4. SENTIMENTAL CAUSES
4.3 Instmmental Rationality
4.3.2 Problems: Reciprocity
Imagine driving down a narrow country-lane when suddenly a car approaches from
the opposite direction. W ithout previous conventions about behaviour on roads,
should you keep to the left or to the right? Obviously you would do best to keep to the
right if the other car does so too, but equally you'd better keep to the left if that is what
the other does. The trouble is that under the maximizing interpretation the other thinks
exactly the same. W ithout additional information there is no successful way to
coordinate your approaches. In practise, this is o f course what you would seek. You
would try to provide each other with indications on which side you intend to pass.
You might hold sharply to the left and see what the other does.
If coordination seems problematic even in cases where both parties pursue a
common goal they can only reach together (not to collide, say) what about reciprocity where both parties have competing interests they can only satisfy through cooperative
activity? The problem has been acutely form alized in the so-called Prisoner's
Dilemma. Two rogues have committed a crime. Now in prison each is faced with the
options to confess thus incrim inating his partner or to keep mum. If only one
confesses he is let o ff lightly for turning in State's Evidence while the other receives
the maximum penalty (say 10 years). If both confess, they receive 5 years each
(m itigating circum stances). If both remain silent they receive a much reduced
sentence for lack o f evidence ( 1 year each).
Predicting their behaviour under the constraints o f instrum ental rationality
seems to suggest that both rogues will confess, thus paradoxically revealing
possible. A must have thought: if B confesses, I do better to confess; if B keeps mum,
I do better to confess — and vice versa.
A variation o f the Prisoner's Dilemma is the more general case o f promising. A
and B, foreseeing the situation, have promised each other to remain silent. But under
the m axim izing interpretation the parties will never successfully establish the
institution o f promising since the second party would always have reason to abort
coordinated activity after it received the benefits advanced by the first party.
Now most o f us are to some degree trustworthy, we keep many promises, we
drive on the right (that is in Great Britain: left) side. This is part o f our socio
economic behaviour and — as one might want to insist — one o f the more reasonable
parts. Admitting the existence o f reciprocity, the theory o f choice may move into two
directions. Either the endeavour must be to show that people in successful
coordination conform in some unexpected way to the constraints o f instrum ental
rationality or the theory has to accept that people characteristically do not act
rationally. In the latter case, the theory o f choice gives up its explanatory claims and
may therefore have to rethink the foundations o f the maximizing conception. This is
no easy undertaking since interpretations under the constraints o f instrum ental
rationality were convincing in the first place because o f its explanatory credentials.
O nce the gap between preference and choice reopens, we may ask again more
fundamental normative questions: why should we maximize? Why should we take the
second-order attitude to act out certain preferences and not others? I shall pursue this
line o f thought in time. But first let us return to the first response.
Might not a utility-maximizing explanation o f reciprocity be available? Could it
not be ultimately rational for the two rogues to keep mum and, in a wider sense, for
most o f us to keep promises? David Gauthier has taken this route, most detailed in
Chapter Six o f his Morals by Agreement. ' There he argues that a rational person (in a
Gauthier 1986. Gauthier claim s that M orals by Agreem ent grew out o f a deliberation o f the Prisoner's D ilem m a (v). M y cursory treatment o f Gauthier does not do justice to the intricate argument o f his book. Still, the central difficulties facing normative system s derived from instrumental rationality can be brought out clearly from Gauthier's discussion. R eferences to M orals by A greem en t in this section
strategic setting) "chooses on utility-maximizing grounds not to make further choices
on those grounds" (158). The rational person will do better overall if she disposes
herself to become trustworthy.
The disposition to keep one's agreement, given sufficient security, w ithout appealing to directly utility-maximizing considerations, makes one an eligible partner in beneficial co-operation, and so is in itself beneficial. (162)
This would solve Prisoner's-Dilemma-type situations since A could count on B to
perform after he (A) has undertaken the first move. So the explanatory claim is that a
maximizing interpretation can be given to the tendency to be stably disposed not to
abort cooperation single-handedly.
we do not purport to give a utility-m axim izing justification for specific choices o f adherence to a joint strategy. Rather we explain those choices by a general disposition to choose fair, optimizing actions whenever possible, and this tendency is given a utility-maximizing justification. (189)
There are in effect two distinct actions. The action to dispose oneself to become
trustworthy and the further action o f making specific choices in strategic settings.
While the former can be given a maximizing interpretation under the constraints o f
instrumental rationality, the second defies this attempt. An example invented by Parfit
elucidates the point. W hile it may be rational to dispose m y self to becom e a
threat fulfil 1er (since anybody threatened by me will then be more likely to comply
with my demands) it may not be rational to carry out my threat after it has been
ignored (say, blow up the aircraft). Thus one may reject the claim that
If it is rational for someone to make him self believe that it is rational for him to act in some way, it is rational for him to act in this way. (Parfit, 1984, 23)
The difference between the explanations o f disposing oneself to become trustworthy
and actually performing specific acts in strategic Prisoner's Dilemma situations can be
brought out clearly. For each single strategic choice the party will do best that would
only pretend to be disposed to be a restrained maximizer but abort cooperation after it
received the advanced benefits. ’
are revealed by page-numbers alone. Luce & Raiffa (1985 (1957), 94) attribute the Prisoner's D ilem m a to A.W . Tucker.
Gauthier tries to convince us that this is not the case. Partially transparent as we
are (G authier uses the term "translucent", 174) we cannot expect deceptive
motivations to remain hidden. Thus Gauthier insists that there is good reason why we
should become trustworthy, that is dispose ourselves to constrained maximization.
This is a normative claim which explicitly presupposes a gap between preference and
choice. We prefer in the first place to maximize but, after deliberation, choose not to
do so.
... the capacity to make such choices [among dispositions] is itself an essential part o f human rationality ... At the core o f our rational capacity is the ability to engage in self-critical reflection. The fully rational being is able to reflect on his standard o f deliberation, and to change that standard in the light o f reflection. Thus we suppose it possible for persons, who may initially assume that it is rational to extend straightforward maximization from parametric to strategic contexts, to reflect on the implications o f this extension, and to reject it in favour o f constrained maximization. (183-4)
If Gauthier had given primacy to maximization as an explanatory theory a way out
might have been to ascribe to cooperating parties a different set o f preferences. In
performing reciprocal actions, they reveal preferences, say, for trustworthiness which
again can be given a maximizing interpretation under the constraints o f instrumental
rationality. '
Instead, Gauthier opts for the second, normative route. Though people do not
always act ideally rationally, if they reflected critically they would. In fact they ought
to. Gauthier still insists that the only critical considerations available to rational agents
are maximizing reasons. Gauthier would have to argue why only maximizing reasons
offer a "sure grounding" (17) since maximization, on our reading o f Gauthier's
exposition, just had failed to explain reciprocity. If the normative problem is which
attitudes and dispositions survive critical reflection, we have to set the frame wider
The Prisoner’s Dilem m a, however, w ill remain unsolved i f the utilities are as they have been stated above (cf. Binm ore, 1993). Purely self-seeking parties under the constraints o f instrumental rationality w ill fail to cooperate successfully. This may be sad for moral theory but not necessarily bad new s for the explanatory credentials o f the theory o f choice. Again it has been questioned whether the axiom s o f the theory o f choice can allow m otivations like "trustworthiness". Instrumental reasons are forward- looking reasons where utility is only attached to consequences. "Trust" or "Having promised", on this account, appear to be backward looking utilities referring to the history o f the situation. (See H ollis & Sugden, 1993, 27 ff.)
and make from the start psychological assumptions about what kind o f motivations
people have. Curiously, Gauthier does precisely that. He adopts the m axim izing
constraint o f instrumental rationality subject to Hobbes’ m aterial conditional "that
each seeks above all his own preservation" (159). Thus a person's rational choices, the
choices she ought to make, are those that further her interest - and I must stress again:
'interest' is here not coextensive with the technical term 'utility' in rational choice
theory since people may attach utility to acting altruistically, adopting the best means
to that end. Gauthier introduces the technical sense:
Let us suppose it is agreed that there is a connection betw een reason and interest - or advantage, benefit, preference, satisfaction, or individual utility, since the differences among these, important in other contexts, do not affect the present discussion. (6)
— only to continue with interest in the sense o f self-interest:
Morality, we have insisted, is traditionally understood to involve an impartial constraint on the pursuit o f individual interest. (7)
This terminological uncertainty may account for the confusion between the normative
and explanatory role o f interpretations under the constraint o f instrumental rationality
we have encountered. ' I f the debate was conducted under the supposition that people's
m otivations are purely self-seeking we might reassess some o f the results while
skipping the maxim izing apparatus o f the theory o f choice. W hat G authier has
perhaps shown is that purely prudentially motivated persons cannot cooperate
successfully and that people who are disposed to be trustworthy can.
In identifying purely prudential m otivation as a kind o f sm allest common
denominator o f people's possible choices, we don't say that all people are self-seeking
but ask whether people with such m inim ally conceived ends could be party to
mutually beneficial cooperation. The problems o f the two tier structure o f dispositions
The confusion continues into other parts o f Gauthier's book. In chapter one "Reason and Value" Gauthier locates the norm ative elem ent in standard econ om ic theory correctly as interpretative constraint "expressed by the single injunction, 'Maximize!' To say that one should m axim ize utility as a measure o f preference adds nothing, since utility is sim ply identified with w hatever one's behaviour m ay be interpreted as maximizing." (27) W hile in the introduction econom ic theory is described as part o f normative inquiry itself. "... the role o f econom ics in formulating and evaluating p olicy alternatives should leave us in no doubt about the deeply prescriptive and critical character o f the science." (3)
and single decisions besetting maximizing explanations o f reciprocity reappear now in
much the same guise for normative expectations: People see a gap betw een their
preferences and choices. They ask themselves which choice they ought to make, and
again they may find they would do best (given Hobbesian motivations) to pretend to
be trustw orthy and break rank whenever they can do so at reasonable cost. The
dom inant strategy is only to appear to have a certain disposition. This would
undermine norms o f reciprocity as prudential norms.
Under these gloomy motivational assumptions, it remains strange that in daily
life so many cases o f reciprocity are available. We cooperate successfully in many
ways and even keep promises. So people may not be only self-seeking after all. But is
there a way to show that we are normatively compelled to enter mutually beneficial
cooperations? And does instrumental thinking over minimally conceived ends support
one unique set o f norms? Maybe not. The most we might feel able to do is point out
more precisely which motivations we would have to have in order to overcome the
contractualistic dilemma. This seems to me a worthwhile task, the more since the
justification o f reciprocity is not the only challenge to contractualism. It m ay be
theoretically the most fundamental one (since it targets the possibility o f rational
agreement itself) but practically speaking other consequences o f instrumental thinking
based on a Hobbesian psychology are much more dramatic. I shall briefly sketch two
o f the central scenarios: asymmetric and exclusive agreements.