. and (ii) s(om. and not s(omA. the sine qua non condition 'If omA.w»!:.(u.4.98 (11) RISTO HILPINEN s(om(AB).{u.w» = z(u). According to (11). jp I.{u. then . Ip I· In this example.4.{u. Passivity (in an earlier situation) would (or might) have led to =p.u) and g(A . but the outcome p in which we are interested may depend on only one of these actions. the conditionals (9) and (10) are true at the result state w. w})!:. we have to regard AB.u) !:. I"'pl.u). F If the negative condition is formulated following generalization of (AD3): (AD6) as a might-conditional. and s(om.) Under this analysis.(u.u) U . we can say that George saw to it that he was not hurt by the explosion. s(om. . because the agent can perform several actions in the initial situation./V4. !p!.(u. the concept of (sine qua non) agent causation can be defined as follows: (AD4) (u. the action of praying that the safe would open by itself. I"'pl.{u. w) (i) (ii) w F D4p if and only if there is an action A such that E g(A./(B.u) !:. for example.u). w}) = flA.

_ . we should say. the negative or counterfactual condition of agency can be regarded as a special case of the corresponding requirement for an adequate explanation._a_@~_o!_!_~!!~!_p_qtfly.L ~ xm All the analyses of agent causation considered above entail the following condition: ~y ~_~g_~~~ . .-DiSed. " '. 73)~7 thus the condition of favoring suggests the following connection between the notions of agent causation and explanation: ~-~-~:r~t_ _. the omission of the relevant act driving into the railway crossing without stopping .. For example.~. !: According to the analysis proposed here."t itself does not make George the agent of the safe's being locked if he does nothing else in the situation.1'10 . In the collision example given earlier.. ACTIONS. 19-21). oPen 'by- .ji~ a. concept olprob=- (CA. also gives unsatisfactory results. 21-22).I . Mellor 1995." on the'concept ol'passivity. and the conditions of agent causation discussed above have their counterparts in the theory of explanation. OMISSIONS AND NORMS 99 If George does nothing in the initial situation in which 0p & q is true (or for example prays that the safe would open by itself)._~c1i_vitY:Of_~_' Oidie_~omisSioii-Oi:A-. according to the first proposal.exp) a is an agent of the state of affairs that P only if P is explained by the actions of The concept of probabilistic favoring is usually regarded (at least) as a prima of agency and agent causation (Aqvist and Mullock 1989... - ON srxras. In the present analysis.. this account is inapplicable to the situations which do not include a clear alternative of passivity. _..1. ~{-. and can be regarded as a generalization of von Wright's concept of passivity. In the deductive model of explanation this condition is the requirement that the general laws in the explanans should be essential (necessary) for the derivation of the explanandum (Salmon 1989. the concept of omission is indispensable. an agent of the collision.. that the fact that the safe ~Jocked is due to (or caused : by) George.. Moreover._~-:-~-~~4:-~y~~_~ -!he safe to deto~_the :' expli5SiVeS:-We have seen earlier that the second ·account._[_~OTed by ~ some action A peUormed I?Y a in the sense that p is more likely as a result of A abilistic--favoring is familiar from the theories of causation and explanation (Eells 1991. because in such circumstances the safe would have remained locked even if George had omitted to pray.._!f_p_ ii.. praying that the safe would open by . Most authors who have developed logics of action facie criterion . according to (AD4). chapters 2-3. . 1-5.consists in not driving into the crossing while deviating from the actual course of events in other respects as little as possible. The variety of the concepts of agent causation resembles the variety of the concepts of explanation. The driver of the car satisfies the counterfactual condition and is therefore.~: ..

121.. .exp).. In cases of negligence the agent usually satisfies the sine qua non condition that p would not be the case if the agent had not omitted A (for some A).P? Thus we should say that an agent may be regarded as responsible for some result p if he has caused p or if he has failed (omitted) to cause (or see to it that) 'p. (Cf.have been mainly interested in actions which necessitate their results. that is. because then we should regard a person as an agent of any state of affairs which he could have prevented by choosing a different course of action in some earlier situation. but such judgments of agency or agent causation depend on judgments of responsibility and not vice versa. In our ordinary moral thinking we make a distinction between activity and passivity or between action and omission. it is clear that these analyses apply only to certain interesting special forms of agent causation. Aqvist and Belnap . mentioned in the definitions (ADI)-(AD6) should be positive actions. For example. . cases of negligence in which a person is regarded as an agent of a harmful result because of his omission of some required action are examples of the latter kind.28 the fact that there are borderline cases in which it is difficult to say whether a person does something or is passive (with respect to a certain result p) does not obliterate the distinction. Chellas. XIV It should be observed that agency or agent causation in the senses discussed here (as defined by (AD1)-(AD6») is not a necessary condition of the moral or legal responsibility of a person for some state of affairs. Kanger. Hart and Honore 1959. In the light of condition (CA. .B. The action routines A.e? However.for example. according to which the explanantlum of a good explanation should be a necessary consequence of the explanans.100 RISTO HILPINEN . but this is used as a ground for judgments of agency only in situations in which the agent was required to see to it that =p. etc. in this respect these theories of action resemble the deductive model of explanation. In these cases we should not say that the agent causes p (sees to it that P or brings it about that p). not omissions. Sometimes an agent is regarded as responsible for a state of affairs p because of his omission of an act A which would have prevented the result p.) It is natural to say in such cases that the result is due to the agent. it is clear that letting something happen (being passive) can in some circumstances be morally or legally as objectionable as causing something by means of one's positive actions: letting an animal die may be as objectionable as killing an animal. because he has let it be the case p (has let p become or remain true). Porn.

OMISSIONS AND NORMS 101 xv In this paper.g.iQ~b_e-:iweii©ifi· . the latter include agency statements.~M~. ~_~~~~. permitted or forbidden in a . I '. and A is prohibited if and only if all transitions which exemplify A are illegal. ~e divide the movements from one world-state to another into legal (o!:__~~!.!!_ re~_(Or is obligatory). a state of affairs) is due to the actions of a certain agent. To represent this : rustiiictioii"m-deontlcToglc.ON srATES. in the subsequent work on deontic logic the deontic operators were usually construed as propositional modalities rathertban action modalities. 33 However. j : .31 The distinction between action descriptions and propositions underlies the traditional distinction between two kinds of ought (or two kinds of oughtstatements).i •.ol1gti_t. that is.e_ffect .y~~~equo !be. ought-to-be and the ought-to-do (between Seinsollen and .. The truthcolidlfioIlsofdoonrlc'-sem:encesc3n tlieii--6e''deftned in the same way as in the standard system of deoutic logic: an action A is obligatory (or required) in a given situation if and only if all legal transitions from that situation exemplify A..-~n:. e. Wh.t~~ of affairs ought or ought not to obtain in a given situation.'I. Instead of dividing world-states into deontica1ly perfect (ideal) worlds and deontica1ly imperfect (unacceptable) worlds •. Tunsollen)_31 An ougbt-to-do statement_~_!L[lO~g. we have to distinguish action descriptions from 'ordinary' propositions which are interpreted as sets of situations. s~yj~~_ a.-we" neeti"iWo-kmdiof'dooritic operators: the ought! to-do operators apply to action descriptions and turn them into deontic statements. sentences which state that a certain result (a fact. AcnONS. According to this conception of actions.3s The state transition semantics can be enriched and embellished in the same way as the semantics of the standard system. not to propositional expressions (statements). a simple semantics of directives (ought-to-do sentences) can be obtained by applying the basic ideas of the standard semantics of deoutic logic to such transitious or ordered pairs of possible worlds (world-states). by defining in addition to the ideal transitions various "sub-ideal transitious" or by distinguishing ithe l' . Georg Henrik von Wright's first (1951) system of deontic logic which started the contemporary development of the field was a logic of the ought-to-do: in this system the deoutic operators were prefixed to expressions for action-types or (in von Wright's terminology) "act-qualifying properties". whereas the ought-to-be operators turn 'ordinary' propositions into deontic statements. viz. that a certain ac_!o. and it explains the temporal and situational indeterminacy of actions. This model fits actions which can be said to lead to a certain result and characterized by means of their results.~#Qi. 34 If actions are represented as binary relations between world-states or as transitions from one world-state to another. actions have been analyzed as world-state transitions or as relations between wcrld-states.Je) tiimiiuouSaIidillegai (uiiaCCepti6le) transitions (C~~\ij~![l997).

This view reflects a consequentialist conception of normativity. 101-104).. (Cf. Carmo and Jones 1993. Back 1996. can apply its own criminal law to the case . or Z. and B goes to and as a result of the injury dies in state Z. but on the level of ordinary nonnative argumentation the justification can ron in both directions. see (1983.----""--' objeCtivesOfan University of Turku and University of Miami.Sjatus o{"8ffiiifS -may: be considered deontologica1ly on the basis of the actions which led kdi orcould have"led'tO-ii.. or that p ought to obtain because it would have been the result of an action (or actions) which an agent ought to have performed in an earlier situation.) What is the relationship between the two kinds of ought? Von Wright (1996.36 Thus the n~ve status of an action maybe considered from_l!__£Onsequentialist ieWpoiDt ~ the lighi·o(i:1!f?-_-jnterestS and v age~(oia-nofui7atithoJ:i!y. ._aru{_Qte~ti__QiTiiat!ye of 3_ state . we may safely make the following generalization.'··" . -"'-. For example. Where A in state X sets in force a motion which injures B in state Y. - between different degrees of deontic perfection. Y. ' 4Von Wright sometimes uses abbreviated expressions of this kind. see Kit Fme (1982. Jones and Porn 1985. Sometimes we say that an agent ought to perform an action because it is necessary for a result which ought to obtain. that is. they are plausible candidates for qua-objects which are constituted by an object together with some property possessed by it.. 2 Cook (1942. but normatively acceptable (permitted) if it is the result of honest work. 489 -493). either X. who has argued that individual actions can be identified with the agent's "primitive actions". According to the doctrine of qua-objects. or more likely to lead to such a result than the alternative actions. NOTES I In the recent philosophy of action the Wolfean view has been defended' by Donald Davidson (1980). 'Socrates qua philosopher' refers to an object (a qua-Object) distinct from Socrates (Fine 1982. a person's being rich is a normatively unacceptable (prohibited) state if the wealth bas been obtained by theft. S The explicit reference to the agent can be omitted when the actions of a single agent are being considered. 174). 61-74. in other circumstances we may argue that a certain state of affairs P ought not to obtain (is normatively unacceptable) because it resulted from an illegal or immoral action. if it gets its hands on A. For the application of this view to acts. 3 If individual actions are understood in this wily. •. 13-14) notes: On the basis of actual observation of wbat the courts bave done and are doing. 68) has argued that the norms that prescribe what ought to be (or may be) are more basic than the norms concerning actions.102 rusro HILPINEN . bodily movements which are "intentional under some description" . and that the latter depend on and are justified by the former. Perhaps this view is correct on some fundamental level of morality. 100.

see Hilpinen (1981. 1 We can say. see Hilpinen (1997b). According to Ota Weinberger (1985.309).no one is satisfied with it but it is hard to ignore. and Sandu and Tuomela (1996). she regards both as propositional expressions . any binary relation on W defines a function from W into !he power set of W. Here the D-sentences are action sentences proper. 137 -166). (Cf. 311314). but calls both m and. should be regarded as involving two 'phases:'. 6 OMISSIONS AND NORMS 103 Krister Segerbcrg (1992a) interprets von Wright's theory in this way: he regards (2) as a valid principle of von Wright's logic of action (p. in the state transition semantics this assimilation is apt to lead to the confusions discussed above. !6 In my earlier paper (Hilpinen 1997b) I defined the agency operators as operators which turn 'ordinary' propositions into statements of agency evaluated at the result-states of the agent's actions. 307 ." If the distinction between world-states and transitions is taken seriously. 38). 174.38). the sine qua non analysis gives rise to the familiar problems of causal overdetermination. 1m). 8 Most versions of the slate transition semantics of action and agency proposed in the recent years have followed Chellas's example in this respect. here "conditions" (p. Hamblin has analyzed actions or deeds in terms successive world-states (Hamblin 1987. IS For a more detailed discussion of the work of NueJ Belnap and his associates. 67) have observed.. 13 For a discussion of the Principle of Conditional Excluded Middle and counterexamples to it. II However.ON STATES. A state of affilirs p may be caused by an agent's actions without being dependent on Ihem if another agent's actions would have caused p if the first agent had been passive. Von Wright's logic of action can also be revised in this way and regarded (or reinterpreted) as a theory of agency statements. IS The sine qua non analysis is not withont its problems. see Hilpinen (1997b). voa Wright notes: "The fact that an agent performs a certain action is also a state of affairs of a sort. 14 This representation is equivalent to the representa1ion of actions by binary relations on W. of course. For example. agent x sees to it that r (Holmstrllm-Hintikka 1991. but in this paper I shall leave these problems and complications aside. "an action is a transformation of states within the flow of time" involving a subject (an agent).L. 117-118. 19 The formulations of agent causation discussed here resemble Lars Lindahl's (1977) and Gbita Holmstrem-Hinrtkka's (1991) formulas for "instnunental action".) The sine qua noll analysis needs refinements. Belnap and Perloff (1990. that an agent can have sun to it (or can have brought it about) that p only if it is the case thatp. Ch) discussed above. '(an agent] psees to it that F by performing action A' (Lindahl 1977. we can fonn complex sentences from action expressions and propositional expressions by means of suitably defined intensional connectives. For example. a phase when the state of affairs is absent and a phase on which the state is present. actions should not be assimilated to states of affairs. Leo Apostel (1982. Holmstrom-Hintikka does not distinguish action descriptions from propositions. Hart and Honore 1959. 10 In (1996. 69). as Aqvist and MuUock (1989. The cOnditio sine qua non is a perennial problem for all theorists of causation . Czelakowski (1997). evaluated at world-state pairs. 358). 9 See also von Wright (1983. for example. 104) bas observed that "anaction is a transfonnation of namre in order to realize a purpose" . cf. 11 In Ibis respect Kanger's analysis agrees with Cbellas's definition (AD. a state ofaffilirs comes to obtain. . for example. 69) and 'by means of m. 195-196). as will be seen later. Belnap (1991). AGnONS. 12 Many philosophers bave adopted this view of actions. where von Wright DOleS that an occasion on which a change takes place. and in his "action-state semantics" for imperatives C. 37 .

25 For a more detailed discusslon of diis way of analyzing actions. von Wright has adopted the latter approach.H. that directives and norms have non-propositional contents called practitions (Hilpinen 1993b. the interpretation of action descriptions in terms ofthej-function and the g-function makes it possible 10 make normative distinctions which are not possible in die standard system. 22 Krister Segerberg (1989. 1995) distinction between practitiom and propositions is analogous 10 the present distinction between action descriptions and ordinary statements (including statements about agency).'. see Eells (1991. Lindahl calls such an action a "null action" (Lindahl. It is clear that an action of type o5p neceBSitalesp. thus my use of the expression may differ from his. 48). I shall not try to analyze the concepts intennediate between (AD4) and (ADS)-(AD6) here. 127 -168). 70) has observed that the expression 'x sees 10 it diat p' characterizes sometimes merely an intention or preparedness to act in order to sustain the stare of affairs p should die circumstances require it. g8 -89). and an act A is strongly permitted if and only if any transition which consists of A alone is legal. and therefore the expression 'seeing to it that' fits it better than 'bringing it about that' . I). and argued only diat killing someone is not necessarily morally worse than letting someone die. Ingmar POrn (1974) defines the concept of agency by means of a 03-type modality. Lindahl's null action can be regarded as one of !he action alternatives A.104 20 RISTO HILPINEN <-. who was interested in developing deontic logic as a theory of the ought-to-do. 23 It is not entirely clear whether Segerberg wants to understand the concept of routine in the way intended here. 21 Lars Lindahl (1m. . Hohnstrom-Hlnrikka 1991. thoop is necessary for something a does" (1974. . ibid. see Hruscbka (1986). and does not necessarily require any bodily movement. 37 -38). Nuel Belnap (1991) and his associates the former. 1985). 30 The philosophers who have rejected the moral significance of the distinction between killing and letting die have usually assumed that die distinction can be made (in most cases). 28 See Lucas (1993. a weak permission and a strong permission: we can say that an act is weakly permitted in a situation u if and only if it is exemplified by some legal transition from u. It is also possible 10 distinguish between different . and should be distinguished from die concept of zero action or passivity used here. 96). he argued. 29 If actions are regarded as analogous to instruments or tools to be used for resbaping the world. 27 The view that probabilistic favoring is a sign of causation requires various well-known qualifications which need. see Rachels (1975) and Thomson (1985/1986). For example. see Hilpinen (1993b). For a discussion of Castaiieda's theory. but translates his agency sentences by the phrase 'a brings it about that . n This distinction is closely related 10 but not the same as the distinction between impersonal and personal ought-statements (flats and personal directives). 26 It is clear that (ADS) and (AD6) define very strong concepts of necessitating (sufficient) action.. used in (AD 1) (AD3). etc. and Risoo Hilpinen (1993a). 34 One exception to this trend was Hector-Neri Castafieda (1981.. Porn's concept is a concept of necessitating agency. . it is possible to distinguish between two concepts of permission. 1992b) has studied the logic of an 'action-forming' . the distinction between an action and an omission (or passivity) is clear. 11 Hector-Neri Castaneda's (1981. B. ch. because it is defined in terms of the result p. For example. see' Brigitte Buck (1987 . not be discussed here. 3S However. see Hilpinen (1997a).!i-operator which forms from a propositionp an action description op with die reading 'the bringing about of p'. and notes: "It is a plain troth that if a brings it about thatp. Lucas has argued that the failure to do justice to the distinction between action and inaction is one of the weaknesses of consequentialism. 24 G. 33 In this respect von Wright's 19S1 system agreed with the view adopted in the 17th and 18th century literature on nonnative discourse in which deondc concepts were regarded as action modalities.

Aqvist. Universitlit zu Kiel. Wien. "Towards a General Theory of Argumentation" in E. (eds. Jose and Andrew Jones (1993).). Perry Lane Press. Workshop Proceedings: Second International Workshop on Deontic Logic in Computer Science. Dordrecht and Boston. Oslo. . Oxford.). Knowledge 'RepresetlJll1ionand Defeasible Reasoning. OMISSIONS AND NORMS 105 concepts of prohibition. Cook. Cambridge University Press. Kit (1982). Tano A. DordrecbtBoston.S. LePore and B. Czelakowski.). Eells. pp. Honore (1959). "Action and Deontology" in Sten Lindstrom and Eva EjerhOO(OOs.P. (eds.I. Chellas. Oxford. D. Hector-Neri (1981). pp. Actions andEvents: Perspectives on the Philosophy of D. (1987). A Canonical Form for Agentives" in H. Logic. Stanford University). MON '94.. 167-190. 463 -484. Belnap. Oxford. Davidson. REFERENCES Apostel. Mass.L.M. Buck. Belnap. Lennart and Philip Mullock (1989). pp.New YorkKOln. "Seeing to It That. Events and Things" in Werner Leinfellner et al. pp. Fine. Walter de Gruyter. Janusz (1997). Barth and J. Brian (1969). Hilpinen (00. pp." Philosophy fllld Phenomenologicai Research 51.). Clarendon Press. and A. ACTIONS.L. "The Way of the Agent.). Sergot (eds. Essays in Actions and Events. C. Kybwg et al. 294-310. Basil Blackwell. Martens (OOs.l. Basil Blackwell. The Simplest Solution to All of Them in One Fell Swoop" in R.). Dordrecht. Reidel. pp. Jones and M. " Studio Logica 51.ON STATES. E. Leiden .V. "Acts. Amsrerdam. Actions and the Feundations of Ethics. "The Paradoxes of Deontie Logic. Erne deomische Logile auf tier Gnmdlilge dynamischer Aussagenlogik. Belnap. 777 -807 . "Deontic Database Constraints and the Characrerisationof Recovery" in A. Probabili£tic Co. Berlin-New York. Logical Theory and Semantic Analysis. The Logical Form of Imperatives (Dissertation. Donald (1980).) . 73 -91. Nuel and Michael Perloff (1990). "Time and Modality in me Logic of Agency. On RedupUcation. Norms. Hamblin. Hector-Neri (1985). 56 -85. Harvard University Press. Dordrecht. IGeI. cambridge and New York. Clarendon Press. Approaches to Theory Pormaiion. stanford.). Logical Theories of Qualification. Brill. Nuel and Michael Perioff (1992). "Aspectnal Actions and Davidson's Theory of Events" in E. Caslai'ieda. "Backwards and Forwards in the Modal Logic of Agency. McLaughlin (eds. D. Allan (1996). Causation in the Law. Leo (1982). Hart. New Stwiies in Deontic Logic. "A New Approach to the Logical Theory of Actions and Causality" in S. Back. Dissertation. Ellery (1991). in the American judicialsystem a piece of evidence may not be admissible (pennitted) in court if it has been obtained by illegal actions. The Logical and Legal Bases o/the Conflict of Laws. 37 -85. 310-311). 36 In the same way. Herben L. Stenlund (ed. Imperatives.I. Aqvist. K1uwer Academic Publishers. Brian (1992). For the weak and strong concepts of permission and prohibition.A. Lennart (1974). regardless of the probative value of the evidence. K1uwer Academic Publishers. Cambridge. pp. Chellas. 97-105. Ctmsing Harm. "Agency" in D. 43-61. Nuel (1991). ~ StudiaLogicaSl.usality. see Risto HiIpinen (1993a. Language and Onwwgy: Proceedings of the Sixth Inremati01UJlWittgenstein Symposium.. Davidson. pp. HlilderPiehler+Tempsky. Oxford. Action and Cognition. Brigitte (1987). Walter Wheeler (1942).Argumentation. Reidel. John Benjamins B.M. Canno. 93-122. Davidson. 485-517.. Castaneda.

. Philosophy: A New Deontic Logic in Computer Science: Normative System Specification. 275 -295. Basil Books. D.R. A Formal1heory. Segerberg. Dordrecht and Boston. Gabriel and Raimo TuomeJa (1996).). Berlin.. pp. Porn. Perloff.). Logical1heory and SemonticAnatysis. "Four Decades of Sclentlfic Explanation" ---pany. Salmon (eds." Synthae 105. (1989). 3-219. Sub-Ideality and Deontic Logic. Hruschka.). Risto (1993b). Salmon.----. "On Impersonal Ought-Statements and Personal Directives" in E.). Minneapolis. lngmar (1977). "Action Incompleteness.Science. (Minnesota Studies in the Philosophy of Science. Normative Systems in Legal and Moral Theory.(ed.H.lhoff. K1uwer Aeademic Publishers. Segerberg." Stildia Logica Studia Logica 51. Position and Change. bei Gottfried Achenwall im Jahre 1767. "Some Basic Concepts of Action" in SOren Stenlund.) Thalberg. Oxford. undfonru. Joachim (1986). Perception. Rachels. and Obligation. ." The Journal of Philosophical Logic 24. Martinus Ni. Stig (1972).- . Hamburg. Rex (1935/1980). Jones.Com- England Journal of Medicine in Philip Kitcher and Wesley C. Bantam . 275 -290. Blackwell.tsGesellschoft tht: WlSsen. The League of Frightened published in 1935. 185-210.). Sprachpragm4tik. Fatschriftfor Carlos E. 299-335. Clarendon Press. Meyer and R. Oxford. 319~345.. "Applying Krister (1985) Krister "Routines. Dordrecht and Boston. ~ Studia Synthese 65. "Bringing Keister (l992a). (1989). The Hague-Boston-ulDdon. Wesley C. and Ingrnar POrn (1985). -J. and Modality" in Herbert Stachowiak (ed.379-408.-~". Krister (1mb). 4 (2). Ghita (1991). Mellor. University of Minnesota Press. pp.schoften e. Porn. Responsibility.). 85-100. Segerberg. "On Action and Agency" in Sten Lindstr6m ~ Eva Ejerhed (eds.78-80.). Lewis. David (1973). Wieringa (eds. London and New York.. 583 -644. It About. Action. Ability. a Active and Passive Euthanasia. Pragnwtik: Handbuch Pragmatischen Dimkens. "The Deliberative Stit: A Study of Action. Action Theory and Social. SO. Chichester-New York. "Stit and the Language of Agency. Ingmar (1974). D. 93-101. Hamburg. "Actions in Deontic Logic" in J. Sckntific Bxplanation. Irving (1977).. Dordrecht." Stout. Kanger. "Das deontologische Sechseck. "Joint Action and Group Action Made Precise. Omission.327-347. Reidel. Hilpinen. Vol. Reidel Publishing Lucas. pp. D. "On Deontic Logic. Societas Philosophica Fennica. D.G. V. Emotwn and Action: A Component Approach. Risto (l997a). Purpose and Will." Started: Beginnings Logica 39.llive Pragmolik. AIchourr6n and Eugenio Bulygin. Pragmatics. (eds. Hilpinen. 347-378. The Facts of Causation. sophica Fennica Horty. "Ideality. Risto (1993a). Hilpinen.. CounUrfactuals. RISTO HILPINEN Risto (1981). 105-132. Hiipinen. 115-132. Segerberg. Thomson. Acta PhiloRisto (l997b). Segerberg. _ . . Michael (1991). Duncker & Humblot. Oxford. in the Logic of Action. pp. Action and Cognition. "Law and Logic. "Getting Journal of PhiiosophicalLogic 18. John and Nuel Belnap (1995). Krister (1980). Sandu." Theoria 38. Ch." Modal Logic. Band IV: Sprachphilosophie. Contemporary Survey. James (1975)." The New 292. (1993). (1995). Valdes et ai. New York. Routledge. (First 51. "The Time of a Killing. 533'-550. " The Journal 68. XIII). 295-319.J.106 Hilpinen. . " Berichte aus der Sitzungen der Joachim Jungir. Dordrecht." Sy1l1hese 65. "Conditionals and Possible Worlds" in G.-.J. Basil Blackwell. Felix Meiner Verlag. Flaistad (ed. Holmstrfun-Hintikka. Men. of Philosophy Judith Jarvis (1971). N Synthese 86. pp. Andrew U. Helsinki. Jobn Wiley & Sons. LindMl. Reidel. Lars (1977). Logic.

ON STATES. pp. Routledge & Kegan Paul. Mass~. pp. Cornell University Press. and Logic" in G.OUght-to-do.63-70. Co.J. Norm and Action. Thomson. "Deontic Logic. " in G . "Freedom. Yon Wright. Acta PhlIowphica. 1. Six Essays in PhiIosophicalLogic.H. An Essay in Deomic Logic and the General Theory of Action." Synwse 65. Acts and Other Events. Judith Jarvis (1985/1986). "Norms. ACTIONS. 1986. Judith Jarvis (1977). 1-15. Acta Philowphica Fennica 60. von Wright. Vol. Ithaca: Cornell University Press. Rights. London. Helsinki.H. pp. Logical studies. OMISSIONS AND NORMS 107 Thomson. 58-74. Thomson. 130-209. Reprinted from The Yale Law JoumalM. yon Wrigbt. von Wright. von Wright. yon Wright. Amsterdam. Practical Reason: Philosophical Papers. Restitution. "The Trolley Problem" in J . Cambridge. von Wright. London. 1985. Ithaca.H. Georg Henrik (1996). von Wright. Georg Henrik (1983). Weinberger. Georg Henrik (1951). pp. Range for Action and the Ontology of Norms. Georg Henrik (1968). . North-Holland Publ. Societas PhilosophicaPennica. Reprinted in G. Georg Henrik (1963). eta (1985). Fennica 21. Routledge and Kegal Paul. 94-116. and Risk: Essays in Moral Theory. 307-324. Truth. "OUght-to-be . Harvard University Press." Muul 60.

.../ ..........I / --/ / .-.. .