Even intuitively standard
descriptions of what rational behavior consists of face strange difficulties. I
cite two instances. It is fairly common to think that rational action entails
some element of choice. Nobody is rational in breathing, since nobody decides
to breath. Choosing seems to entail a range >1 of potential outcomes. One is
supposed to survey mentally, as it were, the field of available choices.
Secondly to (try to) assess their foreseen or foreseeable consequences. Thirdly
establish a hierarchy by means of a matching procedures between outcomes and
one’s own prior preferences.

A weird remark: much of our mental
life (where "we" comprises many humans in leisure societies) is spent
in finding out what to do at all, not really in assessing means to ends. Often
we do not know what our ends are. This is however an external problem: we do
not really know in which sense people have preferences and why they have those
they do have. Note that in any event there is always an element of choice,
albeit a non conscious one. Lifetime is a scarce commodity and doing "nothing"
is a choice after all on the allocation of such a commodity. The issue of what
forms, if anything, preferences is best left to some other occasion.

Closer to the issues in the text
under review is a second almost paradoxical aspect of rationality thus
conceived. The elementary scheme above entails that one has to make hard
choices. It may very well be the alternatives that come to mind are not
exhaustive. In principle they are never exhausting physical, let alone logical,
possibilities. What tells me that the set of alternatives I consider is the
good one? One may very well concoct an infinity of situations in which the only
"rational" thing to do is keep searching for more and/or better alternatives.
But notice that this process itself is time/energy consuming hence it is a
choice in allocating my one and only lifetime. This is in a semitechnical
language known as a stopping problem. And it leads to a vicious circularity:
one has to perform the very same task to know when it is bet to keep looking
and when it is best to stop looking and make do with what one can come up with.
And this entails another search and so forth. The standpoint taken by this
collection is partly derived from the "satisficing" approach
advocated by one of the founders of modern behavioral science, H. Simon. When a
subject satisfices she stop searching at any time an alternative is found that
is as high or better than a pre-fixed aspiration level. One of the editors (Reinhard
Selten) has improved this approach.

The general idea of the collection
is to explore an adaptive toolbox (yet another mental box for the cognitively
inclined.) The box contains heuristics: methods that can deliver choices within
strict boundaries and that are, on the face of it, fairly irrational in an
intuitive sense. Why should I believe that the best stock to pick in trading
session in the stockmarket is the one whose name I recognise? The present
collection provides some evidence and some answers.

The adaptive toolbox is supposed to
be shaped at once by internal constraints and by environmental pressures. The
constraints are fairly easy to recognize: no (human) agent has infinite memory
or computing power to actually solve optimization problems by brute force. The
environmental pressure is a subtler contribution of Gigerenzer et al.’s
approach. A strategy within the toolbox may match in structural properties an
environment (p. 46) and may be robust in being able to generalize well. The
main idea behind the research program is to look at the evidence we have that
could explain our reliance on heuristics. And many of the papers here do
provide experimental evidence. To the mind of this reader two questions stand
out here. It is fairly clear that we do adopt heuristics, fast and frugal, the
simplest probably is based on mimicry. It is also fairly clear that we can
adopt more traditional methods. The evidence and theory here displayed is
impressive. In Goldstein et al. (Ch. 10) the question comes to the fore. "Which
homunculus selects among heuristics, or there is none?" (p. 188). My own
question is: how do we select between the frugal adaptive toolbox and the more
cumbersome and austere traditional "optimizing" model? It is my
opinion that we do not have a proof yet that only heuristics are used in
decision making.

On the critical side, the book
turns out to be very uneven. Possibly this is not avoidable in a collection
that reports a variety of approaches (from cultural anthropology to algorithm
theory) and research in progress. The most interesting part, for the
psychologically minded are probably those that deal with emotions (see in
particular ch. 15.)

But the absolutely most fascinating
section deals with an extremely clearheaded treatment of a problem that may
baffle us. We often conceive of organisms as single individuals (Adriano chooses
to write an article and Christian decides whether to publish it or not.) Nature
in its weird wisdom appear to display a different kind of choice-making
apparatus. Consider that swarms of bees have to decide where to place their
next colony once they have to move out of one home site. T. D. Seeley (ch. 15)
produces an extremely beautiful account of the fact that completely idiotic
single-bee-level mechanisms (and not well understood ones, see p. 260) crank
out decisions followed by the entire swarm. Alternatives are evaluated, though
each bee scout visits one and only one site, by having all of them dance to
signal locations. The mechanism of the debate is somewhat unclear (see p. 258
on comparative vs. noncomparative tactics); but evidence appears to point in
the direction of the "loser" bees dropping out of the debate far more
than by "persuading" anybody or anything of the superiority of their
finds. The key conceptual point here is a super-organism (the swarm) may reach
a decision (and an effective one at that – bees consistently pick better nests
in experimental settings) without having any central authority and any
particularly clever single-organism member.

The implication for human decision
making may be stark to be stared in the face. We decide nothing at all, we are
assemblies of bee-like mechanisms, some fast and frugal heuristics, some dumb
and cumbersome rationalistic -Leibnitzian computers. The consequences on the
ethics and deontology of the way we treat ourselves and even our own
psychopathologies may be immense.

© 2004 Adriano Palma

Adriano Palma is an associate of
Inst. Jean Nicod, a Cnrs-Ehess-Ens UMR (unite mixte de recherche) in Paris,
France. The institute is reachable at nicod.org

Categories: Psychology