Professional Documents
Culture Documents
Everyday Expertise
Author(s): Hubert L. Dreyfus
Source: Proceedings and Addresses of the American Philosophical Association, Vol. 79, No. 2 (Nov
., 2005), pp. 47-65
Published by: American Philosophical Association
Stable URL: http://www.jstor.org/stable/30046213
Accessed: 01-06-2015 04:58 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
American Philosophical Association is collaborating with JSTOR to digitize, preserve and extend access to Proceedings and
Addresses of the American Philosophical Association.
http://www.jstor.org
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
THEMYTHOF THEMENTAL:
OVERCOMING
PHILOSOPHERS
How
CANPROFIT FROM
THEPHENOMENOLOGY
OF EVERYDAY
EXPERTISE
HUBERT
L. DREYFUS
UNIVERSITY
OF CALIFORNIA-BERKELEY
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
48
...If
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
49
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
50
ProceedingsandAddressesof theAPA,79:2
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
Presidential
Address- PacificDivision
51
rationalizations.
McDowell
seems to agreewhenhe notes,
onlybe retroactive
"I construe Aristotle's discussion of deliberation as aimed at the
reasons suggests that these reasons must have been implicit all along,
whereas, for Heidegger, the phronimos's actions are not in the space of
reasons at all.As Heideggersums it up:"In[phronesis]thereis accomplished
something like a pure perceiving,one thatno longer fallswithinthe domain
of logos."25
McDowell,however, would no doubt reply that there couldn't be any
such logos-free pure perception. So, in taking perception out of the space of
reasons, Heidegger can only be imposing on Aristotle a version of the Myth
of the Given. Heidegger could counter, however, that in assuming that all
intelligibility,even perception and skillful coping, must be, at least implicitly,
conceptual-in effect, that intuitions without concepts must be blind, and
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
52
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
53
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
54
andAddresses
oftheAPA,79:2
Proceedings
In the case of tacticalrules, however, the master may make moves that
are entirelyintuitiveand contraryto anypreconceivedplan.Insuch instances,
when asked why he did what he did, he may be at a loss to reconstructa
reasoned account of his actions because there is none. Indeed, as we have
seen, the phenomena suggest that an expert has long since abandoned
generalrulesas bikerssets aside theirtrainingwheels. Thus,when an expert
is forcedto give the reasons thatled to his action,his account will necessarily
be a retroactiverationalizationthatshows at best thatthe expertcan retrieve
frommemory the general principlesand tactical rules he once followed as
a competent performer.
Consequently,if one followed the reconstructedrules articulatedby an
expert, one would not exhibit expertise but mere competence, and that is
exactly what has happened. "ExpertSystems"based on the rules so-called
knowledge engineers elicited from experts were at best competent. 30 It
seems that, instead of using rules they no longer remembered, as the Al
researchers supposed, the experts were forced to remember rules they no
longerused. Indeed,as faras anyone could tell, the expertsweren'tfollowing
any rules at all.31
So it seems clear that rules needn't play any role in producing skilled
behavior.Thisis bad news forCognitivists,but McDowellis not a cognitivist.
He rightly rejects the idea that skilled behavior is actually caused by
unconscious rules. His view is much more subtle and plausible, namely,
that, thanks to socialization, experts conform to reasons that can be
retroactivelyreconstructed.Afterall,theremust be one structurein common
to situationsthatreliablysolicit one type of tacticalresponse and another to
those situationsthatreliablysolicit another.Itseems one ought to, at least in
principle,be able to articulatethis structurein terms of reasons. But all we
have a rightto conclude from our phenomenology of expertise is that there
must be some detectable invariantfeatures in what J. J. Gibson calls the
ambientoptic arrayand thathumanbeings and animalscan learnto respond
to them. Thesefeatures,althoughavailableto the perceptualsystem,needn't
be available to the mind.
We can understandthis inaccessibilityifwe consider the way simulated
neural networks can be programmed to produce reliable responses. For
example, nets have been programmed to distinguish sonar echoes from
mines fromthose fromrocks. Such nets are given no rules,nor are they told
what features of the signal are relevant.They are simply exposed to tens of
thousands of examples and their at-first-randomcorrect responses are
reinforced.
In one very limited sense, any successfully trainedmultilayerneuralnet
can be interpretedin terms of features-not everyday features but highly
abstract ones. But in the case of multilayered nets that are devoted to
implementingexpertise, these highlyabstractfeaturesare not interpretable
as features thata mind could possibly experience. Toconstruct a semantic
account of what a network that has learned certain discriminations has
learned, each node one level above the input nodes could, on the basis of
connections to it, be interpreted as detecting when one of a certain set of
input patterns is present. (Some of the patterns will be the ones used in
that
training and some will never have been used.) If the set of input pattemrns
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
55
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
56
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
57
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
58
options and their utilities and calculating which action is optimal. Such
responses, however, lose the situationalspecificityof expertise, and so are
inferiorto an expert's intuitiveresponse.
Fortunately,the expert usuallydoes not need to calculate. Ifhe has had
enough experience and stays involved,he will find himself responding in a
masterfulway before he has time to think.Justas Aristotle,Heidegger,and
Merleau-Pontysaw, such masteryrequiresa richperceptualrepertoire-the
ability to respond to subtle differences in the appearance of perhaps
hundredsof thousandsof situations-but it requiresno conceptualrepertoire
at all. This holds true for such refined skills as chess, jazz improvisation,
sports,martialarts,etc., butequallyforeverydayskillssuch as cookingdinner,
crossing a busy street, carryingon a conversation,or just gettingaround in
the world.46
V.Nonconceptual Coping and the Justification of Judgments
So farwe have seen that,ifwe understandconcepts as context-freeprinciples
or rulesthatcould be used to guide actions or at least make them intelligible,
a phenomenology of expert coping shows concepts to be absent or even to
get in the way of a masterfulresponse to the specific situation.Norcan such
thinkable content be reconstruction after the fact by deliberation. More
basically,if concepts must be linguistic,Gibson'saccount of our directpickup of affordances as high order invariantsin the optic array,and neural net
considerations as to how the brain might detect such invariants,suggest
thatexpertise does not requireconcepts. Indeed, the basis of expert coping
may well be the sort of features that the expert could not be aware of and
would not be able to think. In both cases, then, masterfulaction does not
seem to requireor even to allow placement in the space of reasons.
But these objections may seem to miss McDowell'sbasic point that,in
so far as perceptionjustifies ourjudgments, it must be conceptual all the
way out. As McDowellputs it:
When we tracejustificationsback, the last thingwe come to is stilla
thinkable content; not something more ultimate than that, a bare
pointingto a bit of the Given.47
Lewiswas, indeed, fundamentallymistakenin thinkingthatthe Givenhad to
be ineffable and indubitablein orderto groundjudgments,but, as we have
seen, the Givenneedn't be understoodas bare. Itcan be pure in the sense of
nonconceptual, and yet, like affordances, still have motivationalcontent.
Conceptualists like McDowell point out, however, that the idea that any
nonconceptual given, be it intuitive,practical, normative, skillful,or what
have you, could make a contributiontojustificationis unintelligible.This is
an important objection to the view Lewis defended but not to the
phenomenologists' claim that to perform its world-disclosing function
perception must be nonconceptual.
In so far as McDowell speaks of "ourunproblematic openness to the
world"48and of how "we find ourselves always already engaged with the
world"49-he seems to agree with the phenomenologists that perception
has a function more basic than justification. As Heidegger and MerleauPonty (as well as Wittgenstein) have argued, we can only relate to objects
and make judgments about them insofar as they show up on the background
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
Address- PacificDivision
Presidential
59
of the world-and the world is not a belief system but is opened to us only
throughour unthinkingand unthinkableengaged perception and coping.
McDowell,however,onlyseems to be in agreementwith these existential
phenomenologists.ForMcDowell,the worldcan onlybe the totalityof objects,
events, and states of affairs.Manyof these are, indeed, directlyperceivable,
and we do depend on beliefs based on perception to justifyour judgments
about them. But this sort of openness to the world is not the most basic
function of perception. We directlyperceive affordances and respond to
them without beliefs and justifications being involved. Moreover,these
affordancesare interrelated,and it is our familiaritywith the whole context
of affordances that gives us our abilityto orient ourselves and find our way
about. As Heideggerputs it:
[W]hat is firstof all "given"...isthe "forwriting,"the "forgoing in
and out,"..."forsitting."Thatis, writing,going-in-and-out,sitting,and
the like are what we are a prioriinvolvedwith.Whatwe know when
we "know our way around."50
Heidegger might seem to be a conceptualist himself since he continues:
"Mybeing in the world is nothing other than this already-operating-withunderstanding."51But when he introduces the term "understanding,"
Heidegger explains (with a little help from the translator)that he means a
kind of know-how:
InGermanwe say thatsomeone can vorstehensomething-literally,
stand in frontof or ahead of it, that is, stand at its head, administer,
manage, preside over it. Thisis equivalentto sayingthathe versteht
sich darauf,understandsin the sense of being skilled or expert at it,
has the know-how of it. 52
Merleau-Pontyalso appeals to a nonintellectualkind of understanding:
We understandthe thingas we understanda new kind of behavior,
not, that is, throughany intellectual operation of subsumption, but
by taking up on our own account the mode of existence that the
observable signs adumbratebefore us.53
And, of course, such givens are not bare givens. As Heideggerinsists:
Everyact of havingsomething in frontof oneself and perceivingit is,
in and of itself,a "having"something as something. ...However, this
as-structureis not necessarilyrelatedto predication. Indealingwith
something, Ido not performany thematicalpredicativeassertions.54
VI.Where Concepts Come In
If nonconceptual perception and coping is necessary for world disclosing,
and there is no way nonconceptual perceiving could ground judgments,
then we must ask how the nonconceptual given is converted into a given
with conceptual content so that perception can do its justificatoryjob.
McDowell seems to rely on the Mythof the Mental, the idea that pure
perception is impossible so perception and coping must somehow always
already be conceptual, to avoid facing this question. In his book Body and
World,Samuel Todes, however, sees the question of how conceptual content
arises from nonconceptual content as the central puzzle bequeathed to
philosophers by Kant.55He, therefore, proposes to work out a detailed
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
60
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
61
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
ProceedingsandAddressesof theAPA,79:2
62
Endnotes
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Ibid., 155.
Ibid.,134.
15.
Ibid., 111.
16.
17.
18.
19.
20.
21.
22.
Ibid.(Myitalics.)
Ibid.,84.
Ibid.
Ibid.,79.
Ibid.,85.
Ibid.,84.
MartinHeidegger,Plato's Sophist, trans. R. Rojcewicz &A. Schuwer
(Indiana University Press, 1997), 101. (My italics.)
23.
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
63
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
ProceedingsandAddressesof theAPA,79:2
64
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
CharlesTaylor,"Merleau-Ponty
and the EpistemologicalPicture,"Taylor
Carman & MarkB.N. Hansen, eds., The CambridgeCompanion to
Merleau-Ponty,(Cambridge:CambridgeUniversityPress, 2005), 34.
This specificity of coping activity corresponds to the situational
specificity of perception that Sean Kelly uses to demonstrate that
perception is nonconceptual. See "The Non-ConceptualContent of
PerceptualExperience:SituationDependence and Fineness of Grain,"
Philosophy and Phenomenological Research (with response by
ChristopherPeacocke), 62:3 (May,2001): 601-08.
Thereare cases, however, when the affordanceis relativeto what the
mind is capable of. ToKasparov,but not to a merely competent player,
a specific situationon the chessboard affordscheckmate.
Since such a response is clearly possible for animals, so it can't be
conceptual in McDowell's sense of the term since, according to
McDowell,animals don't have concepts.
CharlesTaylor,"Merleau-Ponty
and the EpistemologicalPicture,"Taylor
Carman & MarkB. N. Hansen, eds. The CambridgeCompanion to
Merleau-Ponty(CambridgeUniversityPress, 2005), 34.
MarkA. Wrathall,"Motives,Reasons, and Causes,"in The Cambridge
Companionto Merleau-Ponty,T. Carmanand M. B. N. Hansen, eds.,
(Cambridge:CambridgeUniversityPress, 2005), 118.
Phenomenologyof Perception,238.
Ibid.,250.
in The Cambridge
Sean D. Kelly,"Seeing Things in Merleau-Ponty,"
M.
B.
N. Hansen, eds.,
T.
and
Carman
to
Companion Merleau-Ponty,
87.
2005),
Press,
(Cambridge:CambridgeUniversity
Ibid.,97.
Fordetails, see HubertL. Dreyfus,"The Primacyof Phenomenology
over Logical Analysis,"Philosophical Topics,MarkA. Wrathalland
HubertL.Dreyfus,eds., Vol.27, No. 2, Fall1999,(2001).
Merleau-Pontycalls the sort of intentionalitydefinitiveof the space of
motivations,motor intentionality.Sean Kellyworks out the special
features of this kind of intentionality,as opposed to the conceptual
kind, in his paper, "The Logic of MotorIntentionality."The paper is
availableon his website at/-skelly/Research/LogicMI.pdf.
Formore details see, HubertL. Dreyfusand StuartE. Dreyfus,Mind
over Machine:ThePower of Human Intuitionand Expertisein the Era
of the Computer,(New York:Free Press, (1988) 36-41.
Sellarsiansmightcounterthatall coping,even the immediateresponse
to the perceived situationin chess, is, nonetheless, conceptual, in that
the master's play is crucially informed by an understanding of what it
is to "win,"to "capture"a piece, legal and illegal moves, etc. Someone
who did not understand these concepts, including having the ability
to use them properly in other contexts, would not be appropriately
regarded as playing chess. But it is not obvious that the discrimination
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions
PresidentialAddress- PacificDivision
50.
51.
52.
53.
65
55.
56.
57.
58.
54.
59.
This content downloaded from 146.50.98.28 on Mon, 01 Jun 2015 04:58:31 UTC
All use subject to JSTOR Terms and Conditions