Académique Documents
Professionnel Documents
Culture Documents
DOI: 10.1007/s00355-003-0309-8
Abstract. In decision theory goals are usually taken as given inputs to the
analysis, and the focus is on nding the most ecient means to achieve the
goals. But where goals are set with the purpose of achieving them, it is
important to know what properties they should possess in order to be successful (or achievement-inducing). Four such properties (or rationality criteria) are discussed, namely that goals should be precise, evaluable,
approachable and motivating. Precision and evaluability are epistemic
properties that concern what the agent may know. Approachability is an
ability-related property that concerns what the agent can do. Motivity is a
volitional property that concerns what the agent wants to do. Goals may
satisfy the rationality criteria to a greater or lesser extent. Some goals are
achievement-inducing mainly because they guide action towards the end-state
well, others mainly because they motivate the agent to act towards the realization of the end-state.
1 Introduction
Setting goals, both nal goals and interim targets, plays an important part in
many decision processes. However, although decision theory encompasses a
vast literature on eciency and optimization in relation to given goals, not
much has been written by decision theorists on goal-setting. A possible reason
for this may be that, as John Searle has pointed out, our intellectual culture has a
quite specic tradition of discussing rationality in terms of means and ends.1
Searle (2001), p. 5.
344
K. Edvardsson, S. O. Hansson
This tradition may have its roots in a statement by Aristotle that deliberation is
always about means and never about ends. 2 Modied versions of this statement
have been put forward by several writers in our own time. Hence, Herbert
Simon writes that reason is wholly instrumental. It cannot tell us where to go;
at best it can tell us how to get there. It is a gun for hire that can be employed in
the service of whatever goals we have, good or bad.3 Bertrand Russell states
that reason has a perfectly clear and precise meaning. It signies the choice of
the right means to an end that you wish to achieve. It has nothing whatever to do
with the choice of ends.4 Similarly, Maurice Allais writes that it cannot be too
strongly emphasized that there are no criteria for the rationality of ends as such
other than the condition of consistency. Ends are completely arbitrary. To
prefer highly dispersed random outcomes may seem irrational to the prudent,
but for somebody with this penchant, there is nothing irrational about it. This
area is like that of tastes: they are what they are, and dier from one person to
the next.5 In decision theory goals are usually taken as given inputs to the
analysis, and the focus is instead on what means are most ecient in achieving
the goals. It is also very often assumed that the goals, or ends, in question are
fairly well-determined, and that it is not the task of decision theory to dwell
upon the contents of these goals.6
Nevertheless, as the quote from Allais shows, there is still room to
deliberate about the non-substantial, or structural, properties of ends.
Allais himself, however, nds only one such property, namely consistency.
Till now, most discussions on the non-substantial properties of ends
have taken place in management theory. Despite having emerged in the
private business eld, the idea of management by objectives (MBO) is
today embodied in many public sector areas, such as education, health and
environmental policies.7 A crucial part of MBO concerns goal-setting, and
some writers have purported to trace down the non-substantial properties
that characterize private and public management goals. These accounts
include requirements that goals should be clear, concise and unambiguous,
Nicomachean Ethics 1112b 1112. Note however that it has been pointed out that
this standard attribution may be a misunderstanding of Aristotle, and that the passage
should not be taken simply as an instance of ordinary means-ends reasoning, see for
instance Annas (1993), p. 88 and Richardson (1997), p. 58.
3
Simon (1983), pp. 78.
4
Russell (1954), p. 8.
5
Allais (1979 [1952]), p. 70.
6
Deliberation on the substantial content of goals is often taken to be the task of moral
philosophy. The traditional division of labour between decision theory and moral
philosophy is questioned in Hansson (2001).
7
The term management by objectives was introduced by Peter Drucker in 1954. It
usually denotes goal-setting applied to the macro (organizational) level. See Drucker
(1954), p. 62 .
345
within the competence of man, challenging, measurable, evaluable, integrative, complex, dynamic, transdisciplinary, applicable, participatory and
understandable.8 A set of conditions that is often put forward in management literature is summarized in the acronym SMART.9 According to
the SMART conditions, goals should be (S)pecic, (M)easurable,
(A)chievable or sometimes (A)ccepted, (R)ealistic and (T)ime-bound.
In our view, it emerges from management literature that, in addition to the
requirement of consistency which Allais mentions, there is a wide range of
non-substantial properties of goals that should be taken into account in an
analysis of rational goal-setting. However, no systematic philosophical account of these properties and their relations seems to have been made.
Properties are often suggested without or with only a marginal discussion on
the function of goals and the reason behind goal-setting. In this paper, an
outline of such an account will be proposed.
Before presenting our analysis of rational goals we introduce some preliminary remarks about the starting-points and delimitations of our analysis
in Sect. 2. In Sect. 3 we discuss what goals are for and introduce the central
term achievement-inducing. Based on the ndings of section 3 we then
introduce four rationality criteria for goals in Sect. 4. In Sects. 57 we specify
what we mean by each of these criteria and why they can be used as rationality criteria for goals. The relationships that hold between the criteria are
discussed in Sect. 8. In Sect. 9 we give our conclusions.
Odiorne (1969), p. 25, Carroll and Tosi (1973), p. 72, Raia (1974), p. 25, Pihlgren and
Svensson (1989), p. 42, Slocombe (1998), p. 484. See also Rombach (1991), pp. 4750
and Ehn and Sundstrom (1997), pp. 136143.
9
See for instance van Herten and Gunning-Schepers (2000), pp. 89 and Ds 2000:63,
p. 54.
346
K. Edvardsson, S. O. Hansson
10
A goal can of course also be set not with the purpose of achieving it but to achieve
some similar but less demanding goal. A government may for instance set the goal
reduce unemployment to 1% within two years while the true goal is to reduce
unemployment to 3% in that period. In our view, such goals are part of goal systems
and will be dealt with at a later occasion for reasons indicated below.
11
Another procedural aspect of goal-setting that we do not deal with here is whether a
self-set goal is likely to be more achievement-inducing than a goal set by an external
agent. There is however some empirical evidence in favour of such a claim. See for
instance Carroll and Tosi (1973) pp. 78.
12
Strictly speaking, auto-instrumentality is not self-referential. The word goal can
denote either (a) the objective to achieve a specic state of aairs (S) or (b) S itself.
Auto-instrumentality means that the goal in sense (a) is instrumental to the goal in
sense (b). Neither the goal in sense (a) nor the goal in sense (b) is instrumental to itself.
347
3 Achievement-inducing goals
As indicated above, our starting-point is simple enough. We assume that
goals are set with a purpose to achieve them, at least to some degree. By a
rational goal we mean a goal that it is rational to set with the purpose of
achieving it. Hence, in this paper rationality will be closely tied to the notion
of functionality. A rational goal is a goal that performs its achievementinducing function well. In the light of this, our analysis of rational goalsetting will take the form of an investigation into what criteria a goal should
satisfy in order to be achievement-inducing.
For a goal to be achievement-inducing it must both guide and motivate us
to act in ways that further the desired end-state to which the goal refers. A
goal is action-guiding when it directs and coordinates action, over time and
between agents, towards the desired end-state. This action-guiding function is
perhaps most evident when a group of agents sets a goal, or when an agent
sets a common goal for one or several other agents. In these situations the
goal, which is known among the agents, is achievement-inducing because it
coordinates actions temporally and between the agents.13 For instance, assume that in a country with a multi-party system where elections are made on
the basis of proportional representation a group of voters who traditionally
vote for the Social Democratic Party have as their aim to produce an altogether socialist government in the majority position. This can either be done
when the Social Democratic Party obtains more than half the number of
votes, or when the Social Democratic Party together with the Left Party
obtains more than half the number of votes. Since the former alternative
seems improbable, some of the voters in the group might consider it advantageous to vote for the Left Party just to assure that an altogether socialist
government in the majority position can be formed. On the basis of the goal
the voters can coordinate their voting informally so as to produce the desired
outcome.14
In some situations the goal is also action-guiding because it makes the
behaviour of every single agent somewhat predictable to the others, at least in
13
See Bratman (1987), p. 28 . for a similar discussion on plans and planning.
According to Bratman, by making plans we (1) avoid the need for deliberation at the
time of action, and (2) coordinate our activities over time and between us in ways
compatible with our limited capacities to deliberate and process information.
Temporal and social coordination of action facilitates the achievement of complex
personal and social goals that we would not otherwise be able to achieve without
inordinate cost.
14
The same would apply in rst-past-the-post electoral systems such as the British,
where MPs are elected through single member constituencies. In such a constituency
the goal of some minority of Liberal Democrat voters may be to produce a nonconservative government. Instead of wasting their vote on their own candidate and
risking that a conservative MP will be elected, they may choose to vote for the stronger
Labour Party candidate just to facilitate the formation of a non-conservative
government.
348
K. Edvardsson, S. O. Hansson
the sense that it disqualies some actions. The goal is a ltering device. It
orients the agent towards goal-relevant activities and away from goal-irrelevant ones.15 In a situation where the publicly stated goal of A, B, C, and D is
to jointly travel from London to Bath before 10 PM tonight, As actions
become predictable to B, C, and D, at least in the sense that there is a
multitude of actions that B, C, and D can reasonably expect A not to perform, such as taking a ight to Japan. Because B, C, and D can rely on As
behaviour to some extent, they can themselves perform actions whose outcome is dependent on As specic behaviour.16
A possible objection against this reasoning is that goals can have at most a
subordinate role in guiding and coordinating actions, something which is
instead primarily done with another type of intentional objects, namely plans.
If this objection is correct, then our mode of analyzing goal-setting is on
shaky grounds. Therefore, before proceeding with our analysis we need to
have a clear picture of the relation between goals and plans.
The dierence between a goal and a plan is in principle simple. Whereas a
goal species an end-state to be achieved, a plan species a sequence of
actions to be taken. Admittedly, there are intermediate, or mixed, forms
between goals and plans. Goals are often supplemented by action plans, and
plans typically include goals. Hence, Kaufmann denes a plan as a set of
goals and rules that denes the expected contribution of each actor to a chain
of actions in order to reach a desired outcome.17 On the other hand, a
system of goals that contains a series of time-xed subgoals may be dicult to
distinguish from a plan. However, the existence of such intermediate or mixed
forms does not invalidate the distinction between goals and plans. The distinction is clear although its application in individual cases may cause
problems.
A major reason why goals need not be inferior to plans in terms of guiding
and coordinating action, is that they are often better equipped to mobilize the
creative capacities of agents. This is indeed one of the major arguments in
favour of management by objectives.18 A plan prescribes a certain sequence
of actions and usually does not leave much scope for creativity and improvisation. A goal, on the other hand, typically states a desired end-state and
leaves it to the agent to determine the appropriate actions to be taken in order
15
For instance, experiments show that individuals who are given a list of words will
recall them better when told that the goal is to memorize them, than if they are told
that the goal is to nd typographical errors. See Locke and Latham (1990), p. 92.
16
Nozick (1993), pp. 912 has a similar discussion on the ltering function of
principles.
17
Kaufmann (1986), p. 218.
18
Pihlgren and Svensson (1989), p. 17. Very often complex planning systems are
deemed to be insuciently exible for the demands of modern society. MBO is
understood as not only increasing goal realization through improved means-exibility,
but also decreasing the amount of time and money spent on administrating complex
planning systems.
349
19
350
K. Edvardsson, S. O. Hansson
20
351
5 Epistemic properties
A goal can be action-guiding only when the agent knows what the goal is and
to what extent her own actions lead her closer to fullling it. This requires
that the goal is both precise and possible to evaluate. Of these two criteria,
precision is more fundamental, since evaluablility depends on it. In order for
it to be possible to evaluate whether a goal has been achieved or not, the
content of the goal must be (reasonably) clear. An imprecise goal cannot be
subject to constructive examination, assessment and critique.
The requirement that goals should be precise can be divided into three
subcriteria, namely directional, completive, and temporal precision. A goal is
directionally precise if it is specied in what direction one should go in order
to reach the goal. Consider the following example:
(1) The number of unemployed in this country should be substantially decreased.
Assuming that the notion of an unemployed person is suciently precise, (1)
is precise in the sense that it can be unequivocally determined whether or not
it has been approached. If unemployment has decreased, then the goal has
been approached or achieved, otherwise not. However, (1) does not tell us to
what degree the goal should be reached. It lacks completive precision. The
following variant of (1) satises both directional and completive precision:
352
K. Edvardsson, S. O. Hansson
21
22
353
23
24
354
K. Edvardsson, S. O. Hansson
partial success may lead to the choice of quite dierent means to achieve
the goal.
6 Ability-related properties
It is a common accusation against goals that they are unrealistic, or
utopian. Utopian goals are no good, it is claimed, because they cannot
be achieved. In his classic defence of piecemeal social engineering, Karl
Popper argues that the setting of step-by-step goals that may be adjusted to t
future economical, social, and technological change, is preferable to utopian,
or holistic, social engineering.25 Since the piecemeal social engineer reforms
society by carefully comparing the achieved results with the expected results,
she is able to adapt to the uncertainties of the human factor. The utopian, on
the other hand, due to her wish to remodel the whole society at once and her
unpreparedness to adapt to unavoidable surprises, is forced to control all
unwanted eventualities that follow from this uncertainty by institutional
means. She must extend her programme so as to embrace not only the
transformation of society, according to plan, but also the transformation of
man.26 Thereby, Popper argues, the utopian social engineer creates a
totalitarian society.
The notion of a utopian goal is generally used in a rather imprecise way.
Popper and other critics have mostly conceived utopianism as a rather rough
binary distinction: a goal is either utopian (unrealistic) or it is not. An attempt
to specify the concept was made by Larry Laudan in Science and Values
(1984) with reference to goal selection in scientic contexts. In order to assess
scientic goals Laudan introduces a set of conditions, according to which
utopian goals should be abandoned. He distinguishes between three forms of
utopian goals. Goals can be demonstrably utopian, semantically utopian,
and/or epistemically utopian.27 A goal is demonstrably utopian when it
cannot possibly be achieved, given our understanding of logic or the laws of
nature.28 A goal is semantically utopian if we cannot precisely characterize it
in a succinct and cogent way. Imprecise and ambiguous goals are therefore
utopian. Finally, a goal is epistemically utopian when we are unable to specify
a criterion for determining when it is achieved, even though we may be able to
give a perfectly clear denition of the goal.
Laudans category of semantically utopian goals corresponds to those
goals that, in our terminology, lack severely in precision. His epistemically
utopian goals coincide largely with the goals that do not satisfy our criterion
25
355
29
30
356
K. Edvardsson, S. O. Hansson
7 Volitional properties
As we have seen in Sect. 3, goals can be achievement-inducing both by
guiding and by motivating action. Action-guidance is not sucient for the
goal to be achieved. The agent must also be motivated to act towards the
goal. Therefore, the rationality criteria discussed above should be supplemented with an additional criterion: a rational goal should motivate the right
kind of action. The volitional aspect of goal-based human action concerns
what we may want to do, or accomplish, and it focuses on the motivation that
goals may give rise to in agents.
Several objections could be made against our argument that the rationality criteria should be supplemented with this fourth criterion. One could
for instance argue that all goals are internally motivating since the end-states
to which they refer are always desired by the agent. In short, we do not have
to require that goals should be motivating, they simply are motivating. This
may very well be the case when an agent sets a goal for herself, since it would
indeed be strange for an agent to set a goal she does not desire to achieve. But
many goals are set by a person or institution for some other agent or group of
agents. Such goals are not in all cases internally motivating.
A second objection may be raised, namely that inducement to action
should come from sources other than the goal itself. Action could get started
on many grounds, for instance by way of legally enforced rules, threats, or by
giving favours to those who further the goal. On this view, the rationality
357
criteria that apply to goals should not include motivating force. In our
opinion, although motivation to act may come from other sources than the
goal itself, this is no reason why it should not also come from the goal itself.
Therefore, it makes sense to retain this fourth criterion. Findings in management and psychology literature also provide support for the argument that
a motivating goal increases performance towards the desired end-state.31
The motivation that a goal may give rise to in the agent can be characterized according to degree of intensity, or durability, or both. The intensity
of the motivation that a goal gives rise to is in typical cases reected in the
degree of eort that the agent displays in furthering the goal. Its durability is
reected in the extent to which this eort is maintained over time. Intensity
and durability are independent criteria. Motivation may be intense and shortlived, or moderate and durable.
What makes goals motivating, in terms of intensity or durability, is of
course largely their content. But as established earlier, we are only interested in the non-substantial properties of goals. So we leave the content of
goals aside, and focus instead on how the motivating force of goals relates
to the previously introduced non-substantial properties. In psychology and
management literature, studies show that goals are more action-generating
when they are explicit and specic, and that such goals are more likely than
do-your-best goals to intensify eort towards the later part of long work
periods.32 There is also evidence to suggest that specic and challenging
goals lead people to work longer at a task than other goals, and that
employees in initially low motivation groups catch up to high motivation groups when they are given specic goals to accomplish.33 As regards the criterion of evaluability it has been argued that goals that are
accompanied with outcome feedback can have a positive eect on performance.34 This is believed to originate from the motivational eect that
individuals experience when they are able to determine where they stand in
relation to the goal.35 As regards the criterion of approachability it is on
the one hand generally assumed that a dicult, or challenging goal, increases performance.36 On the other hand, some studies suggest that a goal
which is considered to be too dicult to achieve decreases performance
slightly.37 The lesson to be learnt from these eld experiments is probably
31
Locke and Latham (2002), p. 707 argue that the goal-performance relationship is
strongest when an agent is committed to her goal.
32
Jungermann et al. (1983), p. 234 and Locke and Latham (1990), p. 88.
33
Locke and Latham (1990), p. 90 and Bryan and Locke (1967), p. 277.
34
Locke and Latham (2002), p. 708.
35
McCalley and Midden (2002), p. 591. For a discussion of the relationship between
feedback and performance see also Carroll and Tosi (1973), pp. 57, Locke and
Latham (1990), pp. 173205, and Hall and Kerr (2001), pp. 201203.
36
van Herten and Gunning-Schepers (2000), p. 89, Locke and Latham (2002), p. 706.
37
Stedry and Kay (1966), p. 465.
358
K. Edvardsson, S. O. Hansson
359
9 Conclusion
Our starting point is that a rational (functional) goal is one that it is rational
to set with the purpose of achieving it. Starting out from this basic requirement we have identied four criteria for rational (functional) goals, namely
that goals should be precise, evaluable, approachable, and motivating. We
have ordered these criteria into three categories which reect three aspects
that are involved in goal-based human action. Precision and evaluability are
epistemic properties which concern what the agents know. Approachability is
an ability-related property which concerns what the agents can do. Motivity is
a volitional property that concerns what the agents want to do. It follows
from these criteria that a rational (functional) goal also needs to be consistent.
Each of the criteria that we discuss makes a goal function better in the
achievement-inducing sense. The criteria are, however, often present to
varying degrees in dierent goals and must be balanced from case to case in
order to optimize goal realization. One way of balancing the rationality
(functionality) criteria is to adopt a goal system in which goals on dierent
levels supplement each other so that the system as a whole both guides and
motivates action. However, further empirical research is needed before we can
proceed to the task of analyzing our proposed criteria in relation to goal
38
cf. Barber and Taylor (1990), p. 368. Carroll and Tosi (1973), pp. 6970 argue that
one purpose of MBO is to facilitate the derivation of specic objectives from general
ones so that objectives at all levels in the organization are both appropriate to each
level and also linked to each other.
360
K. Edvardsson, S. O. Hansson
References
Allais M (1979 [1953]) The foundations of a positive theory of choice involving risk
and a criticism of the postulates and axioms of the American school. In: Allais M,
Hagen O (eds) Expected utility hypotheses and the Allais paradox. D. Reidel,
Dordrecht, pp 27145
Annas J (1993) The morality of happiness. Oxford University Press, New York
Aristotle, Nicomachean Ethics (1999) Translated by Terence Irwin, 2nd ed. Hackett
Publishing Company, Indianapolis
Barber WE, Taylor JN (1990) The importance of goals, objectives, and values in the
sheries management process and organization: A review. North Am J Fisheries
Manag 10(4): 365373
Bratman ME (1987) Intentions, plans, and practical reason. CSLI Publications,
Stanford, California
Bryan JF, Locke EA (1967) Goal setting as a means of increasing motivation. J Appl
Psychol 51(3): 274277
Carroll SJ, Tosi HL (1973) Management by objectives: Applications and research. The
Macmillan Company, New York
Chen HT (1990) Theory-driven evaluations. Sage Publications, Newbury Park, CA
C ntora A (1999) Critical comments on Laudans theory of scientic aims. Sorites 10:
1938
Drucker PF (1954) The practice of management. Harper and Row, New York
Ds (2000)63 Ekonomisk styrning for eektivitet och transparens. Fritzes oentliga
publikationer, Stockholm
Ehn P, Sundstrom G (1997) Samspelet mellan regeringen och statsforvaltningen. In:
Lindstom S, Ehn P, Sundstrom G, Det Svara Samspelet Resultatstyrningens
framvaxt och problematik, SOU 1997:15
Hall HK, Kerr AW (2001) Goal setting in sport and physical activity: Tracing
empirical developments and establishing conceptual direction. In: Roberts GC (ed)
Advances in motivation in sport and exercise. Human Kinetics, Champaign, Ill.,
Leeds
Hansson SO (2001) The modes of value. Philos Stud 104: 3346
House ER (1980) Evaluating with validity. Sage Publications, Inc., Beverly Hills,
London
Jungermann H, von Ulardt I, Hausmann L (1983) The role of the goal for generating
actions. In: Humphreys P, Svenson O, Vari A (eds) Analysing and aiding decision
processes. North-Holland Publishing Company, Amsterdam, New York, Oxford
Kaufmann FX (1986) The relationship between guidance, control and evaluation. In:
Kaufmann FX, Majone G, Ostrom V (eds) Guidance, control, and evaluation in
the public sector. W. de Gruyter, Berlin, New York, pp 211228
Laudan L (1984) Science and values. University of California Press, Berkeley
Locke EA, Latham GP (1990) A theory of goal setting and task performance. Prentice
Hall International, Englewood Clis, NJ
Locke EA, Latham GP (2002) Building a practically useful theory of goal setting and
task motivation: A 35-year odyssey. Am Psychol 57(9): 705717
McCalley LT, Midden CJH (2002) Energy conservation through product-integrated
feedback: The roles of goal-setting and social orientation. J Econ Psychol 23: 589
603
Nozick R (1993) The nature of rationality. Princeton University Press, Princeton, NJ
361