Science and Experience/ - Recherche en IA spécialisée au MIT

Science and Experience/
Science of Experience:
Gestalt Psychology and
the Anti-Metaphysical
Project of the Aufbau

Uljana Feest
Technische Universität Berlin

This paper investigates the way in which Rudolf Carnap drew on Gestalt
psychological notions when deªning the basic elements of his constitutional
système. I argue that while Carnap’s conceptualization of basic experience was
compatible with ideas articulated by members of the Berlin/Frankfurt school
of Gestalt psychology, his formal analysis of the relationship between two ba-
sic experiences (“recollection of similarity”) was not. This is consistent, given
that Carnap’s aim was to provide a uniªed reconstruction of scientiªc knowl-
bord, as opposed to the mental processes by which we gain knowledge about the
monde. It is this last point that put him in marked contrast to some of the
older epistemological literature, which he cited when pointing to the complex
character of basic experience. While this literature had the explicit goal of
overcoming metaphysical presuppositions by means of an analysis of conscious-
ness, Carnap viewed these attempts as still carrying metaphysical baggage.
By choosing the autopsychological basis, he expressed his intellectual depth to
their antimetaphysical impetus. By insisting on the metaphysical neutrality
of his system, he emphasized that he was carrying out a project in which they
had not succeeded.

1. Introduction
In his 1928 livre, Der Logische Aufbau der Welt, Rudolf Carnap presented
what he called a “constructional system” (Carnap 1967). The aim of this
system was to demonstrate that all of our scientiªc concepts are logically
derivable from more “basic” concepts in a hierarchical fashion. Carnap em-

This article was researched and written while the author was a research scholar at the Max
Planck Institute for the History of Science. The author would like to thank Brigitta Arden
at the University of Pittsburgh Archives for Scientiªc Philosophy, as well as Hans-Jörg
Rheinberger and the Max Planck Institute for their support of this research.

Perspectives on Science 2007, vol. 15, Non. 1
©2007 by The Massachusetts Institute of Technology

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

phasized that different bases for such a system were in principle conceiv-
capable (§§ 62, 63, 67). En effet, he himself had previously made an unpub-
lished attempt to construct a system on a physical basis (see Pincock 2005),
and as is well known, he was later to return to a physicalist position (Car-
nap 1931b). Cependant, in the Aufbau, Carnap chose what he termed an
“autopsychological basis”, c'est à dire., the “mental objects that belong to only one
subject” (Carnap 1967 [1928], § 63). He further limited the domain of
such objects by specifying that what he meant were elements of conscious-
ness, taken from the stream of experience. He then settled on the term “the
given”, which he deemed to be the most neutral of terms. (Carnap, op.
cit., § 64)

Having stated his choice of the autopsychological basis for his constitu-
tional system, Carnap identiªed a possible danger of his approach, c'est à dire.,
that of being caught in subjectivism. This would have been contrary to his
explicit aim to obtain “an intersubjective, objective world, which can be
conceptually comprehended and which is identical for all observers”
(Carnap 1967 [1928], § 2). In response to this problem he asserted that
while no two experiences can be compared with respect to their material
(the immediate quality of the given), all streams of experience share cer-
tain structural features (§66). He explained this claim by repeating an ear-
lier assertion, c'est à dire., that all science, by its very nature, aims at identifying
structures (§§ 11–16). Based on this assertion, it seems that Carnap’s choice
of individual subjective experience as the basis of his system relied on the
assumption that it is in principle possible to have a science of experience
in the sense speciªed by him, c'est à dire., one that is capable of identifying a struc-
ture of experience that is shared by different experiencing subjects.

In this paper, I ask whether this assumption was backed up by psycho-
logical research of the time with which Carnap might have been familiar.
Ce faisant, I will follow up the suggestion that Carnap’s conceptualiza-
tion of the autopsychological basis was informed by Gestalt psychology
(par exemple., Richardson 1998, Majer 2003). I will show that we need to distin-
guish between particular schools of scientiªc Gestalt psychology on the one
main, and the more general epistemological discourse, in which references to
the complex character of immediate experience could be found. Plus loin, je
will argue that it is helpful to distinguish between two senses in which
Carnap might have taken an interest in a scientiªc Gestalt psychology i.e.,
(1) Gestalt psychology as exemplifying a science whose basic concepts
Carnap wanted to ªt into a uniªed picture of all scientiªc knowledge, et (2)
Gestalt psychology as supplying basic facts about human perception,
which allow us to provide a rational reconstruction of human knowledge ac-
quisition, starting in its phenomenal basis. The two possibilities are not
mutually exclusive, and Carnap indeed occasionally justiªed his choice of

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

the autopsychological basis by appeal to the actual “epistemic order”.
Cependant, I will argue that scientiªc Gestalt psychology only backed up
Carnap’s assertion that basic experience was structured. When Carnap’s
provided his own analysis of the structure of experience, cependant, he drew
on a notion of recollection of similarity, which was entirely formal and which
was in fundamental contradiction to the Gestalt psychological (empirical)
analysis of that same notion. This will prompt me to inquire into the ori-
gins of Carnap’s “psychologistic” language, given that his aim was not a
description of mental processes. In investigating this question, I will turn
to some of the philosophical writers that Carnap cites in support of his no-
tion of experience, all of which were proponents of philosophical move-
ments that attempted to overcome metaphysical assumptions by means of an
analysis of consciousness. I will argue that Carnap’s choice of an autopsycho-
logical basis and of terms like “recollection of similarity” should be placed
into this context. It signiªes a certain allegiance to their projects, même
though he judged them as not having succeeded in overcoming metaphys-
ics. It also suggests who might have been the real targets of his repeated
insistence that his own constitutional system was metaphysically neutral.

2. Experience in Its Totality and Unity
When discussing the question of how to conceptualize the phenomenal
basis of the system, Carnap brieºy considers Mach’s elements of sensation,
but argues that such elements are not actually given in, but rather an ab-
straction from, experience. They therefore cannot function as “elementary
experiences”. Plutôt, Carnap opts for a notion of “experiences themselves
in their totality and undivided unity” (§ 67), citing, entre autres, mem-
bers of the Berlin/Frankfurt school of Gestalt psychology in support of
this notion (Köhler 1925a, Wertheimer 1925).

2.1 The Complexity of Immediate Experience
Proponents of the Berlin/Frankfurt school of Gestalt psychology had
pointed to the fact that our sensory experiences (par exemple., when perceiving a
geometrical ªgure or a chord) can have the phenomenal quality of an un-
divided unity even though the experience has an internal structure or shape
(“Gestalt”), due to the complex character of the stimulus (par exemple., when the
stimulus is a geometrical ªgure or a musical chord). Ainsi, any talk of ele-
ments of such an experience is the result of a retrospective abstraction from
the actual experience. It needs to be emphasized, though, that there is a
certain sense in which Gestalt perceptions exemplify a more general phe-
nomenon. Even when the stimulus is not complex (say, when it is only one
line, or one tone), we can distinguish between different aspects of our expe-
rience of the stimulus. Par exemple, in describing our phenomenal experi-

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

ence of a simple line, we can distinguish between color, width, and orien-
tation. De même, in describing our phenomenal experience of a tone, nous
can distinguish between pitch and intensity. Any one of these features is
always going to be accompanied by the others. Ainsi, if we are talking
about the pitch of a tone in isolation, we are necessarily abstracting away
from the experience of the tone. This fact had been discussed in the philo-
sophical and psychological literature in the decades prior to Carnap’s
Aufbau (par exemple., Cornelius 1900a, Lipps 1900).1

Be this as it may, prima facie, it seems clear why Gestalt psychology
might have been attractive for Carnap’s project: This psychological school
was offering a scientiªc conception of subjective experience, according to
which phenomenal experience is structured in accordance with laws,
which describe the functional relationship between types of stimulus
conªgurations and types of Gestalt experiences. These laws can be deter-
mined experimentally and hold across different individuals. Alan Richard-
son has suggested that “Carnap [. . .] looks to [. . .] Gestalt psychology
[. . .] for an account of the structure of human experience in human agents
and uses this as the basis from which to start his constitutional system”
(Richardson 1998, 9). This assessment is plausible, but it is open to sev-
eral possible interpretations. D'abord, it might mean that Carnap looked to
Gestalt psychology to back up his contention that basic experience was
structured and that it was possible to form scientiªc concepts of this struc-
ture. Deuxième, it might mean that Carnap looked to Gestalt psychology for
an account of how experience was structured, hoping to reconstruct the
process of knowledge acquisition from this basis. Dans ce qui suit, I will
argue that there are good reasons to agree with the ªrst interpretation,
whereas the second interpretation is most certainly false. Let me explain
the difference between the two readings by aligning them with two possi-
ble interpretations of the overall project of the Aufbau.

2.2 The Place of Immediate Experience in the Aufbau: Deux
Readings
According to the ªrst reading, an important aim of the Aufbau was to re-
construct the body of our scientiªc knowledge in a uniªed fashion (Carnap
1967 [1928], § 2), thereby overcoming the perceived fragmentation of the
sciences. In this vein, Christopher Pincock (2003) has argued that by fo-
cusing on the construction of concepts, Carnap wanted to overcome the fact

1. Cornelius proposed a theory of abstraction to account for the way in which we form
concepts of such aspects of an experience (Cornelius 1900b). In this and other writings,
Cornelius also addressed the question of Gestalt-experience. We will return to Cornelius
below.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

that Schlick and Reichenbach, with their focus on axiomatic systems, had
not succeeded in giving an account of the unity of knowledge. Now, si
Carnap’s aim was to reconstruct scientiªc knowledge, it follows that “(w)hat
our best scientiªc theories have as their objects determines what needs to
be constructed” (Pincock 2005, 528). This suggests that for a construction
system that uses basic experiences, our best psychological theories need to
be consulted for an adequate notion of experience. It is in this sense, I believe,
that Carnap’s approach can be described as “naturalistic” (Pincock op.
cit.). Given that members of the Berlin/Frankfurt school of Gestalt psy-
chology had their prime time in the 1920s (Ash 1995), it seems only nat-
ural to assume that Carnap would have considered their ideas as “the best
scientiªc theories” about human experience. En outre, Carnap was at
least indirectly acquainted with members of the Berlin/Frankfurt school of
Gestalt psychology.2

Cependant, there is a different sense (and correspondingly, a different
reading of the Aufbau), in which Carnap’s approach might be characterized
as naturalistic. According to this second reading, it was his aim to recon-
struct the actual processes of cognition in accordance with the scientiªc psy-
chology of the day. Carnap’s remark that his choice of the phenomenal ba-
sis was motivated by the desire to capture the “epistemic primacy” of
experiences might be read as pointing in that direction. Cependant, he also
makes it clear that this was merely intended as a rational reconstruction of
the “real” cognitive processes (Carnap 1967 [1928]. § 54). As I will show
in section 4 below, Carnap’s account of how experience is structured was in
fact contrary to that of Gestalt psychology. This incompatibility of natural-
istic description and rational reconstruction speaks against an interpreta-
tion of the Aufbau as attempting to follow psychology in providing an ac-
count of the real cognitive processes of knowledge acquisition. Ainsi, je
argue for the ªrst of my two proposed readings of the role of psychology in
the Aufbau. In sections 5–7 of this paper, I will offer an additional line of
le
evidence for
autopsychological basis was a respectful nod towards previous philoso-

suggesting that Carnap’s choice of

reading,

2. He had met both Reichenbach and Reichenbach’s friend, the philosopher and Ge-
stalt psychologist Kurt Lewin, at a conference in Erlangen in 1923 (see Carnap 1963, 14).
Around the same time, there were efforts to found a journal for scientiªc philosophy
(which was eventually to appear under the name Erkenntnis in 1931, cf. Bernhard 2002).
As documented by extensive correspondence between Reichenbach and Köhler, Reichen-
bach wanted to Köhler to participate (see Ash 1994). À son tour, Reichenbach informed
Carnap of his correspondence with Köhler, which was not entirely free of tensions
(Reichenbach/Carnap correspondence, Archive for Scientiªc Psychology, Univ. of Pitts-
burgh). Köhler attended meetings of Reichenbach’s “Gesellschaft für wissenschaftliche
Philosophie”, founded in 1927 (cf. Danneberg & Schernus 1994, Hoffmann 1994)

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

phers who had attempted to build their philosophical systems on descrip-
tions of the actual processes of cognition, but whose project Carnap re-
garded as failed.

3. Fundamentals of Gestalt Psychology
Above I used the expression “Berlin/Frankfurt school of Gestalt psychol-
ogy”. This name is usually used to refer to Max Wertheimer, Wolfgang
Köhler, Kurt Koffka, and the slightly younger Kurt Lewin, all of which
were students of the Berlin psychologist/philosopher Carl Stumpf in the
ªrst decade of the 20th century. In the early 1910s, the former three col-
laborated on a number of experiments in Frankfurt, and after WWI, les deux
Köhler and Wertheimer had appointments in Berlin (Jäger 2003, King &
Wertheimer 1995, Ash 1995).3 As already indicated, members of the
Berlin/Frankfurt school were not alone in emphasizing that immediate ex-
perience cannot be analyzed into parts. Nor were they alone in investigat-
ing Gestalt phenomena.4 In this section, I will outline the basic tenets of
the Berlin/Frankfurt’s school’s position by contrasting them with those of
two philosophers who had previously written about Gestalt phenomena.
While the story will be presented chronologically, it is not intended as an
historical account, but is only supposed to bring out some basic analytical
distinctions.

3.1 Experience Has No Parts
With their emphasis on the primary character of structured experience, mem-
bers of the Berlin/Frankfurt school of Gestalt psychologists entered into a
thicket of debates that had been going on within the shifting ªelds of
epistemology, philosophical psychology and experimental psychology and
physiology since the middle of the 19th century. Two aspects can be sepa-
rated here for our purposes. (1) One dealt with the question of the rela-
tionship between immediate and more mediated contents of consciousness,
where the former were also often referred to as “sensations”, and were usu-
ally viewed as closely tied to physiological states. This gave rise to ques-
tions about the nature of complex sensations, and about the mechanisms
or mental processes by which sensations are combined such as to result in
perceptions or judgments.5 (2) The other aspect concerned the question of

3. The fact that Carnap does not cite Kurt Koffka is probably due to the fact that
Koffka had held a chair in Giessen since 1912 et, hence, was not part of the intellectual
circle in Berlin, of which Carnap was aware through Reichenbach.

4. Other prominent schools of Gestalt psychology were (1) the Graz School of
Ehrenfels’s friend and teacher Alexius Meinong (see Fabian 1997) and the Leipzig school of
Krüger (see Scheerer 1931).

5. Helmholtz’s writings on physiological optics may serve as a classical point of refer-

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

why two experiences (of hearing a melody, of seeing a geometrical ªgure)
can be phenomenally quite similar even though few or none of the “parts”
of the experience are the same. While my focus in this section will be on
the second of these two aspects, it is important to recognize that both
problems arose from the same underlying assumption, namely that the con-
tent of a sensory experiences can be analyzed into elements, where those elements
have the character of immediacy and correspond in a one-to-one fashion to
elements of the external stimulus.

The problem just identiªed, how to account for the similarity of expe-
riences despite their lack of common parts, is of course the problem of Ge-
stalt perception. Two milestones of this debate were Ernst Mach’s book
Analysis of Sensations (1996 [1886, 1903]), and Ehrenfels’s paper: “Ueber
‘Gestaltqualitäten’” (1890).6 In the ªrst (1886) edition of his book (lequel
was constantly revised and had doubled in size by the ªfth edition) Mach
laid the foundation for his psycho-physical parallelism, which stated that
every state of consciousness corresponded to a physiological state, et
whenever it was possible to analyze a state of consciousness into elements,
these elements, à son tour, corresponded to elements of a physiological pro-
cess (Mach 1886, 27/8). This led Mach to conclude that whenever we have
two similar states of consciousness, they have to share in common at least
some elements, and that for every such element there was also a correspond-
ing physiological process.7 For example, the fact that I recognize two dif-
ferently colored ªgures (“Gestalten”) as similar shows that the state of
consciousness in question can be analyzed into a color and a spatial com-
ponent, and that that the two sensations are similar, because they share in
common the spatial element of sensation (Mach 1886, 28/9). According to
Mach, this also explained why the same geometrical ªgure causes in us a

ence here (par exemple., 1910 [1867]). In that work, he singles out the material delivered by our
senses as providing the basis for the formation of representations. His discussion makes it
clear that the issue of sensations brought to the fore not only issues about the relationship
between body and mind, but also between the scientiªc ªelds of physiology and psychol-
ogy. In this context, it is interesting to note that Carnap, throughout the Aufbau, does not
use the term sensation (“Empªndung”) to refer to his basic objects, but rather the term, ex-
expérience (“Erleben”), even though he is referring to an aspect of experience (immediacy, le
given) that others, including writers like Wertheimer usually associated with sensations.

6. For a discussion about the historical relationship between the ideas of Mach,
Ehrenfels, and Wertheimer (who was brieºy a student of Ehrenfels’s in Prague), see Ley
1994.

7. This thought was expressed most clearly in the third edition of 1902 (55/6), lequel
in the English translation reads that it is “almost a self-evident supposition, that similarity
must be founded on a partial likeness or identity, and that consequently, were sensations
étaient semblables, we had to look for their common identical constituents and for the corre-
sponding physiological processes” (Mach 1996, 69)

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

different Gestalt experience, depending on how the ªgure is rotated (op.
cit., 44 ff.).

Four years later, Christian von Ehrenfels (1890) challenged the very
idea that it is possible to analyze a Gestalt sensation into elements, depuis
this notion (for Ehrenfels) rested on the assumption that the Gestalt sensa-
tion is the sum of these elements. This assumption, if it were true, would
imply that the more elements two sensations share in common, plus
similar they will be. Ehrenfels tried to refute this assumption by showing
(un) that two stimulus complexes (par exemple., two melodies) can prompt very sim-
ilar sensations even if they don’t share a single element of sensation, et
(b) two sensations can be very different even though they share many ele-
ments. For Ehrenfels, this showed that a Gestalt sensation is more than
the sum of its elements; it exists in addition to the elements.

Notice that while Ehrenfels challenged the idea that a complex sensa-
tion is the sum of its parts, he did not challenge the idea that a complex
sensation consits of parts. Plutôt, the Gestalt character of such sensations is
a quality of those parts (hence, “Gestalt qualities”), though not reducible
to them. This assumption was to be questioned by members of the Berlin-
Frankfurt school. Par exemple, as Wertheimer demonstrated in his famous
experiments of 1910, when we see two stationary lights—ºashing in short
succession—as one moving light, we do not ªrst have a sensation of one
stationary light, then of another, and ªnally of a moving light. Plutôt, nous
simply have a sensation of a moving light (see Wertheimer 1912). De même,
while it may be possible to detect individual tones when paying attention
to our sensation of a chord, the sensation of the chord is not built up out of
the sensation of its tones.

3.2 The Scientiªc Study of the Structure of Experience
The previous paragraph shows that the results of the Berlin/Frankfurt
school have to be located in a particular theoretical debate. I would like to
emphasize that their empirical work was vital to their arguments. Ils
saw themselves not only as (1) making their contribution to the above
questions on empirical grounds, mais (2) in doing so challenged some fun-
damental assumptions about the nature of scientiªc method. They con-
ducted rigorous empirical studies in which they varied stimulus condi-
tion, thereby trying to determine objective conditions under which
certain Gestalt phenomena were experienced. Par conséquent, they showed
that the ways in which stimuli were related to each other have an impact on
what is experienced. Par exemple, the way we experience a particular
chord is determined not by the physical properties of the individual tones,
but by the way in which the physical stimuli are related to each other, c'est à dire.,
by the intervals between the tones. De plus, and this was central, ils

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

questioned the assumption that for every element of a stimulus there is a
corresponding element of sensation, which remains the same, regardless of
the other stimuli that it is presented with (Köhler, 1913, referred to this
as the “constancy assumption”). Par exemple, they argued that the way we
experience an individual tone that is part of a chord is determined by, et
secondary to, the experience of the chord.

By providing empirical evidence for these claims, members of the
Berlin/Frankfurt school took themselves to show not only that it is unnec-
essary to assume the existence of elements of sensation, but that such an
assumption, which was the expression of a particular, atomistic, concep-
tion of science, had been positively detrimental to the empirical study of
sensation.8 According to Wertheimer and his colleagues, previous genera-
tions of empirical psychologists had attempted to account for a pseudo-
problem (how we come to have complex experiences, when the basic mate-
rial is atomistic). This pseudoproblem had been generated by a failure to
pay adequate attention to our own phenomenal experience, c'est à dire., a failure to
notice that the basic material is not atomistic.

4. Structured Experience and Recollection of Similarity
In § 7 of the Aufbau, Carnap introduces his “Grundgegenstände”, lequel
are comprised of basic elements (experiences) and basic relations. The basic
relation is the more important of the two (§ 10, § 61), allowing Carnap to
give descriptions of the elements in purely structural terms (structure de-
scriptions being a special case of relation descriptions, § 11). Carnap refers
to this basic relation as “recollection of similarity”.

4.1 Carnap’s Basic Experiences
In support of his claim that Mach’s elements of sensations are not really
given in experience (§ 67), Carnap cited a broad spectrum of epistemolog-
ical writings (y compris, Moritz Schlick, Wilhelm Schuppe, Hans Cornel-
ius, Heinrich Gomperz, Friedrich Nietzsche), before mentioning that re-
lated views had been developed by Gestalttheory (Köhler, Wertheimer),
and had proven to be methodologically fruitful within psychology. I will
argue below that a full understanding of Carnap’s choice and conceptual-
ization of the autopsychological basis requires that we take seriously the
philosophers he cites. De plus, the Berlin/Frankfurt school, lui-même, has to
be historically situated before the background of this older epistemolog-
ical context. This does not mean that Carnap’s reference to Köhler and
Wertheimer is not signiªcant in its own right. Par exemple, a remark Car-

8. See Mayer 2004 for an overview of atomistic and associationist tendencies in philoso-

phy around 1900.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

nap makes in § 71 make it quite clear that his notion of “basic experience”
is that of the Berlin/Frankfurt school (more speciªcally: their rejection of
Ehrenfels’s version of Gestalt quality). Là, he writes that even though
we think that we hear the c in the c-e-g chord, this apparent sensation of
the c is a quasi-element, not a real element. “Otherwise, one would come
to the conclusion (which has indeed sometimes been maintained) that the
chord c-e-g consists of the individual tones c, e, g, et, in addition to
eux, of something which comprises the actual character of the chord”
(Carnap 1967 [1928]).

Clairement, if basic experience has no parts, it cannot be analyzed into its
parties. Cependant, prima facie the notion of “structure” suggests the existence
of parts, as was also recognized by Wolfgang Köhler who pointed to the
issue of “how such a formation as a whole can have a peculiar phenomenal
structure . . . which cannot be derived from its so-called parts” (Köhler
1925b, 381, my translation). As we saw, Köhler and his colleagues an-
swered this by conceptualizing basic experience in a way that allowed
them to describe experience as structured, without thereby conceding that
it is comprised of elements. Contrary to that, Carnap proposed to capture
the structure of experience by means of what he called a “quasi-analysis”,
which proceeds as if experience had parts (“quasi-parts”). Dans ce qui suit,
I will brieºy contrast the quasi-analytical and the Gestalt-psychological
account of the structure of experience, in order to highlight the senses in
which Carnap was—and was not—expressing Gestalt psychological ideas.

4.2 The Basic Relation
Carnap’s tool of analysis is relational logic, as developed by Russell and
Whitehead in their Principia Mathematica (Whitehead & Russell 1925
[1910–1913]). In order to apply this analysis, Carnap required a basic re-
lation. He called the basic relation of his choice “recollection of similar-
ity”. For Carnap this expression referred to a relation between two experi-
ences, a relation that was presupposed as a primitive component of
Carnap’s constitutional system. While the expression “recollection of
similarity”—as the basic relation of the constitutional system—was a
purely formal notion, Carnap made certain assumptions as to the question
by virtue of what two experiences may be said to be related in such a way.
These assumptions appear to be psychological assumptions, which are sur-
prisingly similar to the above-mentioned Machian framework, according
to which when we have two similar sensations, it is reasonable to assume
that they can each be analyzed into similar parts (Mach 1996). Cependant,
as we saw, this was precisely the assumption that had been questioned by
members of the Berlin/Frankfurt school of Gestalt psychology. According
to Gestalt psychology, two experiences are similar, because they have the

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

Recollection of similarity according to Carnap and the Berlin/Frankfurt school of
Gestalt Psychology (based on the example provided in Carnap’s manuscript,
“Quasizerlegung”). The lines above the chart group together those musical units
that are similar according to Carnap’s analysis. The lines below the chart group
together those musical units that are similar according to Gestalt psychology. je
am not distinguishing between major and minor chords here.

same Gestalt. And to have the same Gestalt is not the same as having the
same elements—or quasi-elements, d'ailleurs.

Let me brieºy illustrate this by way of an example of quasi analysis,
which Carnap offers in an unpublished manuscript from 1922/3 (“Quasi-
zerlegung”), one of two prior versions of the Aufbau. Là, he introduces
12 musical “elements” (individual tones and chords)9, which he calls h-t,
to distinguish them clearly from the names of the notes (c, d, e, F, g, un)
that occur in them: d[h], d/f/a[je], c/e/g[k], c/e[je], f/a[m], d/f[n], c/e/a[o], c/
F[p], c[q], d/a[r], g[s], c/g[t].

In the process of quasi-analysis, (je) pairs of elements are formed on the
basis of similarity, where by “similarity”, Carnap means “identity of at
least one partial tone” (p. 8), (ii) similarity classes are formed, c'est à dire., pour
every element, the class of elements to which it stands in a relation of sim-
ilarity, et (iii) and classes of elements that share a particular attribute,
thereby deªning the quasi-part. Now, if we compare the experiences that
are—on Carnap’s analysis—grouped together as similar, with those that
are similar according to the Gestalt psychological analysis, it is quite clear
that the two diverge profoundly. Carnap conceptualized similarity of expe-
rience as “matching with respect to parts of experience”, whereas for
Gestalt psychologists two experiences are similar if the two stimulus con-
ªgurations are similar in relevant respects: Two major triads would be rec-
ognized as similar even if they share no common note; a major and a minor

9. Bien sûr, the term, “element”, is a little misleading. Presumably, Carnap chooses it
to indicate that they are the smallest units of experience, as opposed to the quasi-elements
that are the result of the analysis.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

triad would be experienced as very dissimilar, even if only one note differs
slightly.

5. The Analysis of Consciousness as Weapon Against Metaphysics
I would like to emphasize that the point of the above analysis is not to sug-
gest that Carnap’s usage of Gestalt ideas is inconsistent. Quite on the con-
trary, if we recognize that his aim was to provide reconstruction of the
unity of scientiªc concepts, it is clear why he might have been interested in
the Berlin/Frankfurt conception of basic experience, without thereby be-
ing committed to following their empirical account of the way in which
such experiences come about, or what makes two experiences similar. Il
merely needed Gestalt psychology in support of his contention that con-
cepts of immediate experience can be scientiªc (c'est à dire., structure) concepts.
Carnap was now free to subject the basic experiences to a relational analy-
sis. En outre, his methodological solipsism immunized him against the
criticism that he didn’t take into account the way in which structured ex-
perience is functionally dependent on the structure of the stimulus com-
plex, as members of the Berlin/Frankfurt school had shown. He conªden-
tially stated that, in close analogy to Husserl’s bracketing of the external
monde, he was starting with a phenomenal analysis of the content of con-
sciousness, neglecting, for methodological purposes not only the existence of
an external world, but also the fact that the very concept of a subject of ex-
perience had not yet been constructed within his system.10

Having pointed out that there is no internal inconsistency in Carnap’s
project, we might nonetheless wonder why Carnap used such a psycholog-
ical expression (“recollection”) to describe a relationship that was, for his
purposes, purely formal, c'est à dire., not intended to capture any real mental pro-
cesses. That this was not accidental becomes clear when we look at an ear-
lier manuscript of Carnap’s Aufbau, in which the psychological language is
much more pronounced (Carnap 1922). This manuscript, which is enti-
tled “Vom Chaos zur Wirklichkeit” (“From Chaos to Reality”), has an
added handwritten remark (by Carnap himself), which states that this is
the core of the Aufbau. Carnap starts out by remarking on a certain ªction
that prevails in many epistemological writings at the time. According to
this ªction, we face a world of chaos, and it is the task of epistemology to
show how we arrive at an orderly world. En fait, he claims, we always ex-
perience the world as already ordered and meaningful. While Carnap
doesn’t cite any references here, I would like to argue that this insight is

10. For an analysis of the ways in which Carnap’s project is similar to Husserl’s, voir
Mayer 1991. Malheureusement, a more in-depth historical analysis of the two projects cannot
be provided in this article.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

informed by a wide array of literature that was already available to him.
Speciªcally, it was informed by some of the philosophers that he later
cited in § 67 of the Aufbau. While it would be historically incorrect to say
that these are Gestalt-psychological writings, I hope to have indicated above
that the phenomena of complex immediate experience and Gestalt percep-
tion were widely known and discussed both amongst empirical psycholo-
gists and epistemologists around the turn of the 19th/20th century. Par
the time the Berlin/Frankfurt school appeared on the scene, there were (comme
we have seen) not only several individuals and groups of empirical psy-
chologists, who offered competing accounts of the phenomenon in ques-
tion, but the issue was also discussed by more traditional philosophers. je
would like to suggest that we can gain additional insight into Carnap’s
project by inquiring (1) what motivated these philosophers to take an in-
terest in phenomena of immediate complex experience, et (2) which as-
pects of this motivation was shared by Carnap.

With respect to the ªrst question, these different philosophers may
generally be said to have shared a certain antimetaphysical sentiment, com-
bined with the assumption that grand metaphysical projects should be re-
placed by detailed accounts of how the world actually presents itself to us in our
consciousness. The contrast was sometimes expressed as one between a dog-
matic vs. an empirical philosophy (par exemple., Cornelius 1903), where the former
were the “big systems” of traditional philosophy (par exemple., Hegel, Fichte,
Schelling), whereas the latter was meant to be less speculative and more
true to a detailed description of “the facts”. Bien sûr, the bigger issue
looming in the background of these 19th century debates was what were
aims and methods of philosophy. A particularly provocative answer had
been provided by Franz Brentano in his 1866 habilitation thesis, où il
stated that the true method of philosophy is that of the sciences (voir
Münch 1997).11

It is important to note, though, that those philosophers who proposed a
scientiªc analysis of the mind were by no means all in favor of experimental
methods in psychology. Two philosophers who exempliªed the different ap-
proaches (empirical and non-empirical) to the analysis of consciousness
were Carl Stumpf and Edmund Husserl (both of whom, like Meinong,
were students of Brentano’s; see Stumpf, 1891; Münch 2003/4). For exam-
ple, the philosopher Hans Cornelius, to whose 1903 travail, Einleitung in die
Philosophie, Carnap refers in § 67 of the Aufbau, had previously published a
livre, entitled, Psychologie als Erfahrungswissenschaft, which had the explicit

11. For background regarding the 19th century development of positivist and idealist
versions of neo-Kantian schools of philosophy within the German-language philosophical
landscape, see Köhnke (1985).

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

aim of laying the epistemological foundation for “a purely empirical the-
ory of mental facts, free of any metaphysical presuppositions” (Cornelius
1897, IIV, my translation). But in this book he draws on assumptions
about experience that do not appear to be the results of experimental anal-
yses. Autrement dit, the call for an empirical psychology that was guided
by an anti-metaphysical sentiment did not necessarily imply the use of ex-
perimental methods.12

The tension between proponents of an experimental vs. non-experi-
mental analysis of consciousness led to increasing calls for an institutional
separation of empirical psychology from philosophy (Kusch 1995). By the
time Carnap completed his ªrst major philosophical work, Der logische
Aufbau der Welt, the institutional separation of psychology from philoso-
phy had largely taken place. Carl Stumpf perhaps has to be seen as a tran-
sitional ªgure, as he was still ªrmly rooted in a philosophical tradition,
yet was instrumental in instituting an experimental psychology in Berlin
between the 1890s and his retirement in 1920 (Sprung & Sprung 1995,
2003). The same can be said, though to a lesser extent, of his students,
Wertheimer and Köhler. Néanmoins, I believe that if we want to do full
justice to what Carnap was doing with his phenomenal basis, we need to
take seriously his references to the older philosophical literature, c'est à dire., a lit-
erature that viewed the (largely introspective, non-experimental) analyse
of consciousness as a weapon against metaphysics. In arguing this, I do not
mean to imply that Carnap was agreeing with every aspect of that earlier
agenda. Quite on the contrary, his ideas about alternative construction sys-
tems make it quite clear that he did not view the phenomenal basis as inte-
gral to a critique of metaphysics. En outre, he believed that these
older attempts had not succeeded in overcoming metaphysics. Cependant,
he shared their conviction that metaphysics had to be combated by means
of a more scientiªc philosophy, and by basing his system in the Aufbau on
an autopsychological basis, he demonstrated that his own version of a
scientiªc philosophy was at least compatible with their focus on immedi-
ate experience, while not running into some of their problems.

6. Formal and Psychological Notions of “Recollection of Similarity”
As we saw above, Carnap’s reconstruction of what makes two experiences
similar bears some resemblance to Mach’s 1886 analyse. Before returning
to Carnap’s relationship to phenomenalism (in section 7 below), I will now
turn to his choice of the term, “recollection of similarity”. I will argue that
that was taken from the work of Hans Cornelius. While the term, for Cor-

12. This was also very clearly expressed by Dilthey’s views about psychology (Dilthey

1894, Feest 2007).

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

nelius, still denoted a psychological (as opposed to Carnap’s formal) concept,
I will show that nonetheless the two philosophers shared some philosophi-
cal intuitions.

6.1 Cornelius and the Relationship Between Similarity and
Perceptual Judgment
The German philosopher Hans Cornelius (1863–1947) is classiªed in var-
ious ways by different sources, most commonly however as a neo-Kantian-
ism with certain positivist and pragmatist ingredients.13 Cornelius, OMS
was a professor in Munich from 1903 and in Frankfurt from 1910 (où,
incidemment, he was the dissertation advisor of both Horkheimer and
Adorno) emphasized the analysis and description of immediately given
facts of consciousness as a basic task of epistemology, believing that gen-
eral judgments about an object can be grounded in the recollection of an expe-
rience of the object, and that any recollection of an experience of an object involves a
recollection of similarity between this experience and other experiences. A judgment
about an object is thereby justiªed in terms of a relational judgement,
which in turn is based in an analysis of consciousness. He took it to be the
task of philosophy to clarify the meanings of concepts, believing that they
were derivable from the immediately given. Cependant, he rejected the
classiªcation as a phenomenalist, since he believed in the existence of
things in themselves, which are governed by laws that can be grasped by
the human mind (Cornelius 1921, Ziegenfuß & Jung 1950).14

In developing his ideas, Cornelius (1903) starts with a basic taxonomy
of what he calls “elements of experience”, very much along empiricist
lines, but then proceeds to argue that the British empiricists have been
unable to account for the ways in which more complex phenomena of our
mental life are constructed out of simple ones (Cornelius, 1903, 192). Il
continues by arguing that this inability is a home-made problem, result-
ing from the atomistic presuppositions of associationism: If one starts out
by artiªcially conceiving of experience as atomistic, then its not surprising
if one runs into problems when trying to regain the internal unity of com-
plex experience by putting the parts back together again in an additive
manière.

The immediately given [. . .] is the complex in which the parts ap-
pear in their unity. We can distinguish between building blocks
[. . .] by abstraction from the context in which we ªnd them. But if

13. As pointed out by Köhnke (1984), neo-Kantian was hardly a uniªed position.
14. En effet, he rejected any kind of philosophical labeling, conªdently declaring that
his work could not be subsumed under any of the usual schools, “because it is not a repeti-
tion of other theories”. (1921, footnote, p. 96)

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

we want to [. . .] understand [. . .] the structure of our mental life
in terms of its elements, we may not engage in this abstraction, mais
have to pay attention to the relations in which the unity is founded.
[. . .] We have to inquire about the features, which create the unity
of the elements of our stream of consciousness at any moment. (Cornelius
1903, 206, my translation)

Cornelius argued that the reason why we always perceive a unity of parts
was that the contents of our experiences are never entirely new, but are al-
ways based on a recognition of a previous experience (op. cit., 213) et en
order to have such an experience of a recollection of similarity, there have to
be two experiences, such that a similarity can be recognized. Being able to
have such an experience is due to some activity of the mind. Cornelius at-
tributes this basic idea (c'est à dire., the notion that individual experiences are
brought into the unity of consciousness by virtue of some cognitive fac-
ulty) to Kant (op. cit., 217 ff.). Our ability to recognize two experiences as
similar is a precondition for two experiences being associable, and there-
fore associationism in itself is insufªcient to account for the connection of
ideas. En outre, two experiences are recognized as similar by virtue of
their complex character:

The identiªcation of the contents of our consciousness presupposes
their recognition . . . Ainsi, every element is a speciªc element for
our consciousness only insofar as we recognize it. Since those con-
tents are never experienced in isolation . . . recollections and recog-
nitions are not directed towards individual contents, but to com-
plexes. (Cornelius, 1903, 226)

6.2 Carnap’s Critique of the Metaphysics of “Antimetaphysical”
Epistemology
Canap cites Cornelius several times in his Aufbau, in particular in § 67,
when mentioning previous critics of an atomistic psychology and episte-
mology and in § 74, where he mentions that not individual elements of
experience, but the law-like relationship between them is of interest. Dans-
terestingly, he does not give any reference with respect to the origin of the
notion of “recollection of similarity”, Néanmoins, I would like to make
the case that he borrows the term from Cornelius. Notice, cependant, que
while Cornelius uses the expression to refer to a mental act whereby two ex-
periences are related to each other, Carnap uses it to refer to the relation
between two experiences as such. This shift is signiªcant, because it points
to the fact that Carnap’s is indeed an entirely different project. This is
quite clear from the fact that Carnap, in his 1922 manuscript, “From

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

Chaos to Reality” essentially presents us with the core ideas of the Aufbau,
where ultimately the relationships in question are formal, yet develops
this from a little taxonomy of types of experience. This taxonomy (lequel
is somewhat reminiscent of one presented by Cornelius) contains “living”
and “dead” parts of experience (sensations and representations). In the
realm of “dead” experiences, he further distinguishes between “ªnished”
ones (memory representations) and “neutral” ones (representations that are
not memories). Carnap furthermore tells us that experiences can be related
to each other in terms of being the same (G), similar (UN), or marginally
identical (Nebengleichheit: G2), where identical experiences form quality
classes (Q). Having laid out his three basic relations (G, UN, G2) Carnap
now states that even though the experiences in question are unanalyzable
building blocks, we can now use the three relations to construct a new
realm of elements that do not have individual attributes, but are mere re-
lation terms.

While Carnap clearly moved beyond Cornelius’s project—insofar as he
did not treat recollection of similarity as the psychological precondition
for a judgment—I would like to argue that Cornelius’ work stood for a
type of philosophy that was formative for Carnap, because he shared
the antimetaphysical attitude that motivated it. Cependant, it is notewor-
thy that Cornelius—while ostensibly appealing to empirical facts about
experience—was still very much engaged in philosophical speculations
about the nature of he human mind and its relation to the physical world.
These were precisely the kinds of speculations that Carnap wanted to get
away from. He contrasted his own approach with others which, he told his
readers, seemed to be epistemological, but were in fact metaphysical, par exemple., concernant-
alistic, idealistic, or solipsistic (Aufbau, § 52). A few years later, in his arti-
cle about the logical analysis of language as a means for overcoming meta-
physics, Carnap again mentioned certain schools that were incorrectly
referred to as epistemological schools, even though they in fact make
metaphysical assumptions (realism, idealism, solipsism, phenomenalism,
positivism) (Carnap 1931a, 237). These remarks are signiªcant, because
they suggest that initially the target of Carnap’s antimetaphysical analysis
was a lot closer to home than Heidegger (who is his explicit target). While
Carnap and Heidegger clearly had incompatible philosophical projects
(see Friedman 2000), I argue that in his Aufbau, Carnap took as his
point of his departure philosophers whose philosophical projects (anti-
metaphysical, pro science, pro clarifying basic terminology) he saw as com-
patible with his, while believing their efforts to have been hindered by
metaphysical residues. Some of the philosophers he had in mind here were
Schlick and Cornelius (as carrying realistic residues), and proponents of

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

various forms of phenomenalism (Mach, Avenarius, Schuppe, Schubert-
Soldern, Gomperz).

7. The Analysis of Consciousness in Immanence Positivism
In his autobiography, Carnap writes that “The choice of a phenomenalistic
basis was inºuenced by some radical empiricist or positivist German phi-
losophers at the end of the last century whom I had studied with interest,
in the ªrst place Ernst Mach, and further Richard Avenarius, Richard von
Schubert-Soldern, and Wilhelm Schuppe” (Carnap 1963, 18). With all
due caution regarding retrospective statements like this, we should not
disregard this one. I want to argue that, as did Cornelius, these writers in-
tended their analyses of consciousness to be in the service of an antimeta-
physical project, with which Carnap was sympathetic. He saw himself as
proposing a method of analysis that didn’t have the metaphysical residues
that he saw in theirs, while being compatible with (though not inherently
tied to) the phenomenalist language used by them. Ainsi, while Carnap’s
choice of “units of experience” was similar to those of his phenomenalist
and positivist predecessors, this was not where their systematic impact lay.
Dans ce qui suit, I want to demonstrate this by following his references
(again, in § 67 of the Aufbau) to members of immanence philosophical
movement and to Heinrich Gomperz.

Immanence philosophy was a philosophical school that aimed to re-
strict itself to an analysis of the immediately given (Schuppe and Schu-
bert-Soldern).15 In a similar vein, Avenarius’s “empiriocriticism” at-
tempted a “critique of pure experience”, by which he meant an analysis of
experience which not only purged it from all metaphysical ingredients
(Avenarius 1888; Eisler 1910, Ritter 1974), but also thereby reconciled a
scientiªcally minded philosophy with the prescientiªc, experiential basis
of our ordinary conception of the world. For a brief while (1895–1899),
there was even a Zeitschrift für immanente Philosophie, edited by Schubert-
Soldern and Schuppe. This school, while largely forgotten today, était
clearly a force that proponents of a scientiªc philosophy engaged with, comme
indicated by contemporary responses to it (par exemple., Wundt 1896; Schlick
1997 [1918]). Essentially, Schlick’s critique reads like a standard refuta-
tion of phenomenalism. We can assume that Carnap was familiar with this
critique. Given his repeated assertion that his constitutional system was

15. One proponent of immanence philosophy, goes even further and posits that if we
take the term, “immanent philosophy” to refer to all attempts to free epistemology of its
metaphysical presuppositions, then a great number of thinkers need to be included, dans-
cluding neo-Kantians like Rickert (Schubert-Soldern 1896, 305).

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

supposed to be neutral with respect to metaphysical positions like
postivism or phenomenalism, we can assume that it was not this part of
immanence positivism that he was sympathetic with. Given furthermore
his discussion of other possible constitution systems, we can assume that
his choice of the phenomenal basis had nothing to do with the immanence
positivist tenet that the only way to achieve an anti-metaphysical episte-
mology was by means of an analysis of human experience. Ainsi, as in the
case of Cornelius, I believe that it was the anti-metaphysical impetus itself that
drew him to their work. Ce, bien sûr, does not rule out that their con-
ceptualization of simple experience had an impact on (or ªt in well with)
his own, as is clear from his references in § 67, for example to Schuppe,
who wrote, “simple sensations usually do not come to consciousness in an
isolated fashion; an individual’s thinking begins with total impressions,
which are only analyzed into its simple elements by means of reºection”.
(Schuppe 1894, 49)

The anti-metaphysical impetus that I have been pointing to was also
shared by another philosopher cited by Carnap in relation with the com-
plex character of basic experience, c'est à dire., the Austrian philosopher Heinrich
Gomperz (1873–1942). Gomperz, who came from a distinguished family
of Jewish Viennese intellectuals (Haller 1994), was strongly impressed by
both Mach and Mill and interested in empiriocriticism, but critical of
Avenarius’s version of. As early as 1891, he had initiated an interdisciplin-
ary philosophical discussion group (“Circle of Socratics”), in which many
of the members of what was later going to become the Vienna Circle par-
ticipated (Stadler 1994).16 Dans 1905, he published the ªrst (of what were
projected to become three) volume of his work, Weltanschauungsleere, lequel
he characterized as a science that attempts to reconstruct scientiªc knowl-
edge into a conceptual system that is free of contradictions (Gomperz
1905, p. 17).17 Gomperz also states that such an explication of scientiªc
knowledge can avoid all metaphysical speculation, as the idealist and the
realist usually do not disagree about the scientiªc facts (Gomperz 1905,
15). Cependant, he cautioned philosophers to keep in mind that they could
only analyze concepts, not things. The aim of his Weltanschauungslehre was
to reconstruct traditional concepts in a way that freed them of contradic-
tion, such that they could be used to describe the facts (Gomperz, op. cit.,
31). Like Cornelius, he believed that in order for real experience to occur,

16. À son tour, he was later going to participate in some of their meetings, though he kept

a critical distance.

17. The concept of Weltanschauung may be said to have enjoyed a great popularity in a

great range of philosophical and popular at the time.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

form had to be supplemented by some kind of intellectual unifying func-
tion, c'est à dire., an “emotive” type of cognition. Based on the nature of this type
of Gestalt perception, where the unifying factor in conscious experience is
its emotional “ºavor”, Gomperz referred to his position as “pathempiri-
cism”. Traces of this latter conceptualization of experience can be found in
the following quote from Carnap’s Aufbau:

We do not frequently assign qualities of emotions or volitions as
properties to things in the outside world. [. . .]. We must assume,
cependant, that to decline this assignment is only the result of a pro-
cess of abstraction and does not hold from the outset. In the uncrit-
ical conception of a child, the apple does not only taste ‘sourish’ but
‘delicious’. (Carnap 1967, § 133)

Summing up, apart from shared ideas about characteristics of basic experi-
ence, we ªnd in Gomperz’s writings not only the antimetaphysical senti-
ment of the Aufbau, but also the idea that an antimetaphysical philosophy
proceeds by reconstructing the concepts of scientiªc knowledge. C'est
this sentiment, rather than the mentalistic way in which it was carried
through, that Carnap sympathized with.

8. Conclusion
In this paper, I have investigated the background of Carnap’s references to
Gestalt perception. Given Carnap’s personal acquaintance with members
of the Berlin/Frankfurt school of Gestalt perception, and given further his
aim of (un) basing his constitutional system in phenomenal experience, et
(b) providing an intersubjective, structural account of such experience, je
followed up the issue of whether Gestalt psychologists might have pro-
vided Carnap with a scientiªc account of the structure of experience. I dis-
tinguished between two aspects of this question. D'abord, the question of
whether Gestalt psychologists might have provided Carnap with evidence
for the claim that basic experience is structured. Deuxième, the question of
whether Gestalt psychologists might have provided him with a scientiªc
account of how such structuring takes place. I presented an argument in fa-
vor of the ªrst reading. This argument was based on the fact (un) que
Carnap rejects the existence of elements of Gestalt experience in a way
that is similar to the Berlin/Frankfurt critique of Ehrenfels’s conceptual-
ization of Gestalt experience, mais (b) Carnap does not follow the Berlin/
Frankfurt explanation of what makes two experiences structurally similar.
I then linked this analysis to a particular reading of the place of basic ex-
perience in the Aufbau. According to this reading, Carnap’s aim was to
provide a uniªed system of the concepts that constitute our scientiªc
connaissance, where the choice of psychological concepts as the basic concepts

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

was only one of several possible options. With this reading of Carnap’s
project in mind—I argued—it is clear why it would have been in his in-
terest to turn to Gestalt psychology, as representatives of the best current
psychologie. Cependant, I argued that this motivation is to be distinguished
from the motivation to capture the “real” cognitive processes, as psycholo-
gists might describe them. Ainsi, it is neither surprising nor inconsistent
that his formal notion of “recollection of similarity” (and the way he uses
it to provide a quasi-analysis of the units of basic experience) does not ac-
cord with the empirically based account Gestalt psychologists would have
given of this basic cognitive phenomenon.

In the second part of the paper, I raised the question why Carnap uses
terms like “recollection”, which make it seem as if he were analyzing the
esprit (as opposed to merely reconstructing a system of scientiªc knowl-
bord). In investigating this question, I took a closer look at some of the
other writers that he cites in relation to his description of “experience in
its totality and unity”. I identiªed the German philosopher, Hans
Cornelius, as a likely source of the term, “recollection of similarity”. Comment-
jamais, while Cornelius used this term in a quasi-psychological (though not
experimentally based) fashion, basing his epistemological theory of con-
cepts and judgments on it, for Carnap the term did not carry this psycho-
logical connotation. This led me to speculate that we should read
Cornelius (et d'autres, like Schuppe, Schubert-Soldern, or Gomperz) les deux
as a source of inspiration, but also as a point of departure. While Carnap
shared with these philosophers a very profound anti-metaphysical senti-
ment, I argued that he viewed their attempts at overcoming metaphysics
by analyzing the basic facts of consciousness as ultimately being responsi-
ble for why they were still stuck with metaphysical baggage. I argued that
it was philosophers like Cornelius, Gomperz, Schuppe, Schubert-Soldern,
and Mach (rather than Heidegger) that he had in mind when suggesting
that he was capable of presenting an epistemological system that was gen-
uinely free of metaphysics.

Les références
Ash, M.. 1994. “Gestalttheorie und logischer Empirismus”,

in L.
Danneberg, UN. Kamlah & L. Schäfer (Éd.) (1994): Hans Reichenbach und
die Berliner Gruppe. Braunschweig/Wiesbaden: Vieweg.

———. 1995. Gestalt Psychology in German Culture 1890–1967. Holism
and the Quest for Objectivity. Cambridge, ROYAUME-UNI: L'université de Cambridge
Presse.

Avenarius, R.. 1888. Kritik der reinen Erfahrung (Bd. 1). Leipzig: Ô. R..

Reisland (3rd edition).

Bernhard, H. 2002. “Zur Geschichte der Zeitschrift ‘Erkenntnis’ (1930–

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

1940) im Lichte des Briefwechsels von Hans Reichenbach”, in E.
Neuenschwader & L. Bouquiaux (éd.): Science, Philosophy and Music.
Turnhout, Belgium: Brepolis

Carnap, R.. 1922. “Vom Chaos zur Wirklichkeit” (unpublished manu-
scénario, University of Pittsburgh, Archives for Scientiªc Philosophy).
———. 1923b. “Quasizerlegung” (unpublished manuscript, University

of Pittsburgh, Archives for Scientiªc Philosophy).

———. 1979. Der logische Aufbau der Welt. Frankfurt/Berlin/Wien: Ull-

stein (4th edition) (First published 1928)

———. 1931un. “Überwindung der Metaphysik durch logische Analyse

der Sprache”, Erkenntnis, 2, 219–241.

———. 1931b. “Die physikalische Sprache als Universalsprache der

Wissenschaft”, in Erkenntnis 2, 432–456.

———. 1963. “Intellectual Autobiography”, in P. UN. Schilpp (Ed.): Le

Philosophy of Rudolf Carnap. La Salle, Ill.: Open Court.

———. 1967. The Logical Structure of the World. Pseudoproblems in Philoso-

phy. Londres: Routledge & K. Paul

Cornelius, H.
Teubner.

(1897): Psychologie als Erfahrungswissenschaft. Leipzig:

———. 1900un. “Ueber ‘Gestaltqualitäten’”. Zeitschrift für Psychologie und

Physiologie der Sinnesorgane, 22, 101–122.

———. 1900b. “Zur Theorie der Abstraction”. Zeitschrift für Psychologie

und Physiologie der Sinnesorgane. 24, 17–141.

———. 1903. Einleitung in die Philosophie. Leipzig: B. G. Teubner
———. 1921. “Hans Cornelius. Leben und Lehre”. In Raymund Schmidt
(éd.): Die deutsche Philosophie der Gegenswart in Selbstdarstellungen, Vol. 2,
Leipzig: Felix Meiner.

Danneberg, L. & W. Schernus. 1994. “Die Gesellschat für wissen-
schaftliche Philosophie: Programm, Vorträge und Materialien”, in L.
Danneberg, UN. Kamlah & L. Schaefer (Éd.) 1994. Hans Reichenbach und
die Berliner Gruppe. Braunschweig/Wiesbaden: Vieweg

Dilthey, W. 1990b. “Ideen über eine beschreibende und zergliedernde
Psychologie”. In Wilhelm Dilthey, Gesammelte Schriften, Volume V.
Göttingen: Vandenhoeck & Ruprecht. (First published 1896)

Ehrenfels, C. v. 1890. “Über ‘Gestaltqualitäten’”, Vierteljahreschrift für
wissenschaftliche Philosophie, 14, 242–92. (English translation: “On ‘Ges-
talt Qualities’” by Barry Smith, in B. Forgeron. 1988. (éd.): Foundations of
Gestalt Theory. Philosophia Verlag, Munich/Vienna).

Eisler, R.. 1910. Wörterbuch der philosophischen Begriffe. Berlin: Ernst

Siegfried Mittler & Sohn.

Fabian, Reinhard. 1997. “The Graz School of Gestalt Psychology”, in W.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

Bringmann, H. Lück, R.. Miller & C. Early (Éd.): A Pictorial History of
Psychologie. Chicago: Quintessence Publishing Co.

Feest, U. 2007. “‘Hypotheses, Everywhere Only Hypotheses’—On Some
Contexts of Dilthey’s Critique of Explanatory Psychology”. Studies in
History and Philosophy of Biological and Biomedical Sciences. 38, 43–62.
Friedman, Michael. 2000. A Parting of the Ways. Carnap, Cassirer, et

Heidegger. Chicago & Lasalle: Open Court.
Gomperz, H. 1905. Weltanschauungslehre.

Jena & Leipzig: Eugen

Diederichs.

Haller, R.. 1994.

“Heinrich Gomperz und die

österreichische
Philosophie”, in M. Seiler & F. Stadler (éd.): Heinrich Gomperz, Karl
Popper und die österreichische Philosophie. Amsterdam/Atlanta: Rodopi.
Hegselmann, R.. & G. Siegwart. 1991. “Zur Geschichte der ‘Erkenntnis’”,
in Wolfgang Spohn (éd.): Erkenntnis Orientated: A Centennial Volume for
Rudolf Carnap and Hans Reichenbach, Dordrecht-Boston-London
(Erkenntnis 35), 461–471.

Helmholtz. H. v. 1910. “Die Lehre von den Gesichtswahrnehmungen”.
Handbuch der physiologischen Optik. Vol. 3. Hamburg & Leipzig: Leopold
Voss. (First published 1867).

Hoffmann, D. (1994): “Zur Geschichte der Berliner ‚Gesellschaft fuer
empirische/wissenschaftliche Philosophie’”
in L. Danneberg, UN.
Kamlah & L. Schaefer (Éd.). 1994. Hans Reichenbach und die Berliner
Gruppe. Braunschweig/Wiesbaden: Vieweg.

Jäger, S. 2003. “Wolfgang Köhler in Berlin”. L. Sprung & W. Schönpºug
(éd.): Zur Geschichte der Psychologie in Berlin. Frankfurt a. M.: Pierre
Lang, 275–301. (2nd edition).

King, D. B & M.. Wertheimer. 1995. “Max Wertheimer at the University
of Berlin”, S. Jäger, je. Stäubele, L. Sprung & H.-P. Brauns (éd.):
Psychologie im soziokulturellen Wandel—Kontinuitäten und Diskontinuitäten.
Frankfurt a.M.: Peter Lang.

Köhler, W. 1913. “Über unbemerkte Empªndungen und Urteil-

stäuschungen”. Zeitschrift für Psychologie, 66, 51–80.

———. 1925un. “Gestaltprobleme und Anfänge einer Gestalttheorie”,
Jahresbericht ueber die gesamte Physiologie und experimentelle Pharmakologie,
3 (Bericht ueber das Jahr 1922), 512–39. (Reprinted in Gestalt Theorie, 5,
1983, 178–205).

———. 1925b. “Komplextheorie und Gestalttheorie”, Psychologische

Forschung 6, 358–416.

Köhnke, K. C. 1986. Entstehung und Aufstieg des Neukantianismus. Die
deutsche Universitätsphilosophie zwischen Idealismus und Positivismus. Frank-
furt: Suhrkamp.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Science and Experience . . .

Kusch, M.. 1995. Psychologism. A Case Study in the Sociology of Philosophical

Knowledge. Londres & New York: Routledge.

Ley, M.. 1994. “Ernst Mach und die Gestaltpsychologie”. Horst Gundlach
(éd.): Arbeiten zur Psychologiegeschichte. Göttingen, Bern, Toronto,
Seattle: Hogrefe.

Lipps, T. 1900. “Zu den Gestaltqualitäten”. Zeitschrift für Psychologie und

Physiologie der Sinnesorgane, Vol. 22, 383–385.

Mach, E. 1886. Beiträge zur Analyse der Sinnesempªndungen. Jena: G.

Fischer.

Mach, E. 1903. Die Analyse der Empªndungen und das Verhältnis des

Physischen zum Psychischen. Jena: G. Fischer.

Mach. 1996 The Analysis of Sensations. Londres: Routledge/Thoemmes

Presse

Majer, U. 2003 “Carnaps Übernahme der Gestalttheorie in den Aufbau im
Lichte heutiger, vor allem computationaler Theorien des Sehens”. T.
Bonk (éd.): Language, Truth, and Knowledge. Kluwer Academic Pub-
lishers, 65–78

Mayer, V. 1991. “Die Konstruktion der Erfahrungswelt: Carnap und

Husserl”. Erkenntnis 35, 287–303.

———. 2004. “Ganze und Teile. Konstitutionsbeziehungen in der Philo-
sophie um 1900”. Litterature et theorie de la connaisaance 1890–1935.
Strassbuorg: Presse Universitaires de Strassbourg.

Münch, D. 1997. “Franz Brentano and Phenomenology”, in L. Embree et
al. (éd.): Encyclopedia of Phenomenology, Dordrecht: Kluwer, 71b–75b.

———.

2003/4.

Psychologie. Die
wissenschaftliche Weltauffassung Carl Stumpfs”. Brentano Studien 10,
11–66.

“Erkenntnistheorie

und

Pincock, C. 2003. “Carnap and the Unity of Science: 1921–1928”. T.
Bonk (éd.): Language, Truth, and Knowledge. Kluwer Academic Pub-
lishers, 87–96.

———. 2005. “A Reserved Reading of the Aufbau”. Paciªc Philosophical

Trimestriel 86, 518–543.

Richardson, UN. 1998. Carnap’s Construction of the World : The Aufbau and the
Emergence of Logical Empiricism. Cambridge: Cambridge Univ. Presse.
Ritter, J.. 1974. Historisches Wörterbuch der Philosophie. Darmstadt:

Wissenschaftliche Buchgesellschaft.

Scheerer, M.. 1931. Die Lehre von der Gestalt. Ihre Methode und ihr psych-

ologischer Gegenstand. Berlin: de Gruyter.

Schlick, M.. 1997. Allgemeine Erkenntnislehre. Frankfurt: Suhrkamp. (D'abord

published 1918).

Schubert-Soldern, R.. v. 1896. “Erwiderung auf Wundt’s Aufsatz “Ueber

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Perspectives on Science

kritischen und naiven Realismus”, Philosophische Studien 13(2), 305–
317.

Schuppe, W. 1894. Grundriss der Erkenntnistheorie und Logik. Berlin: R..

Gaertners Verlagsbuchhandlung.

Forgeron, B. 1988. “Gestalt Theory: An Essay in Philosophy”, in Smith (éd.)
Foundations of Gestalt Theory, München, Wien: Philosophia Verlag.
Sprung, H. & L. Sprung. 1995. “Carl Stumpf (1848–1936) und die
Anfänge der Gestaltpsychologie an der Berliner Universität. In S.
Jäger, je. Stäubele, L. Sprung & H.-P. Brauns (hrsg.): Psychologie im
soziokulturellen Wandel—Kontinuitäten und Diskontinuitäten. Frankfurt:
Lang.

———. 2003. “‘Wir brauchen einen Mann, welcher heimisch ist in . . .
der experimentellen Psychologie’—Carl Stumpf
in seiner Berliner
Zeit”. L. Sprung & W. Schönpºug (éd.): Zur Geschichte der Psychologie in
Berlin. Frankfurt a. M.: Peter Lang, 201–125. (2nd edition)

Stadler, F. 1994. “Heinrich Gomperz und Karl Popper im Kontext des
logischen Empirismus,” in M. Seiler & F. Stadler (éd.): Heinrich Gom-
perz, Karl Popper und die österreichische Philosophie. Amsterdam/Antlanta:
Rodopi.

———. 1997. Studien zum Wiener Kreis. Ursprung, Entwicklung und Wir-

kung des Logischen Empirismus. Frankfurt a. M.: Suhrkamp.

Stumpf, C. 1891. “Psychologie und Erkenntnistheorie.” Abhandlungen der
Königlich Bayerischen Akademie der Wissenschaften. Philosophisch-Philo-
logische Classe, Bd. 19, 2, Abt. München: Verlag der Akademie, 465–
516.

Wertheimer, M.. 1912. Experimentelle Studien über das Sehen von

Bewegung. Zeitschrift für Psychologie, 61, 161–265.

———. 1925. “Über Gestalttheorie.” Philosophische Zeitschrift für For-

schung und Aussprache 1, 39–60.

Whitehead, UN. N. & B. Russell. 1925. Principia Mathematica. Cambridge:
la presse de l'Universite de Cambridge. (First published 1910 (vol. 1), 1912, (vol
2), 1914 (vol. 3)).

Wundt, W. 1896. “Ueber naiven und kritischen Realismus,” Philosophische

Studien 13(1), 307–408

Ziegenfuß, W. & G. Jung 1950. Philosophenlexikon. Handbuch der Philo-

sophie nach Personen. Berlin: de Gruyter.

D
o
w
n
o
un
d
e
d

F
r
o
m
h

t
t

:
/
/

d
je
r
e
c
t
.

je
t
.

e
d
toi
p
o
s
c
/
un
r
t
je
c
e
–
p
d

F
/

1
5
1
1
1
7
8
9
4
4
7
p
o
s
c
.
2
0
0
7
1
5
1
1
p
d

b
oui
g
toi
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3 Science and Experience/ image

Télécharger le PDF