Você está na página 1de 18

The Logical Assessment of Practical

Reasoning
Georg Spielthenner
University of Zambia

Abstract
There is no doubt that in addition to theoretical reasoning there is practical
reasoning. We do not only want to work out what is the case but also what we
ought to do. In this paper I shall discuss when practical reasoning is logically
valid. This issue has not received much careful attention even though it is of
vital importance to the theory of practical reasoning. After some preliminaries
I define when practical reasoning is valid. Then I discuss the main accounts of
the logical structure of valid practical arguments and show that they are defec-
tive. After explaining my own account of valid practical arguments, I outline a
logical system for evaluating practical reasoning and apply it in the last sec-
tion to an example to demonstrate its use in practice.

Keywords: practical reasoning, instrumental reasoning, practical arguments,


practical inference, validity, practical logic

Since Aristotle it has been common among philosophers to distinguish between


two fundamental types of reasoning, theoretical and practical. We do not only
want to answer such theoretical questions as ‘What is the case?’ but also such
practical questions as ‘What is to be done?’ Consider the following simple
piece of practical reasoning:
We must oppose the idea of converting food into fuel. The use of maize to make
ethanol biofuel will drive up the price of maize. As more land is used to grow
maize rather than other food crops, such as soy, their price also rise. And since
maize is used as animal feed, the price of meat goes up, too.

Reasoning of this kind is ubiquitous and in daily life all of us encounter it


frequently. It will be my object in the present study to investigate how we can

  The example has been extracted from The Economist, April 7th, 2007, pp. 11-12.

Sats - Nordic Journal of Philosophy, vol. 9, no. 1, pp. 91-108. © Philosophia Press 2008
92 The Logical Assessment of Practical Reasoning

determine whether it is logically valid if there is any valid practical reasoning at


all. The problem of how to assess practical reasoning from the logical viewpoint
has received scant attention in the literature, despite the fact that this assessment
is not only an integral part of the general theory of practical reasoning but is
immediately relevant to ethics also. For instance, we cannot intelligibly discuss
ethical reasons for acting unless we know the logical status of such reasons.
The theory of practical reasoning is a live topic and there can be little doubt
that it is important. However, since practical reasoning is a many-faceted activity
that involves a variety of distinct types of inferences and can be assessed in
different ways, some preliminary clarifications will be helpful.
1  We need to distinguish between different kinds of practical reasoning.
For instance, there is normative reasoning, which sets out from norms and is
mainly studied by deontic logic; there is decision theoretical reasoning; there is
reasoning that includes other agents and their decisions, studied by the theory
of games; and there is imperative reasoning, which contains imperatives among
its premises. In this paper, I have nothing to say about these forms of practical
reasoning. I shall confine my attention to only one type of practical reasoning,
namely so-called instrumental reasoning, which is undoubtedly a very basic
and important form of practical reasoning. (Hereafter, ‘practical reasoning’
means ‘instrumental reasoning.’)
2  Many authors have too narrow a conception of instrumental reasoning as
they tend to identify instrumental reasoning with means-ends reasoning. This
view is inadequate for at least two reasons. First, practical reasoning can, like
theoretical reasoning, be hypothetical. In this case there is no end for which
appropriate means can be sought. If Jones argues ‘If we want a drink we ought
to find a café,’ he does not need to endorse the end of having a drink; and
often we engage in practical reasoning based on premises we do not believe
also. Second, even if a piece of instrumental reasoning has an endorsed end,
the conclusion need not be a means to it, as this example can illustrate: ‘I must
listen to Fidelio since I want to listen to an opera written by Beethoven and
it is the only opera Beethoven has written.’ Listening to Fidelio constitutes
listening to an opera written by Beethoven; calling it a ‘means’ to it would
stretch the meaning of this term. Moreover, practical reasoning includes what
is sometimes called ‘constitutive’ reasoning. If my end is outdoor exercise and
I have the opportunity to go hiking (which is outdoor exercise and not a means
to it), then I can conclude that I should go hiking.
However, much of our instrumental reasoning does have a valuational premise
and the reasoning is about how to satisfy this valuation. But philosophers are
divided over the nature of this valuational premise. They have claimed that the
so-called ‘major premise’ of a practical inference must express an intention
Georg Spielthenner 93

(Searle 2001, p. 249), a goal (Audi 1982, p. 25), a Fiat (Kenny 1978, p. 64), a
want (Anscombe 1974, p. 19), or an interest (Grice 1978, p. 174), to list only
some suggestions. While I do not doubt that practical reasoning can have such
premises, I hold that there is no need to restrict instrumental reasoning to any
one of them. In the example at the beginning, the reasoner may very well hold
that it is ethically inappropriate to use maize for fuel production. We therefore
need to take a broad view on what the valuational premise of instrumental
reasoning is. Some authors claim that the major premise expresses a desire.
If, as usual in philosophy, ‘desire’ is taken as a term of art covering wanting,
valuing, having a goal, preferring, and many other things (see Dreier 1997, p.
87), I can accept this view because this notion of desire corresponds to what I
mean by ‘valuation.’
There is also fundamental disagreement in the philosophical literature over
what the conclusion of instrumental reasoning is. Some hold that it is an action
(e.g. Anscombe 1963, p. 60), while others reject this as inadequate and contend
that it must be an intention (Broome 2001, p. 175), a decision (Searle 2001, p.
91), or an imperative (Gensler 1996, p. 47); and while some (e.g. Clarke 1985,
p. 46; Raz 1978) claim that the conclusion must be a normative judgement
– a view I adopt here as will be apparent shortly – this is denied by J. Broome
(1999, p. 414) who holds that the conclusion of instrumental reasoning is not
normative. There is no room here to discuss these disagreements. I agree with
Broome that instrumental reasoning is as such not ought-giving, as will be
argued shortly, but it is a plain fact that our practical deliberations have often
normative conclusions (see the examples above), and philosophers are not
warranted to exclude such reasoning from their investigations.
I will not press this point here. However, what I want to emphasise is that
instrumental reasoning can have conclusions of different strengths. In this
study, I will restrict my attention to reasoning that has an obligatory conclusion
– often referred to as deontic (or practical) necessity – that can be expressed
by using terms such as ‘must,’ ‘ought to,’ or ‘should.’ Instrumental reasoning is
not always of this kind. Sometimes our conclusion is weaker, for instance when
we conclude that φ-ing might be better than ψ-ing or that φ-ing is preferable
to ψ-ing.

  There is an extensive debate about whether instrumental reasoning is always “desire-


dependent“ or whether there are also “desire-independent“ instrumental reasons, to use Searle’s
(2001) terms. Searle and other so-called externalists argue that there are desire-independent
reasons also. In this paper, I am concerned with desire-dependent reasoning (whose valuational
premise is, however, often not explicitly stated). But this does not commit me to the view that
there are no other instrumental reasons.
94 The Logical Assessment of Practical Reasoning

3  Instrumental reasoning can be evaluated in numerous ways; for example,


logically, epistemically, or rhetorically. This paper is about a logical appraisal of
practical reasoning. In a logical investigation we study the relationship between
the premises and the conclusion; and since I deal here only with deductive
practical reasoning, we ask whether the former entail the latter. In other words,
in a logical evaluation of practical reasoning we want to know whether the
acceptance of the premises logically commits one to the acceptance of the
conclusion in the sense that it would be inconsistent to assent to the premises
but deny the conclusion. In short, the purpose of this essay is to investigate the
criteria that make instrumental reasoning logically valid.
Since logical appraisals have often been conflated with other kinds of
assessments, a clear demarcation from these other topics of the theory of
practical reasoning is important. (i) This paper is not about practical reasons.
Valid practical reasoning does not ipso facto provide reasons (even not pro
tanto reasons) for doing what the conclusion prescribes. In this respect it is not
different from theoretical reasoning. To infer from the premise ‘Everyone is
happy’ that Smith is happy is valid; but to state the obvious, this premise does
not provide a reason for believing that Smith is happy because it is false and
unjustified. Having a reason requires also in practical reasoning more than a
logical support of the conclusion by the premises. The logical connection is
only a necessary condition for a good practical reason.
(ii) I am not concerned in this paper with practical rationality either. It can
be quite irrational to believe the conclusion of a valid theoretical argument and
acting on the conclusion of a valid practical reasoning can be irrational also.
Practical rationality depends on a number of issues that we cannot here explore.
For instance, it implies epistemic problems and, according to Searle (2001, p.
141), it even involves metaphysical issues such as the problem of free choice.
Consequently, this paper is not about instrumentalism, which can be roughly
defined as the view that acting rationally simply means acting in a way that
contributes to satisfying the agents intrinsic desires whatever they are. Since
my concern is not with practical rationality this paper has no implication on
the problem of instrumentalism.
(iii) Another source of confusion is the fact that we tend to attach deontic
modalities to the conclusion of our reasoning. Consider this simple example: ‘If
you want to reduce you ought to exercise.’ This is grammatically correct but it
  For instance, von Wright (1978, p. 48) contends that third person practical inferences
are ‘not logically conclusive’ because the agent need not know that the premises are true. This
confounds logical with epistemic problems. Similarly, Searle (2001, p. 246) confuses logical
considerations with problems of justification of the premises and with the rationality of actions.
Similar mistakes can be found in Clarke (1985).
Georg Spielthenner 95

misrepresents the logic of what is said. What we should rather say is something
like ‘It ought to be that if you want to reduce you exercise.’ More technically,
the ‘ought to’ should govern the entire conditional rather than its consequent. We
attach modalities to the conclusion in theoretical reasoning also; for instance,
when we say that Smith must be happy if everyone is happy. We are aware
that strictly speaking we should rather say ‘It is necessary that Smith is happy
if everyone is happy’ since we know that the premises of a valid argument do
not necessitate the conclusion. As I have already mentioned, Broome (1999,
2002) holds therefore that the conclusion of practical reasoning cannot be a
normative judgement. He is, of course, right in insisting that the fact that a
piece of practical reasoning is valid does as such not imply that the conclusion
should be acted upon (even not prima face) because, for instance, the premises
might be unjustified or false. But this does not warrant the view that practical
reasoning cannot have ought-statements as a conclusion. Reasoners do usually
not only hold that the premises logically support their conclusion, they are (often
rightly) also convinced that they have good reasons for their premises and in
this case they are indeed justified in drawing normative conclusions.
This is not the place for a full-scale discussion of this problem. However,
what should be clear by now is that a logical assessment of practical reasoning
is not concerned with the question whether the conclusion ought to be acted
upon. In calling practical reasoning valid we are just saying that it is inconsistent
to conjoin the premises with the denial of the conclusion. It says nothing about
whether the conclusion ought to be acted upon or whether the premises are a
reason for doing what the conclusion prescribes.
In this paper I shall first define when practical reasoning is valid. The
Sections II to IV are concerned with the problem when a practical argument
can be said to be valid. I shall investigate this first by proposing an informal test
for the invalidity of such arguments, second by criticising the main attempts
to explain the validity of practical arguments, and finally by outlining my own
account of the logical structure of a valid practical argument. In Section V I
propose a logical system of practical reasoning and apply it in the final Section
to the example given at the beginning of this essay.

I
We are convinced that some of the practical reasoning we encounter is valid
and some is logically defective. This raises the question of how to determine
the validity of this kind of reasoning. One way to accomplish this is to regard

  Broome (1999, 2001, 2002) prefers to speak of ‘correct’ practical reasoning instead of
valid reasoning. But the latter term is by far more common.
96 The Logical Assessment of Practical Reasoning

practical reasoning as valid if the underlying argument, which is expressed by


this reasoning, is valid. Let me briefly explain this: There is a distinction, not
always noticed, between reasoning and an argument. Reasoning is an activity
that takes place in the reasoner’s mind. It takes us from mental states which
we can call ‘premise-states’ to the ‘conclusion-state,’ to use Broome’s (2001,
p. 176) terms. This passage is guided by a sense of a support relation that is
assumed to hold between the premises and the conclusion. It is now important
to distinguish between reasoning as a process and the argument (roughly, a
sequence of statements) that is expressed by a piece of reasoning. An example,
which we have already considered, can illustrate this: We need to make a
distinction between the practical argument that I must listen to Fidelio since it
is the only opera Beethoven has written and I want to listen to an opera written
by him and the process of reasoning from the premises to the conclusion of this
argument. The argument is valid (as can be proved after reading this essay)
and therefore the reasoning can aptly be called valid also. In the light of these
considerations we can now define:
Practical reasoning is valid iff the underlying argument, which is expressed by
this reasoning, is valid.

This definition is intended to be a precising definition. It confers a precise


meaning to a term that has neither in everyday discourse nor in philosophy
a clear meaning. Obviously, this aim can only be achieved if I can explain
when a practical argument is valid. Much of the remainder of this paper will
be devoted to this topic.

II
Since the next section will be concerned with the main attempts to determine
the validity of a practical argument, it will be useful to have a method of testing
their adequacy. A theoretical argument is valid iff it is contradictory to have
the premises all true and the conclusion false. It is, however, obvious that the
validity of practical arguments cannot be defined in this way because their

  Be it noted that ‘premise’ and ‘conclusion’ refer to mental states when I use them in the
context of reasoning but to propositions when I refer to arguments. But I do not think that this
ambiguity causes any problems.
  Some authors speak of ’practical inference’ (e.g. Clarke 1985) or ‘practical syllogism’
(e.g. Anscombe 1978 or von Wright 1978). However, since ‘practical inference’ is ambiguous
as it can also mean the process of reasoning and ‘practical syllogism’ suggests that a practical
argument has only two premises (the ‘major’ and the ‘minor’), which is not correct, I think that
the term ‘practical argument’ is more appropriate.
Georg Spielthenner 97

conclusions need not be true or false (holding the contrary would commit us
to ethical cognitivism). We need therefore a broader account of ‘valid’ which
avoids reference to truth-values. Since the concept of consistency is central to
the idea of validity we can state the following principle:
If a practical argument is valid, then assenting to the premises and denying the
conclusion is inconsistent.

The key notion is now “inconsistency” which may be regarded as a suspect


notion because it appears that in explaining it we must again refer to truth-values
(a set of statements is inconsistent if it is impossible for all of them to be true).
Rather than explicate practical inconsistency here, I will attempt to make it
intuitively clear by paraphrasing it and by giving some examples.
In short, if a practical argument is valid, accepting the premises and rejecting
the conclusion is a logical mistake. This basic idea has been expressed in
different ways. Searle (2001) holds that the acceptance of the premises of a
valid practical argument ‘commits one to the acceptance of the conclusion’ (p.
241); according to Gensler (1996, p. 16), inconsistency of the premises and the
conclusion means that we ought not to combine accepting the premises with
accepting the conclusion; and Richard Hare contends that ‘he who assents to
the premises is compelled not to dissent from the conclusion, on pain of logical
inconsistency.’
Let us consider some examples: Since ‘Everyone is happy’ entails ‘Smith
is happy,’ accepting the premise but denying the conclusion would be a logical
mistake. Put differently, it would be inconsistent. Similarly, if I want to reduce
and believe that I reduce only if I exercise, but deny that I should exercise,
then I am practically inconsistent. (It is understood that there are no hidden
premises, as this would change the argument.) We can say that accepting
the premises commits or compels me to accepting the conclusion, or that I
ought not to combine accepting the premises with denying the conclusion
as this would be inconsistent. All these phrases express the same idea that
assenting to the premises and rejecting the conclusion would be illogical.
(More examples will be given below and this will further clarify the notion of
practical inconsistency.)

III
In the light of these considerations, let us now consider the main attempts to state
the formal logical structure of a valid practical argument. I shall discuss these

  Quoted from Kenny (1978, p. 75).


98 The Logical Assessment of Practical Reasoning

patterns here because it is important to understand why they are unsuccessful


and because this discussion provides the background for my own account.

1  The necessary condition pattern: Some philosophers think we get a valid


practical argument if acting on its conclusion is (causally) necessary for
achieving an end. I shall discuss here only John Broome’s view in some detail
because it is the most elaborate account of this group.
According to Broome (1999, 2001, 2002), reasoning to a necessary condition
is valid (‘correct’) if its propositional content constitutes a valid assertoric
argument. Let us consider one of his examples to illustrate this. To argue ‘I
want to open the wine. In order to open the wine, I must fetch the corkscrew
and therefore I shall fetch the corkscrew’ is valid, on Broome’s account. What
makes it valid is the fact that its contents – consisting of the premises ‘I open
the wine’, ‘In order to open the wine, I must fetch the corkscrew’ and the
conclusion ‘I fetch the corkscrew’ – are a valid argument (according to the
standards of classical logic).
I admit that at first sight the argument appears to be valid. But on closer
examination it is obvious that arguing in this way is not valid. In Broome’s
example, acting on the conclusion is more than a necessary condition for
achieving the end. The argument states explicitly that fetching the corkscrew is
necessary for opening the wine, but it implies10 that fetching it also contributes to
this end. The argument appears to be valid only because of this implication.
This will become evident, I conjecture, if we consider an example where
acting on the conclusion is purely necessary. Assume that you want to turn the
radio on, and you know that to turn it on you must press the power button.
Furthermore, you know that they have switched off the power. Pressing the
power button is still a necessary condition for turning the radio on, but obviously
it is now pointless, and no one would contend that it follows from the given
premises that you should press the power button. The reason for this is that
  For instance, Audi (2004, p. 128) holds that the ‘necessary condition schema’ of practical
reasoning is valid and Gensler (1996, pp. 46-7) contends that means-ends reasoning is valid
if the means is causally necessary for the end. G. H. von Wright (1978, p. 48) regards the fol-
lowing argument as valid: ‘I intend to make it true that this hut is habitable. Unless I heat the
hut, I shall not achieve this. [Therefore] I will heat the hut;’ and Clarke (1985, p. 59) thinks that
reasoning to a necessary condition is at least ‘prima facie valid.’ Brandom (2001) distinguishes
between ‘formally valid’ and ‘materially valid’ and holds that necessary condition arguments
are materially valid (p. 469).
  Broome needs some restrictions on this claim such as that the conclusion is not true anyway
(2002, p. 5), but we can safely ignore them here.
10  Throughout this paper, I use ‘imply’ in a non-technical sense. It means suggesting
something indirectly rather than stating it directly.
Georg Spielthenner 99

doing so is purely necessary – that is, it does not contribute to the achievement
of your end in any way. In general, whenever acting on a conclusion is only
purely necessary for achieving an end, a practical argument is not valid (it is not
even inductively strong). All examples of reasoning to a necessary condition that
can be found in the philosophical literature may have a considerable intuitive
appeal, but this is so only because the context implies that the premises are
more than a necessary condition.

2  The sufficient condition pattern: On this view, a practical argument is valid


if acting on its conclusion is sufficient for satisfying a desire.11 I shall focus
here on A. Kenny’s ‘logic of satisfactoriness’ which has received a considerable
amount of attention.
In Kenny’s (1966, 1978) logical system, a practical argument is valid if
acting on the conclusion satisfies what is wanted. I want to illustrate this basic
idea by one of Kenny’s examples (see 1978, p. 76):
I am to be in London at 4.15
If I catch the 2.30 I’ll be in London at 4.15
So I’ll catch the 2.30.
On Kenny’s view, this is a valid argument because from acting on the conclusion
(catching the 2.30) and the minor premise it follows that the major premise
is satisfied (I am in London at 4.15). To make the structure of this argument
more transparent, we can symbolize it using ‘l’ for ‘I am in London at 4.15’
and ‘c’ for ‘I catch the 2.30’ and (as suggested by Kenny) ‘Fiat’ for valuational
statements and ‘Est’ for descriptive ones. We get: Fiat(l), Est(c → l), therefore,
Fiat(c). This is valid because the theoretical argument c, c → l therefore, l is
valid, Kenny tells us.
Let us now change the example slightly:
I am to be in London at 4.15
If I catch the 2.30 or the 2.45 I’ll be in London at 4.15
So I’ll catch the 2.30.
In Kenny’s system, this argument is still valid. But by applying our inconsistency
test, we can clearly see that it is not. If I assent to the premises but deny the

11  Among the advocates of this account are Dreier (1997) who holds that ‘if you desire to
ψ and believe that by φ-ing you will ψ, then you ought to φ (p. 93), Davidson (2001), and Audi
(2004, 128-9), who support this kind of reasoning albeit not unconditionally. This shows that
this pattern has still supporters and I am not flogging a dead horse.
100 The Logical Assessment of Practical Reasoning

conclusion I am not inconsistent because I have another option for achieving


my end (catching the 2.45).
This simple example reveals the basic flaw of Kenny’s system: If an agent
has more than one alternative for satisfying his desire, then it does not follow
that he ought to act on the conclusion. Kenny’s examples appear valid at an
intuitive level because they imply more than a sufficient condition. Claiming
that I will be in London at 4.15 if I catch the 2.30 states explicitly that catching
the 2.30 is sufficient, but it implies that I will not be there at that time if I do
not catch it. Only both together make the argument putatively valid.12
That practical arguments with a purely sufficient conclusion are not valid,
can easily be seen if we return to the example which we have already considered
and change it slightly. Suppose now that you want to turn the radio on and you
know that you can turn it on by pressing the power button. But you can turn it
on by using the remote control, too. Pressing the power button is still sufficient
for turning the radio on but it is not justified to conclude that you should press
it. Accepting the premises does not logically commit you to accepting the
conclusion (that you should press the power button) and therefore the argument
in invalid.13

3  The optimality pattern: A number of authors have held that reasoning to the
best means of achieving an end is valid.14 Since John Searle made a real effort
to develop a logic of best-means reasoning and presented in (2001) ‘the best I
can come up with’ (p. 247), I shall concentrate here on his account.
In the example discussed by Searle, it is supposed that you want to go to
Paris; you figure out that the best way to get there is going by plane and you
decide therefore to go by plane. Searle analyses this argument as follows (p.
246):
Des (I go to Paris).
Des (If I go to Paris I go by the best way, all things considered).
Bel (The best way, all things considered, is to go by plane).
Therefore, Des (I go by plane).
12  In addition to this objection, a number of counterexamples can show that Kenny’s logic
is defective. For example, the following invalid arguments are all valid in his system: Fiat(p),
therefore, Fiat(p ∧ q); Fiat(p ∨ q), therefore, Fiat(q); or Fiat (∃x)Px, therefore, Fiat Pa.
13  Among the critics of the sufficient condition pattern are Clarke (1985, p. 25), Raz (1978,
p. 11), and Searle (2001, p. 243). I do, however, not agree with all objections they raise.
14  For example, Audi (2004, p. 129) claims that if a means is the best way to achieve an end
‘it apparently does follow’ that the agent should do it. A similar view has been held by Clarke
(1985), and Broome (2002, p. 16) gives this example of an ‘intuitively correct’ argument: ‘I am
going to buy a boat. The best way for me to buy a boat is to borrow money. So, I shall borrow
money.’
Georg Spielthenner 101

In this analysis, ‘Des’ and ‘Bel’ stand for the intentional attitudes of desiring and
believing and the brackets enclose the propositional contents of these mental
states. Searle then symbolizes the propositional contents of this argument using
‘p’ for ‘I go to Paris,’ ‘q’ for ‘I go by the best way,’ and ‘r’ for ‘I go by plane,’
getting hereby this formal argument: p, p → q, q ↔ r, therefore, r.
Searle is dissatisfied that his way of representing best-means reasoning needs
an extra premise stating that acting on the conclusion is the best way of achieving
the end (p. 247). Furthermore, if we conceptualise best-means reasoning as
suggested by Searle, it seems to be invalid. The propositional contents of the
premises entail the propositional content of the conclusion and acting on the
conclusion is necessary for achieving the end, but this does not make a practical
argument valid, which Searle recognizes correctly when he writes that it ‘is by
no means obvious that a rational person who has all those premises must have,
or be committed to having, a desire to go by plane’ (p. 247).
Searle is rightly disappointed with his attempt of developing a logic of
practical reasoning and he admits that after ‘various unsuccessful tries I have
reluctantly come to the conclusion that it is impossible to get a formal logic of
practical reason…’ (p. 248). By outlining my account of valid practical reasoning
in the remainder of this paper I shall attempt to prove that this conclusion was
too hasty.

IV
The object of this section is to sketch my account of the logical structure of
a valid practical argument. In particular, I shall try to show how we can get
a definition of validity for practical arguments. Even though we use the term
‘valid practical argument’ (and similar phrases) in everyday discourse and in
philosophy, it does not have a clear meaning, and some philosophers doubt that
it has any meaning at all. What is needed, therefore, is a precising definition
of this term. I shall first outline my view informally, leaving details aside for
the present.
In a nutshell, a practical argument is valid if the conjunction of the premises
and the contradictory of the conclusion are inconsistent. This raises, of course,
the question of when this conjunction can be said to be inconsistent. A first
requirement is that doing what the conclusion prescribes guarantees that a
(possible or actual) desire is satisfied. In a valid theoretical argument, it is
impossible that all premises are true but the conclusion false. In a valid practical
argument, by contrast, it is impossible that we do not attain what we value if
we do what the conclusion prescribes and all premises are true. This was the
basic idea of Kenny and I agree that it is part of the correct answer. However,
102 The Logical Assessment of Practical Reasoning

that acting on the conclusion guarantees the realization of a valuation is not


enough for a practical argument to be valid, as our discussion of the sufficient
condition pattern has shown. We can only infer that we ought to do something
if the converse also holds – that is, if not acting on the conclusion guarantees
that this valuation is not realized. For a practical argument to be valid, both must
be guaranteed. Summarizing this we can roughly say that in a valid practical
argument, it follows from acting on the conclusion that we attain what we value
and from not acting on the conclusion, it follows that we do not attain this.
In the light of these remarks we can now informally define practical validity
as follows:
A practical argument is valid if the relation between premises and conclusion
is such that (i) from acting on the conclusion, there follows the consequence
which is most valued by the agent and (ii) from not acting on the conclusion,
there follows that this con­sequence does not occur.

A few words of explanation will be in order: The phrase ‘acting on the


conclusion’ means that the agent does what the conclusion pre­scribes. The
term ‘follows’ is taken in the strict sense of classical logic. Of course, strictly
speaking nothing follows logically from doing something. The passage must
therefore be understood as short­hand for ‘From the propositional content of
the conclusion, there follows…’ For example, if the conclusion is ‘John ought
to exercise more,’ then this can be paraphrased as ‘It ought to be that John
exercises more,’ where ‘John exercises more’ is the propositional content of
this judgement. From this propositional content it follows (together with other
premises of the argument), for instance, the consequence that he keeps fit.
That the consequence ‘is most valued by the agent’ implies firstly, that one
of the prem­ises of a practical argument needs to be about a valuation. ‘John
wants to keep fit’ is a statement about a valuation. Sometimes, especially in
first-person reasoning, this premise is a value judge­ment (e.g. ‘I disapprove of
abortion’), but in hypothetical reasoning it is not itself a valuation. Secondly, the
consequences must be ranked. To keep things as simple as possible, I assume
here that the ordering of the consequences is such that there is a consequence
that is most valued.15 It should be noted, however, that this condi­tion is not
always satisfied; but this assumption simpli­fies my exposition. If there is only
one consequence, ‘most valued’ means that this conse­quence is preferred to
its not occurring. Thirdly, since there is no valuation without a valuer, it must
15  This requires that the ordering is (i) connected. If we choose the symbol ‘≥’ for a valuer’s
preference relation, this relation is connected, if, for any two distinct consequences (x, y), either
x ≥ y or y ≥ x (or both). It requires (ii) transitivity. The valuer’s preference relation is transitive,
if, for all triplets of consequences (x, y, z), x ≥ y and y ≥ z entail x ≥ z. Furthermore, since ‘most
Georg Spielthenner 103

be clear whether valid reasoning requires a restriction on who values the


consequences. Most authors take for granted that in a valid practical argument
the agent is the same person as the valuer. A consideration of this issue would
this would take us beyond the confines of the present work. In this paper, most
examples imply that the agent is identical to the valuer. However, this may not
be required for valid reasoning and my definition of practical validity gives
therefore only a sufficient condition for validity – that is, the definiendum has
only been partially defined.

V
On the basis of these informal explanations I shall now briefly explain a logic
of practical reasoning. This will not only make the remarks in the preceding
section clearer, it will also show how to prove the validity or invalidity of a
great variety of practical reasonings.16
The system of practical reasoning P builds upon propositional logic. Assume
that we have a language of classical propositional logic with
an infinite set of propositional variables a, b, c, … etc.
the monadic operator ¬.
the dyadic operators ∧, ∨, →, ↔.
the brackets (, ).
We add the two one-place operators
O (the deontic operator), to be read ‘It ought to be that.’
V (the valuation operator), to be read ‘x values most that.’
Grammatically correct formulas of P are sequences that we can construct using
these rules:
1  Any propositional variable standing alone is a correct formula.
2  The result of prefixing any correct formula with ‘¬,’ is a correct
formula.
3  The result of prefixing any correct formula that does not contain ‘O’ or
‘V’ with ‘O’ or ‘V’ is a correct formula.

valued’ means here ‘uniquely most valued,’ it requires (iii) that (∃x)(∀y) (x ≠ y → x > y), where
‘>’ stands for the irreflexive relation ‘x is more valued than y.’ That is, there must be a conse-
quence that is more valued than any other.
16  I explained the system in more detail in Acta Analytica 22(2), pp. 139-53, 2007.
104 The Logical Assessment of Practical Reasoning

4  The result of joining two correct formulas with ‘∧,’ ‘∨,’ ‘→,’ or ‘↔’
and enclosing the result with brackets is a correct formula.
Comments on the operators ‘O’ and ‘V’: As mentioned earlier, the deontic
operator ‘O’ is intended to express a ‘deontic necessity’ – or a ‘practical
necessity,’ to use von Wright’s term. I have suggested that we read ‘V’ (the
valuation operator) as ‘x values most that.’ However, other possible readings
are ‘x prefers most that,’ ‘It is most preferred by x that,’ ‘It stands highest in
the valuer’s ranking that,’ ‘The valuer prefers no other outcome to,’ ‘It has the
highest value for x that,’ or ‘x puts the highest value on.’
We can now make our informal concept of practical validity more precise
and define validity-in-P as follows:
A practical argument is valid-in-P if both its corresponding arguments are
valid.

The first corresponding argument is the argument from acting on the conclusion
to the occurrence of the most valued consequence. The second corresponding
argument is the argument from not acting on the conclusion to the not-occurrence
of the most valued consequence. Con­structing these arguments is simple: Drop
the operators ‘V’ and ‘O’ from the original prac­tical argument and replace the
formulae governed by these operators with the adequate for­mulae. To explain
what these ‘adequate formulae’ are, I must specify what acting and not acting
on a conclusion means and what it means that the most valued consequence
does occur or does not occur.
i  If the conclusion is of the form Oφ,17 acting on the conclusion means doing
φ and not acting on it means doing the alternative that follows from ¬φ and
the (exclusive or inclusive) disjunc­tion of the given alternatives. Consider this
valid argument: a → x, b → y, Vx ╞ Oa.18 The argument con­sists of the two
alternatives a and b with their respective consequences x and y. The expression
‘Vx’ means that the agent values x higher than y (i.e. values x most), and the
conclusion ‘Oa’ is to be read as ‘It ought to be that a.’ Doing what this conclusion
prescribes means doing a. Not doing it, is tantamount to doing b which follows
from ¬a and (a ∨ b) – or, if the alternatives are mutually exclusive, from ¬a
and (a + b).19

17  The symbols φ and ψ are here used as meta-variables ranging over formulae of P.
18  A semantic sequent φ , …, φ ╞ ψ, using the double turnstile ‘╞’, says that ψ follows
1 n
validly from φ1, …, φn.
19  I use ‘+’ as the symbol for exclusive disjunctions.
Georg Spielthenner 105

Since practical reasoning is defeasible, system P is a non-monotonic logic.20


That is, if a conclusion follows from given premises, then it need not follow
from a larger set of premises, even if the original premises are included. Let a
simple example serve to illustrate this. If we add the premise c → x to the valid
argument discussed in the previous paragraph then it becomes invalid.
ii  If the conclusion is of the form O¬φ, it is the other way round: Acting
on the conclu­sion means doing the alternative that follows from ¬φ and the
given alternatives, while not acting on it means doing φ (more precisely ¬¬φ).
For example, a → x, b → y, Vx ╞ O¬b. Doing what the con­clusion prescribes
means here doing a which follows from (a ∨ b) ∧ ¬b and not doing it means
doing b. Also this argument is valid.
iii  The rules for the consequences are analogous. Consider this valid argument: a → x,
b → y, c → z, Vz ╞ O¬(a ∨ b) where ‘x,’ ‘y’ and ‘z’ stand again for consequences
and ‘a,’ ‘b’ and ‘c’ for alterna­tives. Acting on the conclusion means here doing c
because it follows from the given alternatives (a ∨ b) ∨ c and ¬(a ∨ b). The first
corresponding argument is thus a → x, b → y, c → z, c ╞ z. Conse­quently, not
acting on the conclusion means doing (a ∨ b). That the most valued consequence
z does not occur, means that (x ∨ y) occurs because this conse­quence follows from
(x ∨ y) ∨ z and ¬z. The second corresponding argument is therefore a → x, b →
y, c → z, a ∨ b ╞ x ∨ y.21
Proving the validity of practical arguments is now possible by using the
standard techniques for testing the validity of arguments in propositional logic
such as the truth tree method or natural deduction. To prove that the argument
(a ∨ b) → x, c → y, Vx ╞ O(a ∨ b) is valid, we only need to show that both of
its corresponding arguments are valid. The first corresponding argument is (a
∨ b) → x, c → y, a ∨ b, ╞ x and the second corresponding argument is (a ∨
20  The usual classification of non-monotonic logical systems, however, does not seem to be
appropriate for the logical of practical reasoning because it has been developed for theoretical
reasoning.
21  There is a further obvious rule for arguments in P. The alternatives and the ranked conse­
quences must be logically independent. That is, an alternative or consequence must not be entailed
by other alternatives or consequences. It does not make sense to prefer φ to ψ if ψ must occur
if φ occurs or if ψ cannot occur if φ occurs. This does, however, not mean that only logically
independ­ent consequences can be considered in P, it rather means that those consequences that
are evaluatively ordered to get the most valued one must not logically depend on each other. An
example that violates both conditions is (a ∧ b) → x, x → y, a → z, Vx ╞ O(a ∧ b). From doing
(a ∧ b) it follows the alternative a, and from doing a it follows that z which is not the most valued
conse­quence. In addition, doing what the conclusion prescribes implies not only the preferred
conse­quence x, but also the not preferred y because the conse­quences are not independent. The
problem can, however, be resolved, if we represent the argument as follows: (a ∧ b) → (x ∧ y),
(a ∧ ¬b) → z, V(x ∧ y) ╞ O(a ∧ b). The alternatives and the con­sequences are now logically
inde­pendent and the argument is valid.
106 The Logical Assessment of Practical Reasoning

b) → x, c → y, c ╞ y. Both can easily be proved to be valid and therefore the


practical argument is valid. Another example is this valid argument: a → x, b
→ y, c → z, V(x ∨ y) ╞ O(a ∨ b). The first corresponding argument is a → x, b
→ y, c → z, a ∨ b ╞ x ∨ y - that is, doing the conclusion entails the most valued
consequence. From not doing the conclusion follows that this conse­quence
does not occur; the second corresponding argument is thus a → x, b → y, c
→ z, c ╞ z. The formula ‘z’ follows from the disjunction of the consequences
and ¬(x ∨ y).
On further remark is called for. On my account, a practical necessity can
only be derived conclusively if an act is sufficient and necessary for the most
valued outcome. If the relationship between an act and its consequences is
probabilistic there follows only a weaker conclusion, for instance, that an act
is preferable to its alternatives. However, since the logical system P does not
include probabilities, I shall not pursue this question here. It would, however,
be profitable to consider this issue at greater length on another occasion.

VI
We are now in a position to test the validity of the piece of practical reasoning
presented at the beginning of this paper. According to the definition given
in Section I, this reasoning is valid if its contents constitute a valid practical
argument. To prove that this practical argument is valid we need to show that
its corresponding arguments are valid. Like most of our practical reasoning, this
example is enthymematic. It states explicitly that the price of maize, other food
crops, and meat will go up if we convert food into fuel, but the tacit premise is
that this will not happen if we refrain from converting maize into fuel. It would
not make sense, for example, to claim that using biofuel made from maize ‘will
drive up’ the price of maize if one did not implicitly hold that this rise will not
happen if we refrain from using this fuel. Furthermore, the valuational premise
has also been left out, as it is often the case in practical reasoning. But it is
obvious that the reasoner prefers the prices not to rise.
Let ‘b’ mean ‘We use maize to make ethanol biofuel,’ ‘d’ mean ‘The price
of maize will be driven up,’ and let us use ‘c’ for ‘The price of other food crops
rises,’ and ‘m’ for ‘The price of meat goes up.’ If we interpret the argument
as suggested, we get this formalized argument: b → (d ∧ c ∧ m), ¬b → ¬(d
∧ c ∧ m), V¬(d ∧ c ∧ m) ╞ O¬b. This argument is valid because the first
corresponding argument: b → (d ∧ c ∧ m), ¬b → ¬(d ∧ c ∧ m), ¬b ╞ ¬(d ∧ c
∧ m) is valid22 and the second corresponding argument: b → (d ∧ c ∧ m), ¬b
22  Please note that even though the conclusion is of the form O¬φ, doing the conclusion
means here doing ¬b because from (b ∨ ¬b) ∧ ¬b follows ¬b.
Georg Spielthenner 107

→ ¬(d ∧ c ∧ m), b ╞ ¬¬(d ∧ c ∧ m) is valid also. (If we symbolize the second


premise as ¬b → ¬(d ∨ c ∨ m), as might be suggested to be more adequate,
the argument is still valid.)

As I have already mentioned, the fact that this argument is valid does not ipso
facto mean that we have a reason for not converting maize into fuel since the
premises might be completely unjustified. Therefore it may not be rational to
act upon the conclusion either; and for the same reason it may be wrong to
say that we must oppose the idea of converting food into fuel (even if this is
interpreted as a prima facie obligation). There was no room here to give these
matters the attention they deserve. The question I have been addressing in this
study was to determine when instrumental reasoning is logically valid.
Georg Spielthenner
Department of Philosophy and Applied Ethics
University of Zambia
georg_spielthenner@yahoo.de

References
Anscombe, G. E. M. (1963)  Intention (2nd edition). Oxford: Blackwell.
— (1974)  ‘Comment’. In: St. Körner (ed.): Practical reason. Oxford: Basil
Blackwell, pp. 17-21.
— (1978)  ‘On Practical Reasoning’. In: J. Raz (ed.): Practical reasoning. Oxford:
Oxford University Press. pp. 33-45.
Audi, R. (1982)  ‘A theory of Practical Reasoning’. American Philosophical
Quarterly 19(1), pp. 25-39.
— (2001)  The Architecture of Reason: The Structure and Substance of Rationality.
Oxford & New York: Oxford University Press.
Brandom, R. (2001)  ‘Action, Norms, and Practical Reasoning’. In: E. Millgram
(ed.): Varieties of Practical Reasoning. Cambridge, MA: MIT Press, pp. 465-79.
Broome, J. (1999)  ‘Normative Requirements’. Ratio 12, pp. 398-419.
— (2001)  ‘Normative Practical Reasoning’. Proceedings of the Aristotelian Society
(Supplementary Volume) 75(1), pp. 175-93.
— (2002)  ‘Practical reasoning’. In: J. Bermùdez & A. Millar (eds.): Essays in the
Theory of Rationality. Oxford: Oxford University Press, pp. 85-112.
Clarke, D. S. (1985)  Practical Inferences. London: Routledge & Kegan Paul.
Dreier, J. (1997)  ‘Humean Doubts about the Practical Justification of Morality’.
In: G. Cullity & B. Gaut (eds.): Ethics and Practical Reason. Oxford: Clarendon
Press, pp. 81-99.
108 The Logical Assessment of Practical Reasoning

Gensler, H. J. (1996)  Formal Ethics. London: Routledge.


Grice, G. R. (1978)  ‘Motive and Reason’. In: J. Raz (ed.): Practical Reasoning.
Oxford: Oxford University Press, pp. 168-77.
Kenny, A. (1966)  ‘Practical Inferences’. Analysis 26, pp. 65-75.
— (1978)  ‘Practical Reasoning and Rational Appetite’. In: J. Raz (ed.): Practical
Reasoning. Oxford: Oxford University Press, pp. 73-80.
Raz, J. (1978)  ‘Reasons for Actions, Decisions and Norms’. In: J. Raz (ed.):
Practical Reasoning. Oxford: Oxford University Press, pp. 128-43.
Searle, J. R. (2001)  Rationality in Action. Cambridge, MA: MIT Press.
Wright, G. H. v. (1978)  ‘On so-called Practical Inference’. In: J. Raz (ed.): Practical
Reasoning. Oxford: Oxford University Press, pp. 46-62.

Você também pode gostar