OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
5
Knowing That I A m Thinking
Alex Byrne
Soc .... I speak of what I scarcely understand; but the soul when thinking
appears to me to be just talking —asking questions of herself and answering
them, affi rming and denying. And when she has arrived at a decision, either
gradually or by a sudden impulse, and has at last agreed, and does not
doubt, this is called her opinion. I say, then, that to form an opinion is to
speak, and opinion is a word spoken, —I mean, to oneself and in silence, not
aloud or to another: What think you? Theaet . I agree.Plato, Theaetetus
I. Introduction
1We
often know that we are thinkin g, and what we are thinking abo ut. Here ‘
thinkin g’ is not supposed to be an umbrella ter m for cognit ion in gen eral, but
shoul d b e taken in roughly the sense of ‘ a penn y for your thou ghts ’ : ment al
acti vities like pon dering, ruminating, wonde ring, musing , and daydrea ming all
count as thi nking. In the intended sense of ‘ thinking ’ , t h inking is not just
propos itional : i n additio n t o thinking that p , there is thinking of (or abo ut) x .
Beli ef is neces sary but not suf fi cient for thinkin g that p : thinkin g that p
entails bel ieving that p , but not conversely .
Cons ider Kylie:
On summer afternoons in Canberra, the baking sun reflects off Lake Burley Grif fin, and
the water shimmers. Up behind the university, in the botanical gardens, a cascading
stream of water helps to maintain the humidity of the rainforest gully. These are just a
couple of Kylie ’ s thoughts on the subject of water, her water thoughts. Amongst Kylie ’s
many other thoughts that involve the concept of water are these: that there is water in the
lake, that trees die without water, that
Thanks to audiences at Umeå University, the University of Melbourne, Monash University, UMass
Amherst, the Institute of Philosophy ( London), and a reading group at the RSSS, ANU. I a m
especially indebted to Martin Davies for discussion.
1In the simple present and past tenses (e.g.), ‘think that p ’ is near-enough synonymous
with ‘believe that p’,asin‘I think/thought that the pub is/was open’.
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
10 6
ALEX BYRN E
water is a liquid and, of course, that water is wet. When Kylie thinks consciously, in a way
that occupies her attention, she is able to know what it is that she is thinking. This is true
for thoughts about water, as for any other thoughts. ( Davies 2000: 384 –5)
As the watery clue sugges ts, Davies use s Kylie to discuss McK insey ’ s puz
zle, the alleged incompa tibility of externa lism and the appa rent fact that w e
can obt ain self-kn owledge from the armchair. But th at is not this paper ’ s
topic. Suppo se — if only for the sake of the argumen t— that McK insey ’ s
puzzle can be solved, and that exter nalism and armcha ir self-knowl edge are
com patible. How does Kylie kno w that she is thinkin g about water “ from the
armcha ir ”? Just by sitting comfo rtably and “ introspe cting ”, Kylie can discover
that she is thinkin g about wate r, and that she is thinkin g that trees die withou t
water. B u t what is “intro spectin g” , exactly? And why is it — as is com monly
assumed — a partic ularly effe ctive way of fi nding out what one is thinkin g?
II. Privileged and peculiar access
In his cla ssic paper “ Anti-Indiv iduali sm and Priv ileged Acces s”, McKi nsey
star ts by saying that
[i]t has been a philosophical commonplace, at least since Descartes, to hold that each of
us can know the existence and content of his own mental states in a privileged way that is
available to no one else . (McKinsey 1991: 9, emphasis added)
This “philo sophica l com monpla ce ” consist s o f two distinc t cla ims, which
McK insey does not clearly distingu ish. The first is that we have privil eged
access to the ex istence and conten t o f our mental states. App roxima tely put:
beliefs about one ’ s men tal states acquir ed th rough the usual route are more
likely to amou nt to knowledge than beliefs about others ’ mental states. Kylie ’ s
belief that she is thinking about water is more lik ely to amou nt to kno wledge
than our belief that she is thinking about water. We migh t judge that Kylie is
thinking abo ut w ater bec ause we see her gazin g pensivel y a t Lake Burley G
rif fi n, or bec ause we overhe ar someon e w e take to be Kylie mutteri ng “W
ater is a liquid ” . W e can easily go wron g i n a n ordinary situ ation of this kind.
Perhaps when gazin g a t the lake Kylie is not thinking of water, but of the
original Canber ra Plann er, Walter Burley Grif fin himsel f; or perha ps Kylie is
not the person muttering .
Kylie herself cann ot go wron g a s easil y, or so it is natu ral to suppose. If we
offer Kylie a penny for her thou ghts, and she sincerel y replies “ I was thinkin
g abo ut water ” ,it would be bizarre for us to dou bt Kylie ’ s claim . And this is
not bec ause we have a blank et policy of ta king avowa ls of mental states to
be beyond disp ute — we plain ly don ’ t. Setting fact ive and object -entailing
states (e.g. regretti ng drinking five pints , and seeing an echi dna ) t o one
side , w e make subtle distinc tions be tween our epis temic acce ss to mental
states in ordinary life. It would not be at all unexcep tional, for inst ance, for us
to wond er whether Kylie is mistake n abo ut what she wants. Perha ps Kylie
falsely believe s that she w ants to join the readi ng group on Bein g and Time
because she has a
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
KNOW ING THAT I AM THINKING
10 7
distorted view of the sort of person she is. She isn ’ t really an excessiv ely
sophist icated intellectual with a taste for Teut onic obscurity , b ut a more
straight forward pl ainspoken type who prefers A Material ist Theor y o f the
Mind .
That w e have priv ileged access to our though ts has some ex perimen tal
suppor t. Subjects have bee n inst ructed to “thi nk aloud ” while perform ing a
variety of cognit ive tasks “in their hea ds ” : multip lying numbe rs, solvin g the
tower of Hanoi puz zle, counting the windo ws in their livin g room s, remem
bering and recalling a mat rix of digits, and so on (Erics son and Simon 1993).
The subjects ’ self-rep orts can then be checked again st theor etical models of
proble m solving, the su bjects ’ res ponse lat encies, and so on. Summ arizing,
N ichols and Stich write that:
both commonsense and experimental studies confi rm that people can sit quietly,
exhibiting next to no overt behaviour, and give detailed, accurate self-reports about their
mental states. (Nichols and Stich 2003: 158)
The claim of privil eged access to our own thought s i s a watered - down
version of what is sometime s cal led infallible , incorr igible ,or indubita ble
access:
In Necessa rily, if S believe s she is in M, then she is in M/ knows that she is
in M What is som etimes called self-intim ating acce ss is a nea r-conve rse of
In:
S-I Necessa rily, if S i s i n M , she know s/is in a posit ion to kno w that she is
in M2
Privileged access shou ld not be confused w ith S-I and its watered- down va
riants, which have little to recomm end them , a t l e ast unless highly quali fi ed
and restricted. Post-Freud , the idea that there are subte rranean mental
currents not readily accessible to the subjec t i s unexcep tionable, even if Freud
’ s own account of them is not.
Priv ileged access is the first part of McK insey ’ s “ philoso phical comm
onplace ” . The second part is that w e have peculiar acce ss to our ment al
states: as McK insey says, we know abo ut them “ in a . . . way that is available
to no one else ” . Kylie kno ws th at she is thinking that trees die withou t w ater
by “ introspe ction ”, but whatev er intro spectio n is, she cann ot empl oy it to
discover that someon e else is thinkin g that tre es die withou t water. In
McKinsey ’ s prefe rred termi nology, Kylie can know “a priori ” that she is
thinking that tre es die without water. (Th is termino logy is hardly apt, sinc e one
leadin g theory of self -knowledg e classi fies it as a variety of perceptual know
ledge: see } III.)
It is importan t t o distin guish privil eged and peculiar access because they
can come apart in both d irections. Consider Ryle , who holds that we have acce
ss to our own minds in the sa me w a y t h a t w e have ac cess to others ’ mind
s— by observ ing beha vior — and th us denies that w e have peculiar acce ss.
“The sorts of things that I can find out about myself are the sam e a s the sorts
of th ings that I can fi nd out abo ut ot her people,
2For
the belief-versions of both In and S-I see Armstrong (1968: 101); as Armstrong notes, the
term ‘selfintimation’ is due to Ryle. In the terminology of Williamson (2000: 95), the
knowledge-version of S-I is the thesis that the condition that one is in M i s “ luminous” .
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
10 8
ALEX BYRN E
the metho ds of finding them out are much the sam e” (Ryl e 1949: 149,
emphasis added). Yet he thinks that we have privileged access to our ment al
states (at least som etimes), because w e (sometim es) have better beha vioral
evide nce about ourse lves — greater “su pplies of the requi site data ” (149) .
Ryle ’ s position shows why privileged access need not be peculiar. For an
illus tration of the convers e, suppose that one often mistak es be eches for elms
by sight. Often one ’ s belief that one sees an elm is mistaken, or at any rate not
knowledge . One ’ s a c cess to one ’ s ment al state of seeing an elm is noth ing
to writ e home abo ut: mu ch kno wledge of one ’ s enviro ment and of others ’
mental states is consider ably easier to attta in. Yet one has peculiar ac cess to
the fact that one sees an elm: one does not need the usual thi rd person
evidence — th at one ’ s eyes are open and poin ting tow ards an elm, one has
an unobst ructed view of the elm, the ligh ting is good, and so on.
It shou ld be empha sized that the alle ged con fl ict betw een self-kn owledge
and externa lism has its source in peculiar acce ss, not privil eged acce ss.
Unles s p rivileged acce ss is taken to be something like infal lible access, there
is not even the appearanc e o f a con flict with external ism. Peculiar acce ss to
the existen ce of water (f or example) is the issu e. McK insey ’ s arg ument
gives rise, as Dav ies sa ys, to the puzzle of armchair know ledge.
3and
III. Inner-sense
Accor ding to Ar mstrong:
Kant suggested the correct way of thinking about introspection when he spoke of the
awareness of our own mental states as the operation of ‘ inner sense ’. H e took sense
perception as the model of introspection. By sense-perception we become aware of
current physical happenings in our environment and our body. By inner sense we
become aware of current happenings in our own mind. (Armstrong 1968: 95; see also
Armstrong 1981, Lycan 1987: ch. 6, 1996: ch. 2, Nichols and Stich 2003: 160 –4)
By using her outer eye, Kylie kno ws that there is a lot of water in Lake Burley
Grif fi n; accordin g t o the inner-sen se theor y, by using her inner eye, Kylie
knows that she is thinking of water.
The inn er-sense theor y does offer a nice ex planation of peculiar acce ss: for
obv ious archi tectural reaso ns, the (presu mably neu ral) mech anism of
inner-sen se is only sens itive to the subjec t’ s own men tal states . I n exactly
the sam e style, our faculty of
3The
claim of “Privileged Access” , i n Ryle’s sense, is this: “ (1) . . . a mind cannot help being
constantly aware of all the supposed occupants of its private stage, and (2) . . . i t can also
deliberately scrutinize by a species of non-sensuous perception at least some of its own states
and operations” (Ryle 1949: 158). In the contemporary literature, ‘ privileged access ’ is often
used approximately for what (in the text) is described as privileged and peculiar access (see e.g.
Alston 1971; Moran 2001: 9 –10). Ryle’s characterization of “ nonsensuous perception” denies
that there are “ any counterparts to deafness, astigmatism, . . .” (1949: 157); this means that
Ryle’s Privileged Access implies but is not implied by the conjunction of privileged and peculiar
access (in the sense of the text).
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
KNOW ING THAT I AM THINKING
10 9
propriocept ion explain s the “pec uliar acce ss” we have to the p osition of our
own limbs (see Ar mstrong 1968: 307).
But it does not explain privil eged access. In fact, it leaves it something of a
myste ry. Why is inner-sen se less prone to erro r than the outer sens es? Why
are (as w e com monly suppose) our opinion s about our own desires more
error-pr one than those about our thought s? Are desi res located in so me dark
and obscure place in the brain?
4And finally , i f there is an inn er-sense , then presuma bly it might be damag ed
or absent, w hile sparin g the rest of a normal subject ’ s cognitive capacities . I
n other words, the conditio n Shoe maker (1988 ) cal ls “self-blin dness ” could
occ ur, if the inner-sen se theory is correct . Perhaps self -blindness is (met
aphysically ) poss ible, but no actual psycholog ical conditio n seems to come
close .
IV. Economy and extravagance
On one (min ority) view of our kno wledge of language , i t requi res noth ing
more than a general purpos e learning mechan ism. On the alternat ive
Chomskean pictu re, it requires a dedic ated faculty, a “ language organ ” . The
minor ity view is an econ omical account of our linguistic knowle dge: no specia l
purpose epis temic capacities are required. The Choms kean view, on the other
hand, is extrava gant : given th e mea ger input — the “povert y o f the stimulu
s”— the genera l purpose mechani sm is supposed incapable of genera ting the
requi red torrent ial output .
5A similar distin ction can be drawn for self-kn owledge. Let us say that a theor y
o f self-knowledg e i s economica l just in case it explain s self-kn owledge solely
in terms of epistemic capacities and abilities that are needed for kno wledge of
other subject matters; otherw ise it is extr avagant . 6Rylean ism is econo mical:
th e capacities for self knowledge are precisel y the cap acities for knowl edge of
the minds of others .The theory defen ded in Shoemake r (1994) is also e
conomical : here th e relev ant capacit ies are “ norm al intellige nce, ration ality,
and conceptua l capac ity” (236) . The inner-sen se
4According
to Nichols and Stich, “certain kinds of schizophrenia might involve damage to the
Monitoring Mechanism [their term for an Armstrongian faculty of inner-sense] that does not affect
other components of the mind-reading system” (2003: 190). Their main examples concern “ passivity
symptoms ” , such as delusions of control (in which patients attribute their own actions to other
agents). Nichols and Stich also cite Hurlburt ’s work in “ experience sampling” in support of the view
that some schizophrenic patients have difficulty reporting their imagery and thoughts when
symptomatic (190–1; cf. Hurlburt 1990: esp. 204–8, 224–5, 238–9). However, none of these cases
remotely approximate Shoemakerian self-blindness. Schizophrenic self-blindness is limited, and
anyway schizophrenia patients suffer from other cognitive impairments.
5A qualification: knowledge of the “ other subject matters” should not itself require mental evidence.
On one traditional view (Locke’s, for example), all empirical knowledge is founded on mental
evidence about perceptual appearances; on that view (assumed false here), the correct theory of
self-knowledge is extravagant, not economical.
6Carruthers (2009) is a recent defense of a sophisticated view of this type. Like the account to be
defended later, inner speech plays an important role, but Carruthers denies that his account is an
extension of the Evans-style transparency procedure mentioned a few paragraphs below.
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
11 0
A L EX BY RN E
theor y, on the other hand, is extravagant : the organs of outer perception , our
gen eral capacity for rationalit y, and so forth, do not account for all our
self-knowle dge — for that, an additio nal mechanism, an “ inner eye ”, i s
needed .
7The inner-sen se theor y has no explan ation of priv ileged access. Ryleanism
implausibly denies that we have pec uliar access. On Shoe maker ’ s theor y,
“there is a conceptual, constitut ive connec tion betw een the existen ce of certa
in sorts of men tal entities and their intro spectiv e acce ssibility ” (Shoema ker
1994: 225), but Shoemaker ’ s case for this very strong cla im is open to
question .
Howev er, in the case of belief there is an econo mical theor y that is (a) Shoe
makeria n in spirit, and that ( b ) (arguab ly) expl ains both privil eged and pec
uliar acce ss.
Consider the follo wing well-know n pas sage fro m Evan s:
making a self-ascription of belief, one ’ s eyes are, so to speak, or occasionally
literally, directed outward — upon the world. If someone asks me “Do you think there is
going to be a third world war?,” I must attend, in answering him, to precisely the same
outward phenomena as I would attend to if I were answering the question “Will there be a
third world war? ”. (Evans 1982: 225)
8[I]n
One apparently finds out that one belie ves that it ’ s raining by dete rmining
whether it’ s raining: kno wledge that one has this belief, insofar as it rests on
perceptua l evidenc e a t all, res ts on perc eptual evide nce about the weat her,
not on perc eptual evide nce of one ’ s behavio r o r anyt hing ment al. That is,
one reasons from the evide nce that it’ s rain ing, to the concl usion that one
believe s that it’ s raining.
Of course, this is not to say that such reaso ning is good reasonin g. In fact,
prima facie it is terrible reaso ning. Ignor ing this dif ficulty, the importan t point is
that the particula r capacity to reason from the evide nce that p to the conclusio
n that one believe s that p does not involv e any specia l mech anism of inn
er-sense — given a stock of ment al concepts , i t rides on the coatt ails of the
genera l capacit y for reasonin g.
The next section brie fl y argues for two claim s. First, that this Evans-s tyle
“transpa rency ” procedure is indeed good reasonin g, and so can yield self
-knowledg e. Se cond, that it can als o expl ain privileged and peculiar access.
Th e res t o f the paper tries to extend th is accoun t t o t h e case of though t.
V. An epistemic rule for
belief
9
Consider a very simple examp le of reaso ning: Kylie hears the kettle whist le,
and conclud es that the water is boilin g. By hea ring that the kettle is w
histling , Kylie know s that th e kettle is whistling ; b y reasonin g, she knows
that the water is boilin g.
See Byrne (2005: 82
–92).
See also Byrne (2005:
93–5).
7
8
9
See also Dretske (1994, 1995) and Gordon (1996). A similar view can be found in Husserl; see
Thomasson (2003) for an interesting discussion.
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
K
N
O W
T
H
I N
I N
K
I N
G T
G
H
A
T
I A
M
1 1 1
We can usefull y think of Kylie acquiring kno wledge of the water ’ s boilin g b
y following a rec ipe or rule. If we say that an epis temic ru le is a conditio nal of
the follo wing form:
R If conditio ns C obtain , believe that p 10
then the epistemic rule that Kylie follows is: Kettle If the kettle
whis tles, believe that the w ater is boilin g
What does it mean to say that Kylie follow s this rule on a particula r occasion ?
The semistipulative answer is this: Kylie believes that the water is boilin g
because she recogniz es that the kettle is whistling . The ‘ because ’ is intend ed
to mark the kind of reason-giv ing causal co nnection that is often discussed und
er the rubr ic of ‘ the basing relation ’ . Kylie might rec ognize that the kettle is
whist ling, and believe that the water is boiling for some other reason: in this
case , she does not for m her belief b ecause she recogniz es that the kettle is
whistling .
So S follows the rule R (‘ If conditio ns C obtain, believe that p ’ ) o n a partic
ular occasion iff on that occ asion:
(i) S believe s that p because she recognizes that conditi ons C
obtain which implies:
(ii) S recog nizes (hence know s) that conditio ns C
obtain (iii) conditio ns C obtain (iv) S believes that p
Following Kettle tends to produce know ledge abo ut the conditio n o f the w ater
(or so we may suppose ), and hen ce it is a good rule. Following bad rules tends
to produ ce false and unj usti fied beliefs, for example:
Astrology If the daily hor oscope predict s that p , believe that p Astrology is
als o a n example of a schema tic rule. One follows a sche matic rule just in
case one follo ws a rule that is an inst ance of the schema tic rule; a schema tic
rule is good to the extent that its inst ances are.
11If the anteceden t conditio ns C o f a n epis temic rule R are not speci fied in
terms of the rule follo wer ’ s mental states, R i s neutral . A schemat ic rule is
neu tral jus t i n case some of its instances are. Thus, the claim that S can follow
a neu tral rule does not presuppos e that S has the capacity for self-kn owledg
e. Kettle and Astrology are neutral rules; ‘ If the type looks blurr y, bel ieve that
you need glasses ’ is not.
should be emphasized that the linguistic formulation of the rule only plays a heuristic role —all
the work is done by the account of following a rule (see immediately below).
11‘You’ refers to the rule-follower; tenses are to be interpreted so that the time the rule is followed
10It
counts as the present.
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
11 2
A L EX BY RN E
Self-knowled ge is our topic , not skep ticism: know ledge of one ’ s enviro
nment (includin g other s’ actions and ment al states) and reaso ning (speci fi
cally, rule- following of the kind just sketc hed) can be taken for grante d. So, in
the present cont ext, it is not in dispute that we follo w neutr al rules, includin g
neu tral rules with mentalist ic fillin gs for ‘ p ’ , lik e ‘ If S b l ushes, believe that S
i s embarrass ed ’ ; neither is it in disp ute that some neutr al rules are good
rules.
Evans ’ s observ ation in the passage quoted in the previou s sect ion ma y b
e recast using the appa ratus of epistemic rules as follo ws. Knowledg e o f one ’
s beliefs may be obtained by following the neu tral schemat ic rule:
Be l If p , believe that you believe that p Since the anteceden t o f Be l express
es the conten t o f the mental state that the rulefollow er ends up believi ng she is in, Be l can be called a transparent rule. One
is only in a position to follow Be l by believi ng that one believe s that p when
one has recogniz ed that p . A n d rec ognizing th at p is (inter alia) co ming to
believe th at p . Be l is self-v erifying in th is sens e: if it is follow ed, the result ing
seco nd-orde r belief is true. Com pare a third- person version of Be l:
Be l-3 If p , bel ieve th at Kylie believes that p Be l-3 is of course not self-v
erifying: the res ult of following it ma y b e (indeed, is very
likely to be) a false belief abo ut Kylie ’ s beliefs. Say that S tries to follo w rule R
iff S believe s that p bec ause S belie ves that conditions
C obt ain. That S follo ws R entails that she tries to follo w R , but not conv
ersely. If one tries to follow Kettle but does not succee d, then one will not kno w
that th e water is boilin g; if one ’ s bel ief abo ut the w ater is true, that is just an
accident.
Sometim es one will not succee d i n following Be l; inst ead one will merely try
to follow it. That is: one believe s but does not know that p , and thereby co
ncludes that one bel ieves that p . Her e one ’ s seco nd order belief that one
believe s that p will be true . Si nce trying to foll ow Bel cannot lead to error, Bel
ca n b e calle d st ro ng ly se lf -verifying.12
13Hence ( we ma y f ai rly s up po se ) t hi s situation will be commonplace: tr
yi ng to follow Bel, one i nv es ti ga te s w he th er p, mistakenly co nc lu de s t
ha t p, a nd thereby c om es to know th at one b elieve s t ha t p. I n these
cases, one w ill k no w t hat o ne beli eves that p on the b as is of no evidence
at all.
An example of a rule that is self-verifying but not
strongly so is: Know If p, believe that you know that p
12
13
One might have the following worry at this point. Haven’t w e learnt from Gettier cases (Gettier
1963) that knowledge cannot be based on reasoning through a false step? So if Kylie infers that
she believes that it’s raining from the false premise that it’s raining, how can she know that she
believes it’s raining? The short answer is that a better diagnosis of the Gettier examples is that
safety (in the sense of Sosa 1999 and Williamson 2000: ch. 5) is a necessary condition for
knowledge, not that no reasoning through false steps is a necessary condition for knowledge. And
beliefs produced by trying to follow Be l will often be safe (see Byrne 2005:
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
KNOW ING THAT I AM THINKING
11 3
Given that we follow rules like Kettle , i t should not be in dispute that w e can
follo w Be l. G iven the plausibility of Evan s’ s observatio n abo ut the procedure
we actually follow, it should not be in dispute that w e do follo w Be l . Be l offers
an obviou s explanation of pec uliar acce ss: as just noted, Be l-3 is a very bad
rule indeed . Privileged access is ex plained — at least on the face of it— bec
ause Be l i s strong ly self -verifyin g.
The accoun t just sketche d i s not a version of the inner-sen se theor y. It is
econ omica l, like beha viorism . Taking the capacity to follo w g o o d neutr al
rules for granted, knowledge of what one believe s com es alon g more- or-less
for free. Since this capacit y belongs to the depar tment of reaso ning, not
perceiving , Shoe maker ’ s idea that the source of self -knowledg e can be
traced to “ration ality ” is vindic ated, albeit not via his preferred route .
VI. Silent soliloquy
Much of ou r t hinkin g “ occurs in inner speech”
,orinwhatR ylecalls an “ in ternal m ono lo gue or s ilen t s oliloqu y
” (R yle 1949: 28). In s om e sense — yet to b e e x pl ai ne d — one
hears oneself thinking, with “ the m ind ’ sear” .Anatural idea , t
hen, is that Ky lie k no ws tha t sh e is t hinking about water, a nd
t h at s h e is th inking that trees d ie with out w ater, becau s e s
he eav esdro ps on h erself u ttering, in the s ilent C artesian
theater, ‘ T rees die w ithout w ater ’ .Atan yr ate,thatm ight be one
way K ylie ca n k now tha t this is wha t she is t h in k ing. T he
rest of t his paper p ursues this suggest ion. The p roject of
levera gin g this in to an eco n om ical accou nt of s elf -knowledg
e m ight in itially seem ho peless — reassurance w ill b e offered
several sections later.
The paradoxic al-seemin g claims of the prev ious paragr aph need to be clea
red up first. When one engag es in silen t sol iloquy, there are no sounds in one ’
s head — that ’ s why the soliloqu y i s silent. But, if there really is inner speech ,
there are sequences of phonemes, and so sounds. Hence there is no such thing
as inner spee ch.
Instead, speech i s (phonologically) represe n te d
.W henKylie“ hears” her “ internal m o nolog ue ” , she
is in a q uasi-perceptual state that represents an
utterance of th e s en t e n c e ‘ Trees di e w ithout
water ’ .(Thisclaimwillreceivesomedefense la ter , in
} IX.)
There is noth ing of w hich Kylie is aware w hen she “hea rs ” her inner
speech. A fortiori, she is not aware of any episode of th inking abo ut w ater.
Still, it seems plausible that this appea rance of an inner mono logue enabl es
Kylie to kno w that she is thinkin g about wate r. Why shou ld this be so?
( Thinking about x i s the subject of the next five section s; think ing that p will
be treated in } XII.)
96–8). A similar point is crucial for the following account of the epistemology of thought, which
always involves reasoning through a false step. (Thanks here to Benj Hellie.)
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
11 4
A L EX BY RN E
VII. Outer and inner speech
Ryle noted that an imp ortant source of informatio n about other s i s provided by
their “unst udied tal k ”, utte rances that are “ sponta neous, frank, and unp
repared ” (Ryle 1949: 173). Chatt ing with Kylie ov er a few beers is the best way
of disco vering w hat she believe s, w ants, and inte nds. “Stu died ” tal k, on the
other hand, is not so revealing. If Kylie is a polit ician defen ding the federal
gover nment ’ s policies on water, she migh t assert that the w ater shortage will
soon be over with out b elieving it w ill be.
Howev er, in the umbrella sens e o f “thinkin g about x ” with which we are
concern ed, both unst udied and studied tal k provide excellent evidenc e about
the utterer ’ s thou ghts. Even if the Hon. Kylie, MP , doesn ’ t believe that the
wate r shorta ge will soon be over, she was presuma bly thinking about water.
Outer speech on suchand-such topic is produced by ment al activity about that
same topic.
If someon e outwardly utters ‘ The water shortage will soon be over ’ then
(usuall y) she says som ething, namely that the water shortage w ill soon be
over. She says — and so thinks — som ething about water. Does the sam e
point hold for inn er speech ? A n af fi rmativ e answer is not trivial , because the
producti on of inn er spee ch, unlike the produc tion of outer spee ch, migh t
have noth ing to do with the seman tics of the w ords. Perhaps an inward utte
rance of ‘ Th e water shorta ge will soon be over ’ is produced in a simila r
semantics-ins ensitive manner as the inwa rd utteran ce of ‘ Dum diddley ’ ,or
other meaning less string.
14But this possi bility can be dismiss ed by notin g that outer speech and inner
speech often perform the sam e func tion, moreo ver one for which the sema
ntics of th e words is crucial. One may cajole or encou rage one self out loud;
one may also do so silent ly. Inner spee ch and outer speech may be seam
lessly interleaved in a conversatio n. One may recite a shoppi ng list out loud to
pres erve it in w orking memory; silent rec itation will do just as well.
By “hea ring ” herself inwa rdly utter ‘ Tr ees die withou t water ’ , then, Kylie
can know that she is thinkin g abo ut water. As Ryle puts it:
We eavesdrop on our own voiced utterances and our own silent monologues. In noticing
these we are preparing ourselves to do something new, namely to describe the frames of
mind which these utterances disclose. ( Ryle 1949: 176)
When Kylie utters (o ut loud) ‘ Trees die w ithout water ’ , there is somethi ng
she is aware of — the utteranc e o f that sent ence. She is not thereby aware of
a mental episode of thinking abo ut water, altho ugh she can thereby be aware
that she is thinking about water. When Kyle utters in “silent soliloquy”, ‘ Trees
die without water’, there is nothing she is aware of, a fortiori she is not aware of
a m ental episode o f thinking about water.
14On
one standard model of working memory, this involves the so-called “ inner ear” , a short-term
memory store for phonological information which is part of the “ phonological/articulatory loop”
(Baddeley 1986: ch. 5).
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
K
N
O W
T
H
I N
I N
K
I N
G T
G
H
A
T
I A
M
1 1 5
And even if there were an “ inner utteran ce ”, occurin g i n som e etherea l
med ium, it would still be wron g t o say that Kylie is aware of any ment al epis
ode. Th e outer utteranc e i s not itself an epis ode of thinkin g, but som ething
produ ced by such an episode; lik ewise, if there were (per impos sibile) an
inne r utteran ce it w ouldn ’ tbean episode of thinking either.
Accor ding to Carruthers :
I n t r os pe ct io n i n f o r ms u s , i n f a c t , t h at ma n y o f ou r th ou g h t s a r e
e x p r e s se d e n t i r el y i n n a t u ra l l an g ua g e . .. I c an r ep o r t t ha t m os t o f m y
t h o ug h ts o cc u r i n t h e f o rm o f im a ge d c on ve r sa t ion s .. . ( C a r r u t h e r s 1
996: 50–1)
But surel y intro spection informs us of no such thing. There is no inn er vo ice,
and there are no inner utteran ces of Englis h words. Introsp ection does not
inform us that thought s are represen ted in natural language , any more than it
inform s u s that visua l imaginatio n involv es picto rial inn er represen tations.
Introspectio n does not count against Fod or ’ s claim that we think in Me
ntalese.
Let us recast the main point of this section using the appa ratus of epis temic
rules. One ca n know what one thinks by hea ring one ’ s outer spee ch. Likew
ise, one can know what one thinks by “hea ring ” one ’ s inner speech. One route
to w hat one is thinking is the rule:
Think-Alou d If you outwa rdly spea k abo ut x , believe th at yo u are thinking
abo ut x
But a much more com mon route, and the rou te of present interes t, does not
requi re using one ’ s ears:
Think* If you inwa rdly speak about x , believe that yo u are thinkin g about x
The argumen t s o far is simply that Think* is a g o o d rule. Nothin g has been
said yet
about p rivileged or pec uliar access. Moreo ver, Think* is not a neutr al rule.
Since “inwa rdly speakin g abo ut x ” is auditorily imaginin g— perha ps better,
phonolog ically imaginin g— words that are about x , Think* could be rephr ased
as follo ws:
15Think**
If you auditorily imagine words that are about x , believe that yo u
are thinkin g abo ut x
Think**, and hen ce Think*, are obviou sly not neu tral — their anteceden ts
concern the rule- follower ’ s mental states. If Think* is the best that can b e
done, then there is no economic al theor y o f the epis temology of though t.
15For
reasons why ‘phonologically imagine’ might be a more accurate expression than ‘auditorily
imagine’, see Mackay (1992: 127–30). But the latter expression is more familiar, and will be used
throughout. It should be emphasized that different sorts of ‘auditory imagination’ (inner speech,
imagining the speech of another person, imagining noises, imagining music) may well involve
distinct cognitive systems.
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
11 6
A L EX BY RN E
VIII. Downgrading to economy
When one speaks aloud, one is in a position to kno w that one is speakin g. The
fact one is in a position to kno w i s a fact about oneself. One is als o i n a
position to kno w the impers onal fact, that som e speakin g (per haps not even
by a speaker) is occ urring. To know the latter, one just needs to identif y
speech; one does not need any capacity to know one ’ s comm unicati ve
intentio ns, or laryngeal or jaw move ments. That is, at least in principle one may
kno w that som e speakin g— w hich is, in fact, one ’ s own speech — is oc
curring, withou t being in a pos ition to know that this is one ’ s own speech .
A simila r point holds for inner spee ch — with one importan t caveat to be
mentioned in the next para graph. When one auditorily imag ines w ords, one is
in a posit ion to know that one is auditoril y imag ining words. The fact one is in a
posit ion to k now is a fact about one self, speci fi cally a fact abo ut one ’ s ment
al acti vity. Pretend — just for the remain der of this paragraph — that the
imagined words are actually produ ced in a crani al concert hall, and hea rd by
an inn er ear; scien tists can dete ct inner speech by attachin g sensi tive microp
hones to th e hea d. Th en one is also in a position to know the impers onal fact,
that som e inn er spee ch (per haps not even by a speaker) is occ urring. (Spee
ch in the crani al concert hall sounds quite distinc tive.) An Evan s-style “transpa
rency ” observatio n applies here: to kno w that inn er spee ch is occ urring , one
just needs to identif y inner spee ch. And this does not require the capacity for
self-knowledg e, for instance the capacity to kno w that one is imagining speech.
Here i s the caveat . One cann ot kno w that inn er spee ch is occurr ing,
because th ere isn ’ t any. Similarly , i f one forms a visual image of a purple
kanga roo, one cannot know that there is a visual imag e o f a purple kan garoo
— som ething close ly resem bling a real picture of a purple kanga roo —
because the re is no such imag e.
16(Of course, visuall y imag ining a purple kangaroo might involve the manipul
ation of a picture- like represen tation in the brain . But this pictu re-like represen
tation is (a) not something that one is aware of and ( b ) not much like a picture
of a purple kan garoo — it won ’ t b e purple , for one thing.) Although one
cannot know that there is an image of a purple kan garoo, one can be in
a related state. One can simply take the appea rance s a t face va lue, and be
lieve that there is a visua l image of a purple kanga roo. And pres umably this is
the inclina tion of many of us, at le ast.
(This inclina tion may be part of the explan ation of the sense datum theory ’ s
appeal: if to imag ine a purple kan garoo is to be aware of an imag e o f a purple
kangaroo , then, since seeing introspe ctively resem bles visually imag ining,
seei ng mus t als o involve being awar e o f a n imag e — presuma bly an
especiall y vivid one.)
The capacit y t o believe that there is an imag e o f a purple kan garoo
whenever one visuall y imag ines a kanga roo is not a capacit y that presuppo
ses the capacit y for
16Notoriously,
pictorialists about mental imagery are often tempted to deny (a): a recent example
is Kosslyn et al. (2006: 48–9).
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
K N O W
I N G T H A T I A M
T H I N K I N G
1 1
7
self-knowledg e; in particular, one may have the for mer capac ity withou t havin
g the capacity to know that one is ima gining a purple kan garoo.
Lik ewise, the capacity to believe that inn er spee ch is occurr ing whenever
one auditorily imagines w ords does not pres uppose the cap acity for
self-knowled ge. So, although Think* is not neutr al, a related neutr al rule is,
namely :
Think If the inn er voice speaks abo ut x , believe that you are thinking abo ut
x If one follo ws Think, one recognizes , hence kno ws, that the inn er voice
speaks about
x . Since there is no inn er voice , there i s no such knowl edge to be had, and
one cannot follow Think. H o wever, one can try to follo w Think (see } V abo
ve).
IX. Telling the inner voice from the outer voice
One can tell the inner voice apart fro m the outer voice. But nothing has yet b
een said about why this i s so.
17} VI claim ed without arg ument that inner speech inv olves represen ting an
utteran ce, akin to the way in which utteranc es are represen ted when outer
speech is heard. A lot of the experimen tal evidence concerns the corres
ponding thesis for visua l imag ination: visually imaginin g inv olves repres
enting an arrang ement of objec ts, akin to the way in which an arrang ement of
objects is represen ted when one sees the scen e before one ’ s eyes. Since the
visual thes is lends plausibility to the inner speech thes is, the lat ter can be
indirectly suppor ted by citing the evide nce for the form er.
18First , seeing and visuall y imag ining overlap subs tantially at the neural level.
A recent functional ma gnetic res onance imaging (fM RI) stud y conclud ed that
“ visual imag ery and visual percept ion draw on mos t o f the same neural
machin ery ” (Ga nis et al. 2004: 226).
Seco nd, th ere are many interferen ce effects betw een visual i magery and
visua l perception , which are not as pronou nced betw een (say) visual imagery
and aud itory perception in different modalit ies (K osslyn 1994: 54 – 8).
Third, there is the phenomen on of eidetic imag ery, in which sub jects repor t
their images to be very mu ch like pho tograph s (Ha ber 1979).
19Fourt h, percept ion can be mistaken for visual imagery (the Perky effe ct).
Asked to imagine a red tomato while staring at a w hite screen, subjec ts w ill
often fail to notice a clearly visible red round imag e subs equent ly projected
onto the scree n, and will take its size and sh ape to be features of th e imagined
tom ato.
Fifth , visua l imagery can (arg uably) be mistak en for percept ion (the reve
rse-Perky effect). For inst ance, a corticall y b l ind subject (H.S.) with spared
visual imagery denied
17
For a summary of the direct evidence see Mackay (1992).
For a similar study for musical imagery, see Zatorre and
Halpern (2005).
18
19
Hume basically noted the Perky effect at the beginning of the Treatise: “ it sometimes happens,
that our impressions are so faint and low, that we cannot distinguish them from our ideas” (Hume
1740/1978: 2).
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
11 8
A L EX BY RN E
she w a s blind (Anton ’ s syndrome) , which ma y have “res ulted fro m a
confusion of mental visua l imag es w ith real percept s” (Goldenb erg et al.
1995: 1373):
H.S. was given a comb and recognized it from touch... G.G.: Are you really seeing it,
or is it only a mental image? H.S.: I think I a m seeing it a little, very weakly . . . G.G.:
What does “weakly ” mean? H.S.: It is vague and . . . somehow farther away,
blurred. (Goldenberg et al. 1995: 1378)20
Sixth, and finally , w e ordinar ily and natu rally speak of visua l ima ges and of
an inn er voice . (An d not so frequ ently, inciden tally, of “ inner smells ” .) This
is straight forwardly explain ed by the hypothe sis that visua l and auditory
imagery share a distinctive sort of represen tational conten t with, respectively ,
vision and aud ition.
It migh t b e objected that if aud itory imagery is a kind of quasi-hearin g, then
this will turn auditory imagery into auditory halluci nation. And if the inner vo ice
is hallucinated , then how can we so effo rtlessly tell that the inn er voice is not
outer ?
But this can be expl ained by assuming th at the represen tational conte nt of
imag ery is seriou sly degr aded, compare d t o the represen tationa l conten t o f
p erception . (Som etimes it is not — that is w hen imagery cr osses over into
hallu cinatio n.) As is often pointe d out, imagery is i ndetermin ate in a w ay that
perc eption typically isn ’ t. For example, one can visua lly imagine a stripe d
tiger withou t, for so me n, imag ining a tiger with n stripe s, and one can aud
itorily i magine spee ch withou t imag ining that th e voice is loud or soft, mas
culine or femini ne, near or far.
Hume noto rious ly claimed that “ ideas ” (images) differ only from “imp
ressions ” (sense data) in “ their degree of for ce and vivacit y” (Hume 1740/
1978: 2). Ideas, on Hume ’ s view, are lik e faded pho tograph s before the mind
’ s eye, or soun ds fro m worn cassette tapes pl ayed to the mind ’ s ear. But
even after ideas and impres sions are rejected as fictio ns, an insight can be sal
vaged from Hum e ’ s sugges tion: imagery and perc eption differ in their degree
of informatio n.
X. Privileged and peculiar access again
The suppositio n that we try to follo w Think can ex plain priv ileged and
peculiar acce ss. Peculiar access is e xplained b ecause “I cann ot overhe ar
your silent colloqu ies with yoursel f ” (Ryl e 1949: 176). Typic ally one ’ s “
inner voice ” will be entirely self genera ted. Admitt edly, som etimes one ’ s
inner voice migh t speak abo ut x because one overhe ars som eone el se
speakin g (and so thinking ) abo ut x . Howev er, if in such circum stances one
conclud ed from the pronou ncements of the inner voice that Kylie (say) is
thinking abo ut x , this would at best be an acci dentally true belief, not know
ledge.
20
For the reverse-Perky effect at the level of memory, see Intraub and
Hoffman (1992).
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
K N O W
I N G T H A T I A M
T H I N K I N G
1 1
9
KWha
t abo ut privil eged acce ss? Co nsider the rule: ThinkIf Kylie speaks about
x , believe that Kylie is thinking about x This is a g o o d rule, as Ryle in effect
observe d. Suppose one follo ws Think Kon a particular occasion : Kylie utters ‘
Trees die without water ’ , one recogniz es (hence knows) that Kylie is speakin g
about wate r, and thereby conclud es that Kylie is thinkin g about water. Let us
grant that if Kylie is speakin g abo ut wate r then she is thinking about water.
One ’ s bel ief that Kylie is thinking about water w ill then be true and (at le ast in
a typical case) will also be kno wledge.
KIf one ado pts the policy of follo wing Think K, one will not always succee d. In
particular, sometime s one will merel y try to follow it. For instance, perhaps one
misidenti fi es th e speaker as Kylie , o r mish ears her; the resulti ng belief abo
ut Kylie ’ s thought s w ill not then be know ledge. The polic y o f follo wing
Thinkmigh t w ell not produce, in perfe ctly ordinar y situ ations, knowl edge of
what Kylie thinks.
For compari son, suppose that Kylie ado pts the policy of following Think. I f
(as Kylie would p u t it) the inner voice utters ‘ Trees d i e without water ’ , Kylie
thereby conclud es that she is th inking about water. Kylie ’ s policy can never
succee d: her belief about the inner voice ’ s pron ouncement s are always false.
Altho ugh Kylie can never follow Think , but only try to follo w it, for the purpo
ses of attaining self-kn owledg e that doesn ’ t matter. If she believe s t h a t she
hears the inner voice utte r ‘ Trees die with out water ’ , i t i s not fl atly imposs
ible that she has not auditorily imag ined that sent ence, but it is very unlik ely.
And even if she has mist aken an outer voice for her inn er one (an example of
the auditory Perky effect), her belief that she is thinkin g abo ut water is stil l
true, and will likel y b e kno wledge. (If you hear someone el se speak about
wate r, the very act of understandi ng their words requi res that you think about
w ater.)
KThe be liefs produc ed by tryin g t o follow Think are not absolu tely guara
nteed to be true (here Think diff ers from the rule Be l intro duced in } V). A
fortiori, they are not absolute ly guaranteed to be knowl edge. But they are very
likely to b e ; much more so than the beliefs produ ced by follow ing thirdperson rules like Think21. That kind of epistemic acce ss may b e privileged eno
ugh.
21A
notoriously puzzling symptom of schizophrenia is “ thought insertion” (one of the “ passivity
symptoms ” mentioned i n note 4). Patients claim that certain thoughts are not their own, despite
being “ in their minds” . O n the present view of the epistemology of thought, such patients “ hear”
the inner voice speak about x , but do not (try to) follow Think, and conclude that they are thinking
about x . Instead, they conclude (paradoxically) that although there is a thought about x in their
minds, they are not the thinker of that thought.
Thought insertion is too complicated to discuss here, but it should be noted that the present view
does fit nicely with accounts of thought insertion on which patients attribute their own inner
speech to an external agent. See e.g. Jones and Fernyhough (2007), and (for an even better fit)
Langland-Hassan (2008).
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
12 0
ALEX BYRN E
XI. A complication: Think for philosophers
It is probabl y safe to say th at many ordin ary people bel ieve that there i s an
inner voice (in som e unspeci fied sense imp ortantl y like an outer voice ) that th
ey can hea r with an inner ear (in some unspeci fi ed sen se important ly like an
outer ear). Since that is just how things seem, what else are they supposed to
think?
But what of the enlightene d few who are of the contra ry opinion? They appa
rently can ’ t try to follow Think, s o does that mea n that th eir enlight enment
com es at the steep price of self -ignoran ce? Or will the enlig htened — unlike
the vulga r — need a faculty of inn er sense to find out what they are thinking ?
Howev er, it is not an immed iate consequ ence that the enlighte ned cannot
try to follow Think. What is an immed iate consequ ence is that if they do try to
follow Think, and hen ce believe that there is an inn er voice, then their beliefs
are incons istent. The belief that there is an inn er vo ice plays a highly cir
cumscrib ed role, con fined to inferences of the Think variet y; th eir belief that
there is no inner vo ice pl ays a muc h more expan sive role, of the sort
associated w ith paradig matic ca ses of belief. When trying to follo w Think , the
enlightene d would brie fl y allow their belie f that there is an inner vo ice, nor
mally firml y suppres sed, to control their reasonin g.
On this view, imagery is not contin gently conn ected to belief; rathe r, it is
const itutively involv ed wi th belief. Is there any reaso n t o suppos e this is so,
other than to save the present theor y? If perception const itutively inv olves
belief, then give n the close kinship between perception and imagery , the para
llel claim for ima gery shou ld not seem implausible. Moreo ver, the id ea that
perc eption is, at the very least, const itutively connec ted to a tendenc y to
believe is very appea ling. As A. D. Smith puts it, “ How can I disbel ieve my
senses if I have no thing else to go on?” (2001: 289).
The phen omenon that is w idely taken to block the stronger conclusio n that
percept ion constit utively involv es, not a mere tenden cy to bel ieve, but belief
itself, is the phenomen on of know n illusion . I t visua lly appea rs that the lines
in the Müller-Ly er figu re are une qual, but one plain ly believe s that they are
equa l. Hence — it is com monly argued — the “conten t o f percept ion ” , the
propos ition that the lines are unequal , need not be believe d.
Consider the alternative hypothe sis that the belief that the lines are une qual
is present, but inh ibited by the contra ry belief that the lines are equal. Th is
need not involv e the postulatio n o f novel psychol ogical mechanisms . First,
incons istent beliefs are com monplac e. Seco nd, certain d elusional beliefs
(as in the Capgras delusion , for instance) can be larg ely disconn ected from
the res t o f the su bject ’ s world view: despite believi ng that their spouse
(say) has been replaced by an impost er, “ [m]ost Capgras patients do not
show much conc ern abo ut what has hap pened to the real relatives ; they do
not search for them , o r report their disa ppearance to the polic e ” (Ston e
and Young 1997: 333). On the alterna tive hypothe sis, when one know ingly
looks at the Mü llerLyer figu re one ’ s belief that the lines are une qual is lik e
a n inferentia lly isolated delusion . One consider ation in favor of th is hypothe
sis is that it nicely ex plains the
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
KNOW ING THAT I AM THINKING
12 1
that, absent any reason for thinking otherwi se, one will believe — in the
usua l inferentia lly promisc uous way — that the lines are unequa l (Byrn e
2009: 450 – 1). In contrast, the ortho dox alternat ive has no expl anation at all.
So the view that imagery constitutive ly involves belief — harmless delusion s
about a shadowy inn er world of (at least) visual and auditory images — is not
as unmo tivated as it may seem. A prelimina ry case has been made that Think
requi res no modi ficati on to accommoda te the self-knowl edge of the enlig
htened .
22fact
XII. Extensions: Imagistic thinking, thinking that p
Much of our thinkin g “occ urs in pictures ” (as w e say), not in w ords. The exten
sion of Think to this case pres ents no further proble m:
Think-Imag e If th e inner picture is abo ut x , believe that yo u are thinking
about x Knowledg e that one is thinking that p — that trees die w ithout water,
say — can be
accommoda ted b y com bining Think with the rule Be l from } 5: Think-that If the
inn er vo ice says that p and p , b e lieve that you are thinking that p
It is w orth noting that the present account resolv es a puz zle abo ut thou ght.
We often say we are thinking of x w ithout being able to come up with any prope
rty that the thought predicates of x . That is argua bly a uni que featu re that
distingu ishes though t from bel ief, desire, and intentio n.
Cons ider an example. One is thinkin g o f Barry Humphr ies, say. Suppose
one know s this because one ostensibly discer ns a visua l image of Humphr
ies. Altho ugh one migh t not be thinkin g that Humphr ies is F, there is a repor
table predic ational compone nt to the thou ght — Hum phries is imag ined
dressed as Dam e Edna, sa y, or as havin g black hair. But som etimes one can
think of Humphr ies w ith no apparent predic ational component at all — one is
simply “thinkin g o f Humphr ies ” (cf. “Think of a numbe r ” ). The explan ation is
that one may simply aud itorily imag ine the name ‘ Barr y Humphr ies ’ (or— i n
the case of “th inking of a numbe r ”— the numer al ‘ 7 ’ ).
XIII. Unsymbolized thinking?
Suppose the accoun t just outlin ed is right. We can, and do, know what we think
by trying to follo w Think. Privil eged and pec uliar access are expl ained, and
our genera l capacity for reasonin g suf fi ces for self-knowledg e. W ith that all
granted, there migh t still be rain on the para de.
22Further
support for the constitutive hypothesis can be found in Smith (2001) and Craig (1976).
For a defense of the view that delusory “ beliefs” are the genuine article, see Stone and Young
(1997: 351–9), and Bayne and Pacherie (2005).
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
12 2
ALEX BYRN E
In “th ought sam pling ” , subjects are equipp ed with a beeper that sounds at
random intervals . The y are inst ructed to “‘freeze ’ their ongo ing experien ce
and w rite a descr iption of it in a notebo ok” (Hu rlburt 1993: 10). Some subjects
repor t what Hurlbu rt calls “ Unsymbo lized Thinking ”—“ image less thou ght ” ,
i n more familiar termi nology fro m the epon ymous controver sy in e arly twen
tieth-centu ry psychol ogy. Unsymbo lized thinkin g i s:
the experience of an inner process which is clearly a thought and which has a clear
meaning, but which seems to take place without symbols of any kind, that is, without
words, images, bodily sensations, etc. (1993: 5; see also Hurlburt and Akhter 2008)
If subject s can “ just kno w ” (5) that th ey are thinkin g about x without notin g
the (appar ent) pres ence of words or ima ges about x , then th is fact has not
yet been explain ed.
One explan ation might be th at subjec ts are doing som e quick self-inte
rpretatio n: given my behavio r, cir cumsta nces, and /or other topics of thou ght,
it is to be ex pected that I a m thinking about x . (This hypothe sis is arg ued for
in som e detail in Carruthers 1996: 239 – 44.) Altern atively, perha ps inne r
spee ch and visual imag ery do provide the necessary basis for infe rence along
the lines of Think, but subjects are reporti ng the conten t without repor ting
(perhaps because they have forgo tten) the spee ch or imag ery itself. As w hen
one repor ts th at such-and- such withou t repor ting whether one heard som
eone asse rt it or whether one read it in the news paper, the subjects are reporti
ng the message but not the medium.
Hurlbu rt has recently cla rifi ed his view of unsymboliz ed thinking , w ith the
result that an explan ation along the latter lines seems the more plausible, at
least for some cases. As far as his data show, imag ery and inner spee ch are
not entirely abs ent in unsym bolized thinking , but are rathe r dim inished in
relev ance and are not at the focus of attention:
Unsymbolized thinking . . . may well have s om e k in d o f (probably subtle) sensory p
resentation . . . . . . The apprehension of an unsymbolized thought may involve the
apprehension of some
sensory bits, so long as those sensory bits are not organized into a coherent, central,
thematized sensory awareness. (Hurlburt 2009: 149 –50)
Provided that th e “sens ory bits ” provide enough clues , which Hurlbu rt does
not deny, then unsym bolized thinking does not present a problem.
Of course, if there is a way of appre hen ding one ’ s thou ghts “ directly ” , w
hich explain s both peculiar and privileged acce ss, then there ’ s n o need to
try to explain know ledge of unsym bolized thinking using the pres ent model.
Ho wever, desp ite the great w eight placed on the epistemo logy of thought in
the Cart esian tradit ion, plausible account s o f how we kno w that w e are
thinking are remark ably thin on the ground.
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
KNOW ING THAT I AM THINKING
12 3
References
Alston, W. P. (1971), “Varieties of Privileged Access” , American Philosophical Quarterly
8: 223 –41. Page reference to the reprinting in Alston (1989).
—— (1989), Epistemic Justi fication: Essays in the Theory of Knowledge (Ithaca, NY:
Cornell University Press).
Armstrong, D. M. (1968), A Materialist Theory of the Mind (London: Routledge and
Kegan Paul). —— (1981), “What Is Consciousness? ”, The Nature of Mind and Other
Essays (Ithaca: Cornell
University Press). Baddeley, A. (1986), Working Memory (Oxford: Oxford University
Press). Bayne, T., and Pacherie, E. (2005), “In Defence of the Doxastic Conception of
Delusions ”,
Mind and Language 20: 163–88. Byrne A. (2005), “Introspection ”, Philosophical
Topics 33: 79 –104. —— (2009), “Experience and Content ”, Philosophical Quarterly 59:
429–51. Carruthers, P. (1996), Language, Thought and Consciousness: An Essay in
Philosophical Psychology
(Cambridge: Cambridge University Press). —— (2009), “How We Know Our Own
Minds: The Relationship between Mindreading and
Metacognition ”, Behavioral and Brain Sciences 32: 121–82. Craig, E. (1976), “
Sensory Experience and the Foundations of Knowledge ”, Synthese 33: 1 –24. Davies,
M. (2000), “Externalism and Armchair Knowledge ”, i n P . Boghossian and C. Peacocke
(eds), New Essays on the A Priori . (Oxford: Oxford University Press). Dretske, F.
(1994), “Introspection ”, Proceeding of the Aristotelian Society 94: 263–78. —— (1995),
Naturalizing the Mind (Cambridge, Mass.: MIT Press). Ericsson, K. A., and Simon, H. A.
(1993), Protocol Analysis: Verbal Reports as Data (rev. edn.,
Cambridge, Mass.: MIT Press). Evans, G. (1982), The Varieties of Reference (Oxford:
Oxford University Press). Ganis, G., Thompson, W. L., and Kosslyn, S. M. (2004), “Brain
Areas Underlying Visual Mental
Imagery and Visual Perception: An fMRI Study ”, Cognitive Brain Research 20:
226–41. Gettier, E. (1963), “Is Justi fied True Belief Knowledge? ”, Analysis 23: 121 –3.
Goldenberg, G., Müllbacher, W., and Nowak, A. (1995), “Imagery Without Perception —a
Case Study of Anosognosia for Cortical Blindness ”, Neuropsychologia 33: 1373 –82.
Gordon, R. M. (1996), “‘Radical ’ Simulationism ”, i n P . Carruthers and P. Smith (eds.),
Theories
of Theories of Mind (Cambridge: Cambridge University Press). Haber, R. N. (1979),
“Twenty Years of Haunting Eidetic Imagery: Where ’ s the Ghost? ”,
Behavioral and Brain Sciences 2: 583 –629. Hume, D. (1740/1978), A Treatise of
Human Nature , ed. L. A. Selby-Bigge (Oxford: Oxford
University Press). Hurlburt, R. T. (1990), Sampling Normal and Schizophrenic Inner
Experience (New York: Plenum
Press). —— (1993), Sampling Inner Experience in Disturbed Affect (New York: Plenum
Press). —— (2009), “Unsymbolized Thinking, Sensory Awareness, and Mindreading ”,
Behavioral and
Brain Sciences 32: 149 –50. Hurlburt, R. T., and Akhter, S. A. (2008), “Unsymbolized
Thinking ”, Consciousness and Cog-
nition 17: 1364–74.
OUP CORRECTED PROOF – FINAL, 11/2/2011, SPi
12 4
ALEX BYRN E
Intraub, H., and Hoffman, J. E. (1992), “Reading and Visual Memory: Remembering
Scenes That Were Never Seen ”, American Journal of Psychology 105: 101 –14.
Jones, S. R., and Fernyhough, C. (2007), “Thought as Action: Inner Speech,
Self-Monitoring, and Auditory Verbal Hallucinations ”, Consciousness and Cognition 16:
391–9.
Kosslyn, S. M. (1994), Image and Brain: The Resolution of the Imagery Debate
(Cambridge, Mass.: MIT Press).
Kosslyn, S. M., Thompson, W. L., and Ganis, G. (2006), The Case for Mental Imagery
(Oxford: Oxford University Press).
Langland-Hassan, P. (2008), “Fractured Phenomenologies: Thought Insertion, Inner
Speech, and the Puzzle of Extraneity ”, Mind and Language 23: 369–401.
Lycan, W. G. (1987), Consciousness (Cambridge, Mass.: MIT Press). —— (1996),
Consciousness and Experience (Cambridge, Mass.: MIT Press). Mackay, D. G. (1992),
“Constraints on Theories of Inner Speech ”, i n D . Reisberg (ed.), Auditory
Imagery (Hillsdale, NJ: Lawrence Erlbaum). McKinsey, M. (1991). “Anti-Individualism
and Privileged Access”, Analysis 51: 9 –16. Moran, R. (2001), Authority and
Estrangement (Princeton: Princeton University Press). Nichols, S., and Stich, S. (2003),
Mindreading: An Integrated Account of Pretence, Self-Awareness, and
Understanding Other Minds (Oxford: Oxford University Press). Ryle, G. (1949), The
Concept of Mind (London: Hutchinson, Peregrine Books). Shoemaker, S. (1988), “On
Knowing One ’s Own Mind” , Philosophical Perspectives 2: 183–209. —— (1994),
“Self-Knowledge and ‘inner sense ’”, Philosophy and Phenomenological Research 54:
249–314. Page reference to the reprint in Shoemaker (1996). —— (1996), The
First-Person Perspective and Other Essays (Cambridge: Cambridge University
Press). Smith, A. D. (2001), “Perception and Belief ”, Philosophy and
Phenomenological Research 62:
283–309. Sosa, E. (1999), “How to Defeat Opposition to Moore ”, Philosophical
Perspectives 13: 141– 53. Stone, T., and Young, A. (1997), “Delusions and Brain Injury:
The Philosophy and Psychology
of Belief”, Mind and Language 12: 327–64. Thomasson, A. (2003), “ Introspection and
Phenomenological Method ” , Phenomenology and the
Cognitive Sciences 2: 239–54. Williamson, T. (2000), Knowledge and Its Limits
(Oxford: Oxford University Press). Zatorre, R. J., and Halpern, A. R. (2005), “Mental
Concerts: Musical Imagery and Auditory
Cortex ”, Neuron 47: 9–12.