<< . .

( 29)

. . >>

having nothing to do with purposes, either of nature™s design or of
mine. I then added a reference to tryings, thus importing a dimension
of purpose “ of what happens not by accident but by design “ and this
helped us quite a lot. But defining an ability as an individual™s likeliness
to succeed should they try in conditions they would likely be in if they
tried snags on exactly the same problem one level up. The conditions I
am in may not be conditions I was designed by nature to be in, or that
my abilities were designed through learning processes to operate in. If
(1) I have a disposition to succeed if I try, or if (2) I fail to have such a
disposition, these could be purely coincidental matters, hence might not
indicate an ability.
Consider the second case first, the case where I have an ability but
fail to have any such disposition. Suppose, for example, that the last
dodo is dead but the dodo stew chef doesn™t know it. And suppose
there are people about still posing as dodo pedlars, and that it is quite
likely the dodo stewmaker will be taken in by them.Then it will not be
true that if the dodo stewmaker tries to make dodo stew he will most
likely succeed. If he tries to make dodo stew it will be because he is
mistaken in thinking he has bought dodo meat. Recalling the distinc-
tion noted earlier between knowing how to do A and being in a posi-
tion to do A, it is clear that something has gone wrong. The dodo stew

chef surely does have the ability to make dodo stew, though he is in no
position to exercise it. But he lacks a disposition to make dodo stew
when under conditions he would probably be in if induced to try.
Now consider a case in which a disposition to succeed if one tries is
present but does not correspond to an ability. Macaffee the cat has
learned how to get himself let in the door by stretching up tall and
scratching by its side window, thus making himself both heard and seen
from inside. (Our cat does this.) Now the house changes owners, but
Macaffee won™t stay in his new home and returns to the old. The new
owners dislike cats and have no disposition at all to let him in. How-
ever, soon after his return they have an automatic door installed for a
wheelchair occupant with a wide push button outside at exactly the
place Macaffee is disposed to scratch. Now the first day after the instal-
lation it is true that Macaffee has a disposition to get the door open for
himself if he tries under the circumstances he is in or is likely to be in
when he tries. But he no more knows how, at this point, to let himself
in than you know how to win the state lottery if, happily, you happen
to do so. What one does successfully only by pure chance is not some-
thing one knows how to do.
Now, in the preface to this book I promised not to play Counter Ex-
amples in Queer Possible Worlds, and I do not mean to be doing so
here. My point is not that we wouldn™t say Macaffee knows how to let
himself in, that this is not how “our concept of an ability” is fashioned.
Rather, my point is that the same principle is involved here as in cruder
cases where dispositions very obviously do not equal abilities. Abilities
are distinct from dispositions in having a necessary involvement in the
purposive and nonaccidental order. A consistent theoretical definition of
abilities will consistently take this into account. What is interesting and
central about abilities will not be captured if we let in cases where re-
sults are achieved coincidentally, or if we exclude cases where failures to
achieve results are coincidental.
Notice also that the attempt to turn abilities into dispositions with
the antecedent “if one tried,” as we came to interpret that phrase, was a
cheat all along. It is not just that trying requires an analysis and is un-
likely itself to unpack into dispositions. Rather, the conditionals corre-
sponding to dispositions are supposed to refer to causes in their an-
tecedents and effects in their consequents, and we did not use “if one
tries” in that way. If the match lit, then it was struck may be a true condi-
tional, but it does not correspond to a disposition for the match to be

struck if it lights. Similarly, our “if one tries” was not used merely as a
reference to a cause.The idea was that if one tries that must be because
one believes one can succeed, and that in turn will likely be because
one recognizes one is in conditions that will make success possible. The
only true disposition here is the disposition to succeed if one tries un-
der certain conditions, namely, the ones one attempts to recognize.
Moreover one attempts to recognize these conditions under that very de-
scription, namely, as certain conditions under which one can succeed.The de-
scription of this disposition thus appears to be empty. One has a dispo-
sition to succeed if one is right that conditions under which one will
succeed are present.


We are still seeking the relation of an ability to the conditions of its
successful exercise. Let me start fresh with an example that may bring
out this relation more clearly. Suppose that you once learned how to
use WordPrefab 2.2, but for the last fifteen years you have had a stenog-
rapher and haven™t looked at any more recent word processing pro-
grams.WordPrefab 2.2 is now completely extinct and nobody anywhere
has kept it either on their computers or on disk. It is clear then that you
will no longer succeed in word processing if (without retraining) you
try, no matter how we interpret the antecedent of the conditional. But
do you still know how to do word processing?
A temptation is to reply that you don™t know how to do word pro-
cessing period; you only know how to do word processing in or by
means of WordPrefab 2.2. But consider: No one knows how to do
word processing without employing some program or other, some
means or other. So from that sort of reasoning it would follow that no-
body knows how to do word processing period. Similarly, since every-
one employs some stroke or other in order to swim, and since if we cut
strokes finely enough, surely no one knows how to employ every pos-
sible swimming stroke there is, it would follow that nobody knows how
to swim period. Indeed, since everything accomplished in, or in relation
to, the world outside one is accomplished by some means or another,
requiring some definite conditions or other to be in place in that out-
side world, it will turn out that no one has any abilities to affect the
world outside period. Rather, all abilities are hedged with specific
means and conditions.

In Section 1.9, I pointed out that we have two ways of distinguish-
ing or designating abilities: by ends, or by ends plus means. Where abil-
ities to identify substances are concerned, this is the distinction between
what I called “concepts” and what I called “conceptions.” What has
happened just above is that we have distinguished abilities by means
rather than merely by ends. If we distinguish abilities by means, then of
course no one ever knows how to do something full stop. The various
means they know must always be described in saying what they know
how to do. It is also true that if we cut designations by means finely
enough, it may well be that two people seldom if ever have exactly “the
same ability.”
Now, there is nothing wrong with distinguishing abilities in this
manner for some purposes. But that should not blind us to the legiti-
macy of distinguishing them by ends alone for other purposes. For
many purposes, it is quite legitimate to say that some people know how
to swim whereas others don™t, and that all those people who know how
to swim have the same ability. Similarly, it is quite legitimate to say of
the person who knows how to do word processing using WordPrefab
2.2 that they know how to do word processing. They can have an abil-
ity to do word processing, even though, unfortunately, the conditions
required for the means they employ (presence of a WordPrefab 2.2 pro-
gram on an accessible computer) are unavailable, so they are not in a
position to exercise that ability. Having an ability is not the same thing
as being able.
To see the distinction between counting abilities by their ends and
counting them by their ends-plus-their-means clearly, it is important to
keep another distinction made earlier in mind (Section 4.2). This is the
distinction between aiming at an end while also knowing to use certain
means to that end in certain circumstances, and aiming at getting to the
end in those very circumstances or by the use of those very means.
Aiming at word processing and aiming at word-processing-with-Word-
Prefab 2.2 are two different things to aim at; knowing how to do these
two things corresponds to two different abilities when abilities are
counted by ends only. This is very easy to confuse with the fact that
there are two different ways to count abilities, either by reference to
ends only or by reference also to means. Similarly, abilities as counted by
ends often have, for a given individual, many alternative means, em-
ployed under different conditions. This is not the same as the person
having many alternative ends. It is not the same as the person™s having
many alternative abilities as counted by ends. For it may be that the

person never aims at any of the specific means, or at being in conditions
in which these means specifically can be used. Thus, you walk on a
mountain trail constantly adjusting your means to the conditions com-
ing up underfoot, but you aim at none of these means nor at being in
any of the conditions requiring these means.


We have settled then that the person who learned how to use WordPre-
fab2.2 learned and still knows how to do word processing period,
whether or not WordPrefab2.2 still exists in any form. But there is no pos-
sibility at all in their current situation that their ability to do word pro-
cessing will manifest itself, whether or not they try. Thus it appears that
knowing how to do word processing has nothing at all to do with this
person™s dispositions in their current situation. At the same time, it is clear
under what possible conditions this ability could be manifested “ namely,
under the condition that this person would have WordPrefab2.2 on the
computer they used.What then IS the relation between this person™s abil-
ity and the circumstances under which it would necessarily be manifested?
The relation, it seems, is historical.The conditions in which this abil-
ity would be manifested are the conditions in which it was historically
designed as an ability. In general, the conditions under which any abil-
ity will manifest itself are the conditions under which it was historically
designed as an ability. These are conditions in which it was learned, or
conditions in which it was naturally selected for. They are conditions
necessary to completing the mechanisms by which past successes were
reached by the systems or programs responsible for the abilities. Past
presence of these conditions helped account for the selection or main-
tenance of the systems or programs constituting the abilities. What I
know how to do I must once have learned how to do. Otherwise it
would not be knowledge but mere luck. Or what I have an ability to
do is what my systems were maintained or selected for doing. That is
my suggestion. I know of no other way to understand what abilities are
that is consonant with the case advanced above against their merely be-
ing current dispositions of some kind. (This can also be taken as a chal-
lenge, of course, to come up with alternative suggestions.)
To an ability there always corresponds a disposition, but it does not
follow that an ability IS a disposition. If an ability to do A were a dis-
position to do A under specified conditions, then we should be able to

specify the conditions under which anyone must be disposed to do A if
and only if they have an ability to do A. But these conditions cannot be
specified for the general case. Because different people who have
learned to do the same thing may have learned to do it in different
ways, relying on different conditions, the conditions under which a
given ability (defined by its end) might manifest itself may be entirely
different for different people.There is no such thing as the set of condi-
tions under which it is necessary that any person be able to manifest
their ability to do word processing or to swim. For each person, there is
an independent reference here to personal history. Each person™s ability
to do A rests on a disposition defined through their particular past. Each
has a disposition to do A if they try to do A under the conditions that
accounted for their own past successes in doing A. If they have no such
disposition, of course, then they have lost the ability to do A. It does not
follow that their ability IS a disposition. Rather, which disposition(s) it
is that can manifest the ability to do A is determined by which condi-
tions helped account for the acquisition of this particular person™s abil-
ity. To attempt to define the ability by reference to its historically en-
abling conditions would move one in a circle.1
Another question that has been running in the background is now eas-
ily answered as well. How should we understand the notion “trying” such
that an ability is manifested when one succeeds in what one tries under
historically enabling conditions, yet such that there can be mental abilities
the goals of which we do not represent to ourselves or even understand,
hence do not “try” to reach in the most oridnary sense of “try”? “Trying”
to do A, in the needed sense, is simply the initiation and running to the point
of success or failure of a mechanism or program that is designed to do A.


What then determines what a learned ability is an ability to do? The
difficulty, recall, was that the kinds of parts, systems, and programs that
embody abilities have innumerable dispositions that are not abilities, and

1 The conditions under which a person with the ability to do A necessarily has a disposi-
tion to do A if they try can be thought of as a peculiar sort of normal conditions, namely,
conditions that were normally present and active on the occasions of their past successes.
These are conditions of the sort that I labeled with the capitalized term “Normal” in Mil-
likan (1984).“Normal” conditions for proper functioning of any mechanism are conditions
that obtained and were active on occasions of successful past performance leading to se-
lection of the mechanism.

also that real and even strong abilities can fail, the outer conditions nec-
essary to support them being absent and the absence remaining unde-
tected. So a look only at the mechanism that has or embodies abilities
“ at what its various dispositions are “ will not reveal what its abilities
are. What then does determine what its abilities are? I had originally
raised this question as a more general form of the question: What de-
termines, in the particular case, what particular substance one™s perhaps
stumbling, sketchy and inadequate conception is aiming at?
In the case of innate abilities, no matter what dispositions a mecha-
nism happens to have, what determines its abilities is what it was se-
lected for doing.2 In the case of learned abilities, what natural selection
selected for was the ability to learn in a certain way. It selected for
mechanisms that became tuned through interaction with the environ-
ment to do things of useful kinds. For an organism to know how to do
A as a result of learning is for it to possess an intact mechanism that is
biologically designed to be tuned to do things like A and that has been
tuned to do A as designed. That is, it became tuned in the same man-
ner, following the same principles, as its successful ancestors when they
were learning to do similar kinds of things.3 We humans possess at least
a number, possibly very numerous, different kinds of learning mecha-
nisms, including various mechanisms for trial-and-error learning, learn-
ing by association, by imitation, by figuring something out, and so forth.
Each of these mechanisms works in accordance with its own principles,
tailoring learned abilities in its own manner.To know how to do some-
thing as a result of learning, one must have a disposition to succeed in
doing it under the conditions one learned under that afforded previous
successes or under the conditions that helped to tune one to do it by
affording successes, and this learning or tuning must have been of the
kind the learning mechanisms involved were selected for doing. Other-
wise one does not know how, but does what one does only by accident.
Some learning mechanisms are extremely general in their possible
applications and some are extremely quick. Of very general application,
for example, are our abilities to learn something by figuring it out or

2 Things that a mechanism was selected for doing are its “proper functions,” as defined in
Millikan (1984, Chapters 1 and 2). Those concerned about the heavy but informal use of
the notions of “function” and “design” throughout this book are referred to these chapters,
where a tight definition is offered.
3 Abilities that have been learned correspond to mechanisms that have “adapted and de-
rived” proper functions. For details on adapted and derived proper functions, see Millikan
(1984, Chapter 2) and Millikan (in press b).

thinking it through, using materials gained from previous experience,
including knowledge of our prior abilities. So we often claim correctly
that we know how to do something that we have never tried to do, for
example, that we know how to get to Plainfield, or that we know how
to make a certain kind of repair. If conservative, we may say only that
we think we know how, but we are often right that we know. Similarly,
observing only once that you have a certain capacity can immediately
turn it into an ability. Anything that you find out you can effect imme-
diately becomes an ability. Having observed that stirring the red straw-
berries into the vanilla ice cream turns it pink, the child knows how to
make pink ice cream.
We need not think of abilities, then, as characteristically derived from
elaborate specialized training or tuning histories. They are, however, of-
ten derived from very numerous prior abilities that have been strung or
blended together. The idea that one might count the number of a per-
son™s abilities, or count the abilities that go into a certain activity, often
is not really coherent. Like patterns, however, or like patches of ground,
abilities can be clearly distinguished and designated even when they
have no clear criteria of individuation.


In the last chapter (Section 3.6), I claimed that it is not the purposes of
individuals but the subpersonal biological functions of their inborn
concept-tuning mechanisms that connect their substance concepts with
certain extensions. Similarly, the function of my eyes or my liver is not
determined by my intentions for their use.We must suppose that natural
selection has endowed us, specifically, with the ability to learn to iden-
tify substances, or, more likely, with a variety of abilities, each specific to
learning to identify members of a different domain of substances (see
Chapter 5) or, at the very least, the ability to acquire abilities of the lat-
ter kind through learning. Considering the centrality of the ability to
reidentify substances for any animal that acquires and applies either
practical or theoretical knowledge (Section 1.4), the likelihood that nat-
ural selection has tailored capacities very specifically to this purpose in
such animals is about as high as the likelihood that it has tailored their
capacities specifically for obtaining nourishment and mates.
Granted this, we can begin to answer the question how specific sub-
stances become assigned to specific substance conceptions “ how the
extensions of substance concepts are fixed. Chapter 5 will fill the pic-

ture out more with some speculations about how one™s abilities to re-
identify specific substances are learned or develop, and Chapter 14 will
complete the discussion of how conceptual abilities are focused on cor-
responding substances.
We encounter substances, better, we encounter natural signs of sub-
stances (see Appendix B) in perception (or in the speech of others “ see
Chapter 6).We bring to bear certain innate abilities, or we bring to bear
general skills acquired earlier in life (grasp of substance templates, grasp
of general methods applicable to specific substance domains), or we
proceed by trial-and-error learning in attempting inductions over en-
counters. In these ways we often manage to keep track, and we learn
better how to keep track, of when we are encountering the same sub-
stances again. Thus the information gathered about each substance is
brought together and brought to bear on future encounters with it.The
important point is that the way this all happens is no accident. That is
what it means to say that it is done using “abilities” and “skills” and by
“learning.” It is all done according to principles by which feats of the
same abstract sort were accomplished by our ancestors, thus accounting
for selection of the innate abilities and learning mechanisms responsible.
Now there will be times when a substance is misidentified, and for
every substance, of course, innumerable unrealized dispositions to
misidentify it under adverse conditions. There also may accumulate
considerable misinformation about certain substances. These mistakes
occur because conditions on which past successes in keeping track or
learning to keep track depended “ historically enabling conditions for
the cognitive mechanisms™ development and use “ are not currently
present. Unless the cognitive mechanisms are malformed or damaged,4
mistakes are always caused by the presence of abnormal conditions, not
necessarily in the sense of unusual conditions, but merely of conditions
other than those under which the utilized mechanisms were operating
on those specific occasions that historically afforded successes, hence ac-
counted for their selection or for their tuning through learning. If cur-
rent conditions were exactly the same as the historically enabling con-
ditions for these mechanisms, they would obviously succeed. This is
what distinguishes one™s dispositions to identify from one™s abilities to
identify. It separates acts of misidentification from correct identifications
and dispositions to misidentify from dispositions to identify correctly.

4 A definition of normal structure for devices with proper functions, including the functions
corresponding to abilities, is offered in Millikan (1984, Chapter 1).

In clear cases, then (I will soon discuss some that are less clear), a
substance concept has as its referent or extension a substance encoun-
tered at the very start of such a process of keeping track nonaccidentally
and that fits the general abilities to keep track that have been brought
to bear. It causally originates from encounters with the substance to
which it refers, and the general abilities that are brought to bear in at-
tempting reidentification determine the substance™s ontological cate-
gory. They determine, for example, whether it is the function of the
concept to reidentify a stuff or an individual or a kind of individual, and
if the latter, what sort of substance template is involved (see Section
14.1). The concept is a concept of A, rather than B, not because the
thinker will always succeed in reidentifying A, never confusing it with
B, but because A is what the thinker has been roughly keeping track of
and picking up information about not accidentally but in accordance
with abilities, skills, and know-hows.
Here is another way to approach the same matter, but that leaves it
more open whether to buy into the historical theory of the nature of
cognitive abilities I have proposed.Traditionally, it is supposed that what
a substance concept is of is whatever fits certain features or properties it
represents its extension as having. Its extension is what fits these prop-
erties, even though one may not always be able to recognize exactly
when something does. In this way the concept can have a definite ex-
tension, even though you sometimes make mistakes in recognizing that

<< . .

( 29)

. . >>

Copyright Design by: Sunlight webdesign