Experiments in Thought - Am MIT spezialisierte KI-Forschung

Experiments in Thought

Walter Hopp
Boston University

What are thought experiments, and how do they generate knowledge? More
speciªcally, what sorts of intentional acts must one perform in order to carry
out a thought experiment, what sorts of objects are such acts directed toward,
and how are those objects made present in such acts? I argue on phenomeno-
logical grounds that the proper objects of thought experiments are, in certain
Fälle, uninstantiated universals and relations among them. I will also argue
Das, in the best of cases, we intuit or “see” these universals and their rela-
tions to one another, and respond to some objections to this view.

Thought experiments in science, philosophy, and everyday life give rise to
all of the following: belief, conviction, justiªcation, perplexity, Und, manche-
mal, Wissen. Wie? More speciªcally, what sorts of intentional acts
must one perform in order to carry out a thought experiment, what sorts
of objects are such acts directed toward, and how are those objects made
present, or not, in the carrying out of those acts? My view is that a careful
and initially metaphysically unbiased phenomenological description will
support the view that the proper objects of thought experiments are, In
certain cases, uninstantiated, often mind-independent universals and rela-
tions among them. I will also argue that, in the best of cases, we intuit or
“see” these universals and their relations to one another, and respond to
some objections to this view.

ICH.
Following Sorensen, I will understand an experiment to be “a procedure
for answering or raising a question about the relationship between vari-
ables by varying one (or more) of them and tracking any response by the
other or others” (Sorensen 1992, P. 186). Every actual experiment requires
that the experimenters undergo at least some perceptual experiences that

Perspektiven auf die Wissenschaft 2014, Bd. 22, NEIN. 2
©2014 by The Massachusetts Institute of Technology

doi:10.1162/POSC_a_00129

242

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

243

are appropriately related to the variables whose relationship the experi-
ment is designed to test. In the best case, the experimenter is able to per-
ceive the variables themselves. Quite often, Jedoch, we must rest content
with perceiving things that we know to stand in quite intimate informa-
tional relationships with the variables being tested. We cannot directly
perceive the readiness potential, zum Beispiel, nor can we reliably perceptu-
ally discriminate among very short periods of time. But we can perceive
electroencephalographs and clocks, and we have good reasons to believe
that the states of those instruments co-vary in lawlike ways with the phe-
nomena that elude our unaided senses. Our conªdence in the obtaining of
informational relationships between our instruments, on the one hand,
and the target phenomena, on the other is typically itself grounded in
other perceptual experiences.

It is not enough to perceive the variables or items suitably related to
them to acquire empirical knowledge by means of an experiment. The rel-
evant perceptual experiences must also be appropriately related to the con-
tents of the experimenter’s beliefs and judgments. I have a view on what
this relationship consists in, which is really just a modiªcation of
Husserl’s, and which I will simply lay out here. Suppose I merely think
that the window in my ofªce is clean. By itself, merely thinking this pro-
vides absolutely no evidence whatsoever for the proposition (cid:2)my window
is clean(cid:3). Suppose that I turn to the window and take a look at it. After a
sufªciently but not obsessively close visual inspection, I judge that the
window is clean. What was, Vor, a mere thought with no epistemic
worth has given way to knowledge. This is fulªllment. In such an act, ICH
“experience how the same objective item which was ‘merely thought of’ in
symbol is now presented in intuition, and that it is intuited as being pre-
cisely the determinate so-and-so that it was at ªrst merely thought or
meant to be.”1

Acts of fulªllment are quite often cases of knowing. In them, an act
in this case—
of perceiving and an act of thinking—a judgment,
converge upon the same object. Allgemeiner, in every act of fulªllment
(A) the object is intuited, (B) the very same object is thought about, Und
(C) there is an appropriate synthesis between the acts of
intuiting
and thinking. The object in this example—that which the act is about—is
not just the window. Eher, what is presented in perception and then rep-
resented in the judgment is a state of affairs—the being-clean of the win-

1. Husserl 1970, P. 694. Also see Husserl 1970, P. 720: “What the intention means,
but presents only in more or less inauthentic and inadequate manner, the fulªllment—
the act attaching itself to an intention, and offering it ‘fullness’ in the synthesis of
fulªllment—sets directly before us, or at least more directly than the intention does. In ful-
ªllment our experience is represented by the words: ‘This is the thing itself’.”

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

244

Experiments in Thought

dow. Endlich, these acts do not merely occur side-by-side. They are related
in such a way that I am aware of the window’s being the way it is repre-
sented as being in the judgment.

Although the case of fulªllment above resulted in me knowing that my
window is clean, not all acts of fulªllment confer justiªcation on empirical
beliefs. Had I imagined my window as clean, my thought that it is clean
would have been intuitively fulªlled, but imagining would not have con-
ferred any justiªcation on the judgment that the window is clean. Das
would be a case of merely illustrative fulªllment. The reason is that while
imaginative acts present, rather than merely represent, their objects, Sie
do not present actual objects and states of affairs in the ºesh. Perception is,
while imagination is not, an “originarily presentative” act (Husserl 1982,
P. 327) with respect to actuality; it “is that mode of consciousness that
sees and has its object itself in the ºesh” (Husserl 2001, P. 140). Percep-
tion is a self-giving act with respect to actual individuals and states of
affairs.

Fulªllment is, in every case, completely unlike inferential justiªcation.
My belief that P cannot be inferentially justiªed solely on the basis of an-
other belief that is about exactly the same state of affairs that P itself is
um. Eher, inferential justiªcation involves coming to believe the
world is one way because of other ways one believes it to be; reasoning is a
matter of determining what must, might, or cannot be the case given that
something else is the case. In fulªllment, Jedoch, the fulªlled act and the
fulªlling act in virtue of which it is justiªed must be about the same state
of affairs; we believe the world is a certain way because it is that way.
Epistemic fulªllment is that unique intentional act in which we conªrm a
truth-bearer by consulting its truth-maker.

This does not mean that experimental results are cases of non-
inferential knowledge. Most experiments, as is well known, depend upon
a wide range of background assumptions in order for the theoretical im-
portance of their results to be interpreted, and often also depend on beliefs
about the absence of potential defeaters—and the more complex and the-
ory-laden the experiment, the more vast the array of possible defeaters.
James McAllister has argued persuasively that the evidential value of an
experiment is not an intrinsic property of it. This should not force us into
his position, Jedoch, according to which “the evidential signiªcance of
experiment is conferred on it in particular areas of science at particular
times by the persuasive effort of scientists” (McAllister 1996, P. 237).
Eher, we can and should insist that all experiments do, at some level or
andere, involve the noninferential veriªcation of at least some propositions.
We may not be able to determine anything noninferentially about free
will by observing EEG readings while a subject reports the time he or she

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

245

felt a conscious decision to press a button, since the interpretation of the
results depends on beliefs about such things as the relationships between
EEG readings and the readiness potential, the readiness potential and neu-
ral activity, neural activity and conscious intentions, conscious intentions
and the self-conscious noticing of conscious intentions, und so weiter. Never-
theless, we can acquire noninferential knowledge of such propositions
as “The subject pressed the button” and “The EEG reads 5 microvolts.”
Knowledge of this sort does not evidentially depend on the persuasive ef-
forts of scientists, but is presupposed by it. If scientists themselves did not
encounter such evidence in experimentation, it would be difªcult to see
where they could summon the rational motivation for engaging in any
persuasive efforts or experimentation at all.

Endlich, because fulªllment is a more complex act than mere perception
(or merely thinking), it is susceptible to a wider array of failures. One ob-
vious type of failure stems from nonveridical perceptual experiences. Für
Beispiel, one’s thought that there is a car coming down the street might
seem to be fulªlled by a hallucination of a car coming down the street. Aber
there are other ways fulªllment can fail. Insbesondere, the object one is
thinking about might not actually be identical with the object one is per-
ceiving. Twin Earth cases exhibit this possibility. If transported to Twin
Earth, your singular thoughts about earthly individuals would seem to be
fulªlled by your experiences, but they would not be. Your perception of
twin-Kripke might be perfectly veridical, yet your thought “there’s
Kripke” would not be genuinely fulªlled. Other familiar sorts of errors
can occur. One’s thought that a student is cheating might seem to be
fulªlled on the basis of his sidelong glances, Wann, schließlich, one has only
observed sidelong glances. One might hear the note C veridically, Aber
misidentify it as a G. And so on.

With respect to actual individuals and states of affairs, only perception
and introspection are self-giving and are, entsprechend, the only ones that
can serve as constituents in acts of epistemic fulªllment directed upon
such objects.2 Actual experiments essentially involve perceptual or introspec-
tive experiences, and those experiences must ªgure as constituents of acts
of epistemic fulªllment. Wie wir sehen werden, Jedoch, perception and intro-
spection are not the only kinds of self-giving acts. Rational intuition and
even imagination and hallucination can be self-giving with respect to

2. Or, more precisely, in acts of primary epistemic fulªllment, which alone concern us
Hier. Intuitive memories can also fulªll empirical beliefs, as can certain acts of image-
Bewusstsein (looking at pictures, videos, and maps), but they are each, in differing ways,
dependent upon perception. They provide derivative epistemic fulªllment. For more on
these distinctions, see Hopp 2011, §7.2.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

246

Experiments in Thought

properties or essences and the relations among them, and that is what
sometimes occurs in the case of thought experiments.

II.
Was, Dann, about thought experiments? Is a thought experiment just
like an actual experiment minus the perceptual experiences and acts of
fulªllment that are essential to the latter? Is a thought experiment a case
in which those experiences of fulªllment are replaced by mere thoughts,
devoid of fulªllment? I am not sure anyone holds this view, but Sorensen’s
claim that thought experiments are a type of unexecuted experiment
might suggest something along those lines (Sorensen 1992, S. 213–4).
One obvious problem with this claim is that thought experiments them-
selves can be executed or unexecuted. The view is also suggested as a real
possibility by the leading question that frequently ªgures in this debate:
how can merely thinking about something provide us with knowledge
about it? The answer, I suggest, is that it cannot, at least if by “merely
thinking” we intend to contrast that with “having an experience of epi-
stemic fulªllment.” A thought experiment is neither a mere thought
about an actual experiment, nor is it a mere thought about the entities ob-
served and manipulated in the course of an actual experiment. Merely
thinking about carrying out an experiment in which one measures the
timing of the readiness potential and subjects’ reports of becoming con-
scious of making a decision provides no evidence whatsoever about the re-
sults of that experiment. Allgemeiner, merely thinking about actual
states of affairs provides no reason to believe that they obtain. But thought
experiments do, or at least can, provide evidence for propositions. So ei-
ther thought experiments provide some other sort of access, distinct from
mere thinking and from fulªllment via perception (on pain of being actual
experiments), to actual states of affairs, or they do not have actual states of
affairs as their principal objects.

What might thought experiments be if the ªrst option is correct—if,
das ist, they are directed towards actual states of affairs by means other
than mere thinking or fulªllment by means of perception? The only an-
swer I can think of is that they are really, despite appearances, arguments.
This view has been endorsed by John Norton, and is susceptible to some
rather serious objections. Different conclusions can be drawn from the
same thought experiment (Bishop 1999). Thought experiments are not
formally valid or invalid (Häggqvist 2006). And many thought experi-
ments look an awful lot like actual experiments rather than arguments
(Sorensen 1992).

I suggest we provisionally take this similarity at face value and see
where it leads us. I further suggest that we take the experiences involved

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

247

in thought experiments at face value. What I mean by that is that we pro-
visionally set aside whatever metaphysical or empirical presuppositions,
however established we might take them to be, and take a frank look at
the experiences involved in performing thought experiments. Das ist, Wir
should exercise the phenomenological epoché, welche, despite Husserl’s
long-winded and often foggy descriptions of it, is no great mystery. Wir
examine our experiences and determine what they are aiming at and what
they’ve acquired (Husserl 1969, P. 177) or seem to have acquired, und in
that order. What we do not do is determine in advance what they must or
cannot acquire, and then deduce what they must have been aiming at.

An example might illustrate the difference. We can imagine a philoso-
pher convincing himself that it would be irrational to attempt to do some-
thing unless one knew for sure that one was going to succeed. Suppose
such a philosopher observes me shooting free throws and asks me what I
am trying to do. “I’m trying to make free throws,” I answer. “But I see you
miss quite a few of them. And yet you’re rational, Rechts? So you must not
be trying to make free throws. You must really be trying to do something
else—maybe release the ball from your hands.” This philosopher has not
consulted the experiences of trying to do something to determine what
they are aiming at and then determine what they have acquired. Eher,
he has determined in advance what such acts must acquire, and then in-
ferred what they must be aiming at. The results speak for themselves.

It is this latter, non-phenomenological approach that generates such
counterintuitive theories as the sense-datum theory: since perception and
sensation simply must get things right, they cannot be directed towards
physical objects and states of affairs. The corresponding danger in this
arena should be obvious, and it consists in reasoning as follows: since pla-
tonic objects do not exist, and since we simply could not have a faculty for
accessing them if they did, thought experiments must not be aiming at
those. Naturally these are objections one must deal with, and if it turns
out that we really do aim at physical objects in perception, universals in
thought experiments, or objective moral values in our moral discourse,
and that such aimings simply could not hit their mark, perhaps an error
theory of those domains is called for. But such philosophical views should
not be allowed to distort the content and character of our intentional
Erfahrungen.

III.
My view is that thought experiments, at their best, are in fact founded on
acts of fulªllment in which we intuit universals and the relations among
ihnen, and that the actual instantiation of those universals and relations is
immaterial. I begin with a thought experiment intended to show that this

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

248

Experiments in Thought

is indeed possible. Suppose that I am faced with the task of moving a
rather bulky couch out of a cramped apartment and through a narrow
doorway. I know it won’t be easy, but I also know it can be done. After all,
I got it in here. As I survey the scene before me, something remarkable
happens: the couch begins to ºoat through the air of its own accord,
makes a surprisingly simple sequence of twists and turns, and exits
through the doorway and down the stairs. In disbelief, I rub my eyes, Und
upon opening them again ªnd the magical couch sitting in its original po-
sition, stubbornly unmoved.

It’s easy to say what went wrong here. I underwent a hallucination. Mein
experience presented the world to be a certain way, and it was not that
Weg. But something went right, zu: I now know, or at least have a justi-
ªed belief, that the couch will ªt through the doorway, and I know just
how to make it ªt. How is that possible? What was I right about?

The knowledge I acquired on the basis of this hallucination concerns
the spatial properties of the couch and the doorway. I know that some-
thing with the shape and size of my couch can ªt through something with
the shape and size of my doorway if maneuvered this way and that. Diese
spatial properties and relations, darüber hinaus, were genuinely presented in the
Erfahrung. I did not merely think about those properties and relations—
a procedure which, before the hallucination, did not get me anywhere. ICH
beheld them.

What more can we say about what I beheld and learned? One thing we
cannot say is that I veridically saw a physical individual with the same
properties as my couch moving through a physical space with the
same properties as my doorway. Perhaps, Dann, I beheld different kinds of
individuals, individuals that are non-physical or at least not “external”—a
mental picture or an idea, sagen, or maybe a sense datum. I am fairly con-
ªdent that any answer along these lines is incorrect. Erste, hallucinations
are errors. But if my experience was veridically of actual particulars that
genuinely possessed the properties that I beheld, then my experience was
not erroneous after all. I cannot be wrong about a couch or the state of
things in my apartment if my acts are not even about them.

Zweite, the experience could not possibly have presented such individ-
uals veridically anyway, since my experience presented the relevant spatial
properties as instantiated by middle-sized physical objects out there in
physical space, not as instantiated by little mental pictures in a little men-
tal space or by sense data in a phenomenal space. Perhaps sense data can be
located in physical space. Auch so, my perceptual experience was directed
at something with properties that no sense datum could have. Für in-
Haltung, it looked as though it could be sat on. Most importantly, it ap-
peared, as all physical objects do, as though it had more to it than what

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

249

was presented in my experience—it looked like it had hidden sides that
could come into view, and presented sides that could become hidden from
view. As it rotated in the air, it looked like the same thing that was chang-
ing its orientation and position and whose parts were coming into or re-
ceding from view. But sense data do not have more to them than what is
manifest, Und, in stark contrast to the experience of physical objects, any
change in one’s experience of a sense datum is a change in the sense datum
one experiences. If we insist that some actual individuals must bear the
properties that are genuinely presented in hallucinatory experiences, Dann
we must insist that there is some actual individual that has the same size,
shape, and spatial location as my couch, something that can only be par-
tially perceived at any one time, which can be presented as identical across
experiences that are not identical, and which can be perceived simulta-
neously by many perceivers. But what besides something physical and
“external” could possibly instantiate those properties?

To recap, Dann: ªrst, in this hallucination I am genuinely presented with
spatial properties and relations. Zweitens, there are no individuals that
could plausibly be thought to bear the spatial properties and relations that
are presented to me. When I hallucinated, nothing seven feet long came
into existence. No light was absorbed or reºected by a new gray thing. NEIN
doorway, whether physical or mental, was momentarily rendered impass-
able by a seven-foot long gray obstruction. My experience undoubtedly in-
volved a number of mental individuals, perhaps even mental models or
Bilder. But the object of my hallucination was not a mental model or im-
Alter, but what the model or image (supposing one exists) is a model or
image of.

The third point is that I learned something on the basis of this presen-
Station. My knowledge was not of any particular couch, since there was no
individual veridically presented to me in that act. And yet I know some-
thing about my couch. The reason is that my experience provided me with
the knowledge of any possible individual bearing the properties that were
presented to me, and my couch is one of those individuals. Though
nonveridical with respect to any individual, my hallucination was veri-
dical with respect to something.3

The acquisition of knowledge on the basis of this hallucination was not
a thought experiment, any more than perceptual experiences and fulªll-

3. Compare this with Mark Johnston’s view: “When we see we are aware of instantia-
tions of sensible proªles. When we hallucinate we are aware merely of the structured quali-
tative parts of such sensible proªles” (Johnston 2004, P. 137). My view, which is inspired
by this one, differs from it insofar as uninstantiated sensible qualities and relations are
among the objects we are veridically aware of in hallucination. But hallucination is also
nonveridical insofar as it presents those qualities and relations as instantiated.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

250

Experiments in Thought

ments are actual experiments. But the difference between it and a thought
experiment is just this: in an experiment, one purposefully performs a pro-
cedure in order to answer or raise a question by varying and tracking
the relations among variables. Experiences of the sort I had are among
those that enable one to track those relations. Suppose that I could halluci-
nate like this at will. Dann, when struck with the question of how to get
the couch through the door, I could induce an experience like this on de-
mand. If the couch doesn’t get through one way, I repeat the hallucina-
tion. I thereby deliberately vary one variable, the orientation of the couch,
while keeping others, such as its shape and size and the shape and size of
the doorway, constant, and track their relations. Now I will have acquired
knowledge on the basis of a thought experiment. The nonveridicality of
the act with respect to actual individuals and states of affairs is immate-
rial, provided it is veridical with respect to the properties presented in the
hallucinatory experiences.

If this sounds suspiciously similar to simply saying that I have con-
ducted an experiment by means of my imagination, that’s because it is. Bei
this point, now that I am savvy to the hallucination, the chief difference
between it and a vivid imaginative act is that the hallucination cannot
exist side-by-side with a set of veridical experiences. A hallucination of
a couch in a doorway cannot coexist harmoniously with a veridical ex-
perience of that same couch resting in the room, zum Beispiel. Imagina-
tive acts, Jedoch, can: I can imagine anything whatsoever, and what is
thereby presented to me does not compete with what is presented in my
veridical experiences. This is because the imaginative act is similar to the
hallucination, minus the latter’s positing of actual individuals and states
of affairs. If I imagine a seven foot-long couch ªtting through a doorway, ICH
do not thereby posit, in either thought or imagination, any actual individ-
uals. The object of imagination is not a mental image, entweder, contrary to
some views that, like mine, stress the observational character of thought
experiments (sehen, zum Beispiel, Gendler 2004). If what I were imagining
were a real, mental particular—an image or model in my mind, say—then
I would not be imagining but introspecting, and would not be imagining a
seven foot-long couch, since nothing mental is a couch or seven feet long
and I know this. What imagining and hallucinating have in common is
that in such acts, properties and relations can be veridically presented.

Not only can properties and relations be veridically presented in both
hallucination and imagination, but those acts are self-giving with respect to
ihnen, just as perception is with respect to actual individuals and states of
affairs. Mit anderen Worten, intentional acts directed towards properties, rela-
tionen, and other universals can be epistemically fulªlled on the basis of
such presentations. While hallucination purports to be self-giving with

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

251

respect to actual individuals and states of affairs but is not, and while
imagination does not even purport to present us with actual individuals,
both are acts in which we can, at least in principle, behold properties. Fur-
thermore, as Mark Johnston points out, hallucination can be an original
source of de re knowledge of qualities (Johnston 2004, P. 131). I can ac-
quire the concept of a square in a hallucinatory experience, zum Beispiel.

Acquiring knowledge on the basis of thoughts which are fulªlled, Aber
not fulªlled on the basis of our experience of actually existing individuals,
is utterly commonplace, and lies, to some extent or other, within virtually
everyone’s ken. A good chess player can determine how to force his oppo-
nent to make various moves, how to fork a queen, how to threaten mate,
and so forth by imagining what will happen if he moves thus and so. Er
can and probably will perform various thought experiments by determin-
ing what will, could, or could not occur if he moved this way or that. Das
is not because he is playing a game of chess on a little mental chessboard.
Perhaps he is. The important point is that it does not matter. What he
learns in imagining possible chess moves is invulnerable to defeat by the
factual existence or nonexistence of any individual chessboards, ob
physical or mental. It is this arbitrariness with respect to the identity and
even the existence of the individual exhibited in intuition that allows it to
fulªll thoughts about any possible individual.

Earlier I argued that if thought experiments give us knowledge—and
they sometimes do—then either they do so by means other than mere
thinking or fulªllment via perception, or they do not have actual states of
affairs as their objects. I rejected the ªrst view, and have defended the
zweite: thought experiments provide knowledge of properties and their
relations—relations such as compatibility, incapability, and necessary co-
instantiation. In at least some cases, darüber hinaus, thought experiments are
founded on self-giving intuitions of those properties. Just as actual experi-
ments are founded on presentations of actual states of affairs, thought ex-
periments are founded on presentations of ideal states of affairs.

Before turning to some of the problems with my account, let me point
out two beneªts. The ªrst is that it explains why thought experiments re-
semble actual experiments so closely: both are attempts to raise or answer
a question by manipulating one or more variables and tracking what hap-
pens to the others, on the basis of a self-giving presentation of those variables or
something suitably related to them. Thought experiments and actual experi-
gen, Dann, differ not insofar as one involves “perception” and one in-
volves mere thought. They both involve a kind of “perception,” at least if
by “perception” we mean an act which presents, rather than emptily repre-
sents, its object, and both involve the fulªllment of various thoughts on
the basis of those presentations. They differ, eher, insofar as actual exper-

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

252

Experiments in Thought

iments are founded on presentations of actual states of affairs, while
thought experiments are founded on presentations of states of affairs
whose factual existence is irrelevant.

A second virtue of this account is that it does not conform to a charac-
terization of thought experiments that does make them look genuinely
magical. John Norton, zum Beispiel, writes, “Thought experiments are
supposed to give us information about the physical world. From where can
this information come?” (Norton 1996, P. 333) Roy Sorensen claims that,
according to rationalism, “we can learn about the world without experi-
ence” (Sorensen 1992, P. 15). And that sounds downright spooky. On my
view, we cannot know anything about the actual world on the basis of
thought experiments alone, since thought experiments are founded on ex-
periences that either contain no reference to actual individuals, or whose
reference to those individuals is such that the individuals are treated as
mere possibilia. The best they can do is tell us what is true in some possible
Welt, or any possible world, or any possible world which has such-and-
such features, properties, or kinds. Zum Beispiel, Galileo’s cannonball
thought experiment might be able to inform us that Aristotle’s theory of
motion cannot describe any possible world. But it cannot tell us that our
world contains cannonballs, or even that it contains any objects with mass.
A thought experiment concerning how one can mate in three moves given
a certain conªguration cannot give us any information about our world by
selbst, since it contains no information about actualities at all. It cannot,
without further ado, tell us how Smith can beat Jones. Thought experi-
ments do, Jedoch, tell us about the world in conjunction with informa-
tion that is about the actual world. Once I know that there is an actual
chessboard conªgured in the same way as the one I have imagined, then I
can know something about the actual world, namely that I can beat an ac-
tual opponent by moving actual pieces thus and so.

IV.
So far I have emphasized thought experiments founded on acts with sensu-
ous or quasi-sensuous content—acts of hallucination and imagination.
Not all thought experiments rely upon imagination, and almost none rely
upon hallucination—though many could, as the thought experiment in the
previous section shows. Viele, and perhaps most, thought experiments
succeed in spite of whatever imaginative acts we happen to carry out while
performing them. Think, zum Beispiel, of Twin Earth thought experi-
gen, or Maxwell’s Demon, or the notorious trolley cases in ethics. Nicht
only do imaginative acts seem inessential to such thought experiments,
they might sometimes prove to be a dangerous distraction. How does that

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

253

square with my quasi-perceptual model of rational intuition, according to
which we can “intuit” universals and their relations?

Well, as a stab at an answer, I think it is because not all “intuitions” or
self-giving acts are sensuous or quasi-sensuous, nor are all things which
can be given capable of being presented in sensuous or quasi-sensuous
acts. Fulªllment, at its best, is a matter of having the objects one thinks
about given with the highest degree of adequacy which objects of that
type are capable of being given. This often occurs without anything re-
sembling imagery. And sometimes it occurs without doing anything
mehr, seemingly, than understanding a proposition. When I think that
2(cid:4)3(cid:5)5, I ªnd it instantly and primitively compelling, just as, when I
look at and listen to the television which is on and consider the proposi-
tion that the TV is on, I ªnd myself unable to refrain from believing it. ICH
can also merely think that the TV is on and feel no compulsion to believe.
There is no contrasting case with respect to the proposition that 2(cid:4)3(cid:5)5,
Jedoch. Provided I actually think about the entities represented by the
proposition, I ªnd myself convinced that it is true.

Does this show that in certain cases intuition is dispensable and mere
thinking is sufªcient for rational insight? And if so, then why not settle
for a more modest rationalism, such as Christopher Peacocke’s, Das
grounds a priori insight into the very fabric of concept possession, and do
away with the problematic notion of rational intuition?

In the ªrst place, a subject’s inability to think a given proposition with-
out also ªnding it self-evident might show that merely thinking that
proposition is sufªcient for ªnding it evident. What it does not show,
Jedoch, is that the complex phenomenon of fulªllment—the wedding of
thought and intuition—is not taking place, or that ªnding the proposi-
tion evident is just a matter of understanding it. Consider the cogito. A
proper phenomenological account of why the proposition that I exist is
self-evident whenever I think it is not that merely thinking it just is to
ªnd it evident, or that various other known capacities of mine would be
lacking if I did not. Eher, I ªnd it evident because the state of affairs
that it represents is always present to me in a privileged way.

In the second place, we can draw the contrast between merely think-
ing about something and having it given to consciousness in the case
of logical and mathematical propositions too. I know there was a time
when I did not ªnd certain logical equivalences—DeMorgan’s Laws, für
instance—evident. I had memorized DeMorgan’s Laws, and I could use
them correctly in proofs, but without insight. Eventually I gained insight
into their truth. Now I cannot consider them with due attention without
ªnding them evident. This is not a matter of mere memorization or repe-
tition. I also memorized and repeatedly applied the quadratic formula cor-

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

254

Experiments in Thought

rectly while in school, and no behaviorist would be able to tell the
difference between my understanding of it and my understanding of
DeMorgan’s Laws. But I never achieved insight into its truth. When it
comes to certain propositions, it is not that I have acquired the ability to
know them just by thinking them. It is that I have lost the ability
to think them without having those thoughts fulªlled.

Wieder, I think my four year-old son understands the proposition that
2(cid:4)3(cid:5)5. But he does not appear to ªnd it self-evident. When I ask him
ob 2(cid:4)3(cid:5)5, he goes through a procedure—counting ªngers—to de-
termine the answer. That he understands the proposition is part of the best
explanation for his successful use of this procedure. So even though I can-
not produce the contrast between merely thinking that 2(cid:4)3(cid:5)5 and hav-
ing insight into it in my own case, that contrast does seem to show up be-
tween my four year-old son and me. I possess something that he, as yet,
lacks, namely rational insight into the truth of that proposition, sogar
though both of us possess the ability to think that proposition and, Dort-
Vordergrund, possess the concepts that make it up.

Dritte, even if mere concept possession fully explains our ability to ªnd
certain propositions evident, this is compatible with an account which in-
sists that we bear direct epistemic relations to platonic entities. Suppose
that every time I think that nothing can be red and green all over, I ªnd it
evident. Suppose, weiter, that this is best explained by my possession of
the concepts ‘red’ and ‘green’. This only constitutes an alternative to Pla-
tonism if possessing those concepts does not give me epistemic access to
platonic entities. Ähnlich, if someone found it mysterious that we can
stand in direct perceptual relations with distal objects, appealing to our
possession of a visual system would not constitute, by itself, zu von-
mystifying move. If spelled out in the most plausible way—a way, für
starters, that did not maintain that we see our visual systems—it would
very likely simply constitute an explanation of what, in us, allows us to di-
rectly see distal objects. But what do I possess when I possess the concepts
‘red’ and ‘green’? What I possess, I suggest, is the ability to carry out in-
tentional acts directed at red and green and the various actual and possible
states of affairs into which they enter. In exercising those concepts, Ich bin
not thinking about the concepts or the propositions in which they ªgure
(welche, to complicate matters, might themselves turn out to be pla-
tonic—there aren’t as many concepts of red as there are acts of thinking of
the color red). I am thinking with them. A very simple argument for this
is that concepts and propositions represent things. But what I think and
know about when I know that nothing can be red and green all over does
not represent something. What I come to know about in thinking that
nothing can be red and green all over are the colors red and green and

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

255

their inability to be jointly instantiated. In der Tat, I think I know quite a bit
more about red and green, and circles and bachelors, than I do about the
concepts of them. I bet you do too. List everything you know about cir-
cles. Now list everything you know about the concept of circles. Which
list contains more information?

Außerdem, the knowledge we have of properties and essences is not
epistemically indirect, since we do not typically base our knowledge of
those concepts’ referents on prior knowledge of the concepts themselves.4
We do not ªnd out about circles by examining the concept ‘circle’, but by
using the concept ‘circle’. Most of what we know about the concept ‘cir-
cle’, eher, depends on prior knowledge of circles. But these referents are
the very platonic entities access to which is supposed to be problematic,
and to which we do have access on any theory of concept possession that
gets the objects of our a priori knowledge right. If a theory maintains that
the possession of concepts gives me direct insight into universals and nec-
essary states of affairs, then rational intuition, or something very much
like it, has not been dispensed with, but folded right into the structure of
concept possession. Such a theory is hardly a more conservative position
than my own, according to which concept possession and rational intu-
ition are distinct. And if it renders me incapable of being conscious or
knowledgeable of such entities—if, zum Beispiel, it substitutes the proper-
ties red and green with the concepts of red and green—then it is objec-
tionable on phenomenological grounds for replacing the genuine objects
towards which I am directed with completely different ones.

V.
The account I am putting forward is most similar to that endorsed by
Platonists such as James Brown. We shouldn’t overstate the similarity be-
tween my view and Brown’s. For one thing, as Harald Wiltsche has
pointed out, I endorse Platonism for phenomenological reasons: they show
up in experiences of fulªllment, Und, unless extremely compelling reasons
are presented not to do so, we ought to take such experiences at face value.
Brown’s Platonism, according to Wiltsche, is more theoretically moti-
vated. For Brown, it provides the best alternative to Humean accounts of
the laws of nature, which counterintuitively hold them to be nothing
more than regularities. Außerdem, while Brown and I both hold that we

4. Even moderate rationalists do not think this. For Peacocke, possessing a concept
amounts to “knowing what it is for something to be the concept’s semantic value (its refer-
enz).” But, as he explains, this does not require “that someone can ªrst grasp a concept
and then go on to raise the question of what it is for something to be its semantic value.
On the present theory, grasp consists in knowing the answer to that question, so such a
state of ignorance is not possible” (Peacocke 1992: 22).

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

256

Experiments in Thought

can be acquainted with universals in “some sort of intellectual perception”
(Braun 2002, P. 1131), I do not think any contingent laws of nature are
among the abstract entities that we can intellectually perceive in thought
experiments, or at least not in thought experiments alone. Thought exper-
iments tell us nothing about which things or laws are actual.

Despite these differences between my view and Brown’s, the metaphys-
ical commitments of mine are liable to draw similar objections from simi-
lar quarters. One of the most common objections to this view is that, als
Rachel Cooper puts it, “there is no account of how the Platonic universals
are ‘perceived’” (Cooper 2005, P. 333). Although Cooper does not consid-
er a Husserlian account like that above, the real objection appears to be
the one raised by John Norton to the effect that “The mechanism of per-
ception of Platonic laws is essentially completely mysterious” (Norton
1996, P. 360).

One response to this argument comes from Brown, who raises the same
worries for ordinary sense perception. True, we know a great deal of the
physical story that begins with the stimulation of the retina and ends with
signals being sent to the visual cortex, but there ends our knowledge. Ex-
actly how this process culminates in belief is something concerning which
we are completely ignorant (Braun 1991, P. 65). Das, Jedoch, should
not make us skeptical about the existence of sense perceptual experiences
giving rise to beliefs, nor should we be skeptical of the deliverances of
sense perception itself. Ähnlich, the absence of an account of precisely
how platonic perception works should not make us doubt its existence or
results either.

Brown’s response illustrates something important about mysterious-
ness: a phenomenon might be mysterious insofar as we lack an under-
standing of how it comes about or operates, even while its existence is
fairly uncontroversial. The existence of consciousness ought to be uncon-
troversial, zum Beispiel, as should the connection between consciousness
and events in the brain, even though we have, as yet, no widely agreed
upon account of the nature of that connection. The charge that seeing uni-
versals or essences is mysterious, Dann, had better amount to more than the
mere observation that we lack an account of how it works, since that, von
selbst, gives us very little reason to doubt that it exists.

The real force of the objection, as Brown acknowledges, is that there is
nicht, and cannot possibly be, any acceptable causal story—one couched in
terms of things appropriately thought of as “mechanisms”—about how
our minds (or brains) interact with platonic entities. At least that is
Cooper’s objection to Brown’s account. And if there is no possibility of
causal interaction with such entities, then we cannot even refer to them,
much less know them. As Paul Benacerraf expresses it:

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

257

I favor a causal account of knowledge on which for X to know that
S is true requires some causal relation to obtain between X and the
referents of the names, predicates, and quantiªers of S. I believe in
addition in a causal theory of reference, thus making the link to my
saying knowingly that S doubly causal. (Benacerraf 1973, P. 671)

The argument in favor of this view is that we would not accept the claim
that X knows that p if we had compelling evidence that X could not have
stood in the right causal relations with those portions of reality in virtue
of which p is true. Nostradamus, we reason, could not have known any-
thing about or even made de re reference to Hitler, because he did not
stand in the right causal relations to Hitler. Since we don’t ever stand in
causal relations to such abstracta as numbers and, presumably, platonic ob-
Projekte, reference to and knowledge of them is impossible.

As plausible as this position is, I think causation is a red herring here.
Certainly causal relations aren’t nearly sufªcient to ground either knowl-
edge or reference. Gamma rays and mitochrondrial DNA have causally in-
teracted with every human who has ever lived, but knowledge of them is a
recent achievement. What really seems to matter is that knowledge of
something requires that it be either present to you, present to someone
whose testimony can be relied upon, or inferable from what is present
to you. In the case of empirical objects and states of affairs, it is typically
the case that something can be present in the required way—through
perception—only if one is causally related to it. Causation is, mit
respect to these sorts of entities, necessary for the type of presence-
to-consciousness that, I suggest, really makes Benacerraf’s argument
plausible.

For the causal theory of knowledge to provide an objection to the possi-
bility of perceiving platonic entities, we would need a good reason to
think that, just as in most cases of perception, causal contact is required
to secure the right kind of knowledge- and reference-grounding presence-
to-consciousness to the affairs in question. But I do not think that’s espe-
cially obvious. In the ªrst place, it is not clear to me that all of the entities
to which a causal theorist is committed are things to which we stand in
causal relations, including causation and theories. Does causation itself
have causal powers? And do we really stand in causal relations with such
things as theories? True, we stand in causal relations with the signs and
symbols in whose terms theories are expressed, but so do the windows that
vibrate when we utter them and the sheets of paper on which we print
ihnen, without thereby standing in causal relations to theories. A theory
isn’t identical with this or that inscription or utterance. If I wad up a piece
of paper on which a theory is written and throw it in a ªre, I will not have

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

258

Experiments in Thought

incinerated a theory. We also stand in causal relations with our acts of
thinking theories, but these aren’t theories either; there are not as
many theories of Special Relativity as there are acts of thinking it. Fur-
thermore, the mental acts involved in thinking any theory whatsoever are
part of the subject matter of psychology, but not every theory whatsoever
is part of the subject matter of psychology. Absent an explanation of the
“mechanism” in virtue of which we stand in causal relations to causation
and to theories, I remain unconvinced that a causal theorist can consis-
tently claim to know or even think about the causal theory of knowledge
or reference.

In the second place, there is some reason for thinking that hallucinatory
and imaginative experiences qualify as counterexamples to the causal the-
ory. There is no actual instance of couch-hood, or even shape or color, Sei-
fore me when I hallucinate or imagine a couch, and insofar as causal rela-
tions hold only among actualities, I do not stand in any causal relations, bei
least at that time, with the properties revealed to me in those experiences.
And yet not only do I refer to them, but they are presented to me.

One response to this argument is that even if we do not bear causal rela-
tions to what shows up in hallucinatory experiences and acts of imagining,
we at least bear causal relations to actual instances of relevantly similar
types of objects. That might be right, but any defense along these lines
threatens the other premise of the anti-Platonist argument, namely that
we do not have causal contact with platonic objects. If what explains my
ability to hallucinate the color red is that I have previously been causally
related to the color red, then I am, schließlich, capable of being causally re-
lated to the color red.

Außerdem, even if we set aside the question of whether we causally
interact with platonic entities or not, seeing features or properties or uni-
versals is a thoroughly mundane occurrence in both ordinary perception
and actual experimentation. Suppose that I think, emptily, about the color
Rot. What kind of experience will fulªll my empty intentions towards the
color red? Here is one kind: I perceive a red tomato, and then turn my at-
tention to another red tomato, and then to another. In the course of this
series of acts, I perceive three different individuals. But I also perceive one
identical color, and perceive it as one identical color. My intention towards
the color red is fulªlled throughout a uniªed sequence of experiences.
And even though I was aware of some individual at each stage in that se-
quence, there was no individual such that I was aware of it across that
sequence. I was aware of an identical something, even though there was
no actual or intended individual that was identical. If I were only aware
von
individuals—whether individual tomatoes or individual property
instances—I would not have beheld one thing. Though I might, at this

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

259

Punkt, only have acquired the consciousness of a universal that is common
to this and this and this tomato, it is not difªcult to bring redness itself
into view. One way is to realize that this property common to the three to-
matoes can be instantiated by “an inªnity of possible single instances”
(Husserl 1977, P. 59) oder, more simply, that the “universal is not bound to
any particular actuality” (Husserl 1973, P. 329). Our understanding that
universals are not bound to particular actualities is manifested in such
trivialities as wanting to paint one’s kitchen red; we understand that the
color red is not bound to what is presently red, and that the present color
of one’s kitchen is not bound to it.

This itself is just one arbitrarily chosen example illustrating the perva-
siveness of our awareness of universals. “The truth,” says Husserl, “is that
all human beings see ‘ideas,’ ‘essences,’ and see them, so to speak, continu-
ously” (Husserl 1982, P. 41). Sometimes the particularity of the objects of
perception and thought becomes almost transparent, and they count in
our cognitive and practical economy only insofar as they bear the proper-
ties that really concern us. I don’t particularly care which $20 bill I receive from the ATM, as long as I receive a $20 bill. When I look up the mean-
ing of a word, I don’t care which token deªnition I encounter. I just want
to encounter a token, never mind which, of the relevant type. And when I
want to know whether heavy objects accelerate at the same rate as light
ones, I don’t care which individuals I drop from which tower. I only care
that they instantiate the properties whose relationships I want to discover.
In all of these examples, it matters that I encounter actual individuals—
I don’t want an imaginary $20 bill—but the individuals are only relevant
insofar as they instantiate the relevant properties, not in virtue of being
the very individuals that they are. Any other qualitatively identical indi-
vidual would have served just as well.

As the example of falling objects illustrates, the variables tracked and
varied in actual experiments are almost always properties and relations,
and experiments are typically construed as answering or raising questions
about the relations among those sorts of variables. Suppose we discover
that a 5 lb. weight—let’s call it “Timmy”—and a 10 lb. weight, “Jimmy,”
fall at the same rate in conditions C. The variables that this experiment, Wenn
it is a sensible one, is designed to track are not Timmy and Jimmy, aber die
properties that they instantiate. The knowledge we acquire on the basis of
this experience, darüber hinaus, is not conªned to Timmy and Jimmy, but ap-
plies to any qualitatively identical objects in identical circumstances, Und
we know this. Here is a patently absurd follow-up question to the experi-
ment: “Sure, Timmy and Jimmy fell at the same rate. But would two indi-
viduals exactly like Timmy and Jimmy also fall at the same rate in condi-
tions identical to C?” Experimentally answering this question is a waste of

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

260

Experiments in Thought

Zeit. If we tried out two different weights and they did not behave in the
same fashion, we would simply conclude that either those two individuals
were not qualitatively identical with Timmy and Jimmy, that the condi-
tions were not the same as those in C, or that we performed the experi-
ment incorrectly on one of the trials. The reason, darüber hinaus, is due to our
understanding that individuals do not fall under laws in virtue of being
the individuals that they are, but in virtue of having the properties they
have and standing in the relations that they stand—an insight that quite
obviously depends on our ability to grasp properties. There are several ver-
sions of the problem of induction which are indeed problematic. Das ist
not one of them. Determining whether qualitatively identical objects will
behave in the same way in qualitatively identical circumstances is not like
determining whether all crows are black on the basis of our knowledge of
all observed crows. It’s more like determining whether all crows which are
qualitatively identical to the black ones are black.

The laws and regularities that scientists attempt to discover, und in
whose terms the antics of individuals are explained, hold for individuals in
virtue of their properties and the relations they bear to other propertied
individuals. This holds in the cognitive sciences as well. According to
most scientiªc accounts of perception, it is properties rather than individ-
uals which are detected by our perceptual systems. Wann, zum Beispiel, Es
is claimed that people attuned to popular culture have a “Jennifer Aniston
cell,” it is clear that this cell does not detect Jennifer Aniston. It does not, In
standard conditions, ªre if and only if Jennifer Aniston is visually present.
It ªres when subjects are presented with pictures of Jennifer Aniston (Quian
Quiroga et al. 2005). No doubt it would also ªre in the visual presence of
the real Jennifer Aniston, or a twin of Jennifer Aniston, or a hallucination
of Jennifer Aniston. What it quite plainly detects is Jennifer Aniston-
esque features. In der Tat, our access to features or properties is precisely
what is not mysterious on most contemporary naturalistic theories of con-
tent. As Ruth Millikan puts it, “Failure to account for our capacity to rep-
resent individuals in language and thought has been, vielleicht, the most
serious failing common to contemporary naturalistic theories of content”
(Millikan 2002, P. 43). Informational relations are lawful relations, Und
“there are no natural laws just about individuals” (Millikan 2002, P. 35). ICH
will not try to determine whether Millikan’s own efforts to solve this
problem are successful. What is noteworthy is that it is a problem at all.
Our mundane perceptual experiences, as well as those underlying sci-
entiªc experiments, are saturated with reference to universals. Universals
show up in both thought and experience, and our cognitive achievements,
including those involved in carrying out empirical science, would be in-
comprehensible if they did not. What genuinely sets thought experiments

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

261

apart from actual experiments is not, Dann, that in the former case we are
aware of universals. What sets them apart is that thought experiments,
and the intuitive acts on which they are founded, have lost the connection
to actuality that is essential to actual experiments. By allowing us to fulªll
thoughts which are not rendered true or false by the actual existence
of this or that individual, and therefore by the actual instantiation of this
or that property, they allow us to discover relations that hold, or not,
among instances of the relevant universals in any possible world. If there is
a mystery involved in a priori insight, that mystery is not (just) our access
to universals—which, far from relying upon a special faculty in addi-
tion to thinking and perception, is built into them—but the ability to
“purify” our thoughts of any positing of actuality, to “treat the factual ac-
tuality of the single cases attained in variation as completely irrelevant”
(Husserl 1977, P. 55; also see Kasmier 2010). Even if we do not know just
how this ability works, the existence of the ability is beyond question. Es ist
routinely exercised by the producers and consumers of ªction, lucid
dreamers, savvy hallucinators, chess players of virtually all skills levels,
and those with imaginations.

VI.
To sum up, Dann, on the view defended here, actual experiments and
thought experiments are both genuinely experiments. Both are controlled
attempts to determine the relation between variables on the basis of acts of
epistemic fulªllment and, daher, on the basis of self-giving presenta-
tions of those variables or objects suitably related to them. The difference
is that actual experiments are directed at actual states of affairs, while
thought experiments are directed at ideal states of affairs. What distin-
guishes thought experiments from actual experiments is not that thought
experiments involve the awareness of universals. That’s true of actual ex-
periments too. What sets thought experiments apart is that they, und das
intuitive acts on which they are founded, have lost the connection to actu-
ality that is essential to actual experiments.

Verweise
Benacerraf, Paul. 1973. “Mathematical Truth.” Journal of Philosophy 70:

661–79.

Bishop, Michael. 1999. “Why Thought Experiments are Not Argu-

ments.” Philosophy of Science 66: 534–41.

Bokulich, Alisa. 2001. “Rethinking Thought Experiments.” Perspectives on

Wissenschaft 9: 285–307.

Braun, James Robert. 1991. The Laboratory of the Mind. New York:

Routledge.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

262

Experiments in Thought

Braun, James Robert. 2004. “Peeking Into Plato’s Heaven.” Philosophy of

Wissenschaft 71: 1126–38.

Cooper, Rachel. 2005. “Thought Experiments.” Metaphilosophy 36: 328–

47.

Gendler, Tamar Szabo. 2004. “Thought Experiments Rethought—and

Reperceived.” Philosophy of Science 71: 1152–63.

Häggqvist, Sören. 2009. “A Model for Thought Experiments.” Canadian

Journal of Philosophy 39: 55–76.

Hopp, Walter. 2011. Perception and Knowledge. Cambridge: Cambridge

Universitätsverlag.

Husserl, Edmund. 1969. Formal and Transcendental Logic. Translated by D.

Cairns. Den Haag: Martinus Nijhoff.

Husserl, Edmund 1970. Logical Investigations, 2 vols. Translated by J. N.

Findlay. London: Routledge & Kegan Paul.

Husserl, Edmund 1973. Experience and Judgment. Edited by L. Landgrebe.
Translated by J. S. Churchill and K. Ameriks. Evanston, IL: Norden-
western University Press.

Husserl, Edmund 1977. Phenomenological Psychology. Translated by

J. Scanlon. Den Haag: Martinus Nijhoff.

Husserl, Edmund. 1982. Ideas Pertaining to a Pure Phenomenology and to a
Phenomenological Philosophy. First Book: General Introduction to a Pure Phe-
nomenology. Translated by F. Kersten. Den Haag: Martinus Nijhoff.
Husserl, Edmund. 2001. Analyses Concerning Passive and Active Synthesis.
Translated by A. J. Steinbock. Boston: Kluwer Academic Publishers.
Johnston, Markieren. 2004. “The Obscure Object of Hallucination.” Philosoph-

ical Studies 120: 113–83.

Kasmier, David. 2010. “A Defense of Husserl’s Method of Free Variation.”
In Epistemology, Archaeology, Ethics: Current Investigations of Husserl’s Cor-
pus. Edited by Pol Vandevelde and Sebastian Luft. New York: Contin-
uum.

McAllister, James W. 1996. “The Evidential Signiªcance in Thought Ex-
periments in Science.” Studies in the History and Philosophy of Science 27:
233–50.

Millikan, Ruth Garrett 2004. The Varieties of Meaning. Cambridge: Der

MIT Press.

Norton, John. 1996. “Are Thought Experiments Just What You

Thought?” Canadian Journal of Philosophy 26: 333–66.

Peacocke, Christopher. 1992. A Study of Concepts. Cambridge, Masse.: Der

MIT Press.

Quian Quiroga, R., L. Reddy, G. Kreiman, C. Koch, and I. Fried. 2005.
“Invariant Visual Representation by Single Neurons in the Human
Brain.” Nature 435: 1102–07.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3

Perspektiven auf die Wissenschaft

263

Sorensen, Roy. 1992. Thought Experiments. Oxford: Oxford University

Drücken Sie.

Wiltsche, Harald. 2012. “Platonism and Thought Experiments: Phenom-
enology, How Deep is Your Love?” Paper presented at the Workshop
“Thought Experiments in Science: Four Blind Spots,” University of To-
ronto, Marsch 2012.

D
Ö
w
N
Ö
A
D
e
D

F
R
Ö
M
H

T
T

:
/
/

D
ich
R
e
C
T
.

ich
T
.

e
D
u
P
Ö
S
C
/
A
R
T
ich
C
e
–
P
D

F
/

2
2
2
2
4
2
1
7
8
9
8
9
6
P
Ö
S
C
_
A
_
0
0
1
2
9
P
D

B
j
G
u
e
S
T

Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
PDF Herunterladen