Wolfgang Wildgen
Semiosis based on the principles
„saillance“ and „prégnance“
18-19-20 luglio 2005
Semiosi e catastrofi. L’eredità semiotica di René
Thom/ Semiosis et catastrophes. L’héritage
sémiotique de René Thom/ Semiosis and
Catastrophes. René Thom's semiotic heritage.
1
Thom’s itinerary towards
„semiophysics“
 In the sequel of a letter exchange with the biologist
Waddington Thom saw language in a quasi continuity
with biological development (morphogenesis).
 The model proposed by Waddington and elaborated by
Thom was rather global and concerned the major
topological features and the process of differentiation as
such (bifurcations, appearance of barriers, separation of
specific organs and subfunctions, etc.).
 In this realm, Thom could apply the new results obtained
in catastrophe theory based on the classification
theorem.
2
Controversies
 At this point the geometry of the umbilics was not yet
known in detail and in the second version Thom added
some remarks concerning the umbilics, but he did not
modify the list of semantic archetypes, although those
derived from the umbilics could not be maintained under
the new insights.
 The classification of paths which had been done
intuitively by Thom was also problematic for all
catastrophes beyond the cusp .
 Meanwhile, the catastrophe controversy which regarded
mainly the applications proposed by Christopher
Zeeman (cf. Sussmann and Zahler, 1978) had weakened
the international interest in applications.
3
Thom‘s model of semiosis
 In the same period (after 1978) Thom went one step
further and tried to specify the forces which govern the
process of semiosis and not only the topology of its
outcomes.
 Thom first tried to link the forces of the morphogenesis of
meaning to known basic forces like gravitation, radiation
(light), etc. These universal fields embed the living
beings and govern their environment (ecology). They are
naturally the background of all perceptual and motor
processes.
 In perception, light is at the basis of our visual
perception; gravitation underlies human and animal
motor-processes and the sensation of pressure and
weight; sound waves are registered by the ear and
chemical substances evoke reactions of our taste and
smell organ.
4
Semiophysics
 As the dynamics of such fields (e.g., light) have been the
topic of physics since Newton, and wave dynamics were
the topic of specific mathematical treatments since
Maxwell, it was straightforward for Thom to postulate a
specific field that is registered and filtered by our sensory
organs. He called it “saillance”, i.e., those effects which
stand out, may be selected as informative in a
psychophysical field.
 His program was to extract as much systematic content
as he could from the analogy between physical fields
and perceptual fields.
 Insofar as perception is the basic stratum of semiosis,
any perceptually basis symbolic structure elaborates the
psychophysical fields of our sensory systems.
5
Remaining gap
 Although this strategy: from physics to semiotics
allowed for the transfer of many mathematical
techniques, which had been shown to be
successful in physics, there remained a large
gap between psychophysics (the level of
perception) and linguistic (or cultural) semiosis.
 The term “prégnance” had to fill this gap and to
explain the transition between very basic
perceptual reactions and language.
6
The concept of “prégnance”
 The basic meaning of the term introduced by Thom
comes from German “Prägnanz”, which has been
internationalized by gestalt-psychology. In German the
adjective “prägnant” means that something perceived,
enounced, remembered has outstanding properties,
which catch attention, makes it relevant, important in a
specific situation; one also associates brevity (laconism)
and sudden effect as in jokes or aphorisms with it.
 Other words, that may be associated to this lexical field,
are “prägen”, “Prägung”, i.e., the coining of piece of
money or the imprinting of a sign on an object.
 “Prägnanz” has a positive value in gestalt-psychology
insofar as the term “gute Gestalt” (optimal form) refers
also to “Prägnanz”.
7
 “Prägnanz” has a positive value in gestalt-
psychology insofar as the term “gute
Gestalt” (optimal form) refers also to
“Prägnanz”.
 The term in French “prégnance” (or even
less evident in English “pregnance”)
should be understood as a lean-translation
with the lexical content of German
“Prägnanz”.
8
The scientific use of the concept
 In Pavlov’s experiments with dogs, they salivate if
presented with meat and “learn” the conditioned reflex of
an associated bell.
 Konrad Lorenz observed the process called “Prägnanz”
(imprinting) in birds (ants, geese, etc.). During a short
period after they left the egg, many birds select rather
unspecific stimuli in their environment and quickly
elaborate basic concepts like that of a “mother bird”.
 Jakob von Uexküll. A similar concept of
“Bedeutungswelt” (meaningful universe) was proposed
by another biologist, Jakob von Uexküll. For Uexküll
every animal creates its own “Bedeutungswelt” which
depends first on its windows of perception and then on
its vital need.
9
 In the realm of psychology Gibson elaborated
the concepts of “prägnante Gestalt” and “Valenz”
of his teacher Koffka and coined the term of
affordance. Any object or process in our
environment may have affordances; thus a chair
allows for sitting, a bed for sleeping, etc.
 Cassirer generalized the perceptual/motoric
“Prägnanz” to “symbolische Prägnanz” as
perception is the first level of semiosis (symbol
creation) in his system.
 The application of the concept “Prägnanz”
(imprinting) to language acquisition was
proposed by William Stern, who recognized a
type of goal-oriented, internally controlled
process in language acquisition.
10
Thom adds two new ideas:
 The salience (“saillance”) effect in perception
may be linked via psycho-physical laws to the
dynamics of objective fields in physics and
chemistry .
 The “prégnance” effect applies to the topology of
salient objects and events via a process called
“diffusion de prégnance” (channeling of
“prégnance”). In this channeling the multiple
forms of (perceptual) “saillance” are molded into
the instinctive and inborn (poor) forms and
elaborate them to rich and context-dependent
fields of categorical perception and behavior.
11
Where mathematics come in
It was clear (for Thom) that the psychophysical transition
calls for the application of the laws of physical
dynamics:
 The first line has been further elaborated by the group
of Turvey, Kelso and others and in my own work.
 The second line (diffusion of “prégnance”), calls for
something like fluid dynamics and models of growth in
space and time.
 One may ask however if the generalization of
mathematical models from physics to psychology
(neuropsychology) and from there for linguistics
(semiotic systems) is philosophically sound. The work
of Cassirer (mainly after 1935) may guide this inquiry.
12
Mathematics in psychology and
semiotics: Cassirer’s scepsis
The basic question are:
 How is the mathematical study of the
physical world related to the study of the
mental and the cultural word?
 Is there continuity such that the concepts
of mathematics successful in the scientific
capture of features in the objective world
may be applied (with the same success) to
the mental and the cultural world?
13
Answers
 In mathematical Platonism, the mathematical
model has an existence of its own right and its
application to physics is just one possible field
among many.
 A more synthetic view of mathematics assumes
a specific adaptation to major problems of
physics and technology in the course of
millennia
 An even more radical view (Lakoff) may even
state that mathematics are a specialized
deformation of natural (language) reasoning and
thus are biased for the analysis of language and
culture.
14
Cassirer‘s position
 Cassirer argued in several articles written towards
the end of his life, that mathematics (mainly Klein’s
geometry and topology) describe only a realm of
possibilities (e.g., possible Euclidean and nonEuclidean geometries) and that an empirical study
of visual perception can at its best select, specify,
even combine these geometries or their features.
 The basic selection must be made based on
empirical arguments. The fact that certain
mathematical structures have been invented
historically does not allow any statement about the
structure of visual perception.
15
Leyton‘s critique of the Klein
(Erlangen) Program
 Leyton (2001) even argues that the invariant structures
arrived at by Felix Klein have eliminated the memory and
thus the information contained in visual and linguistic
(symbolic) forms.
 Leyton’s criticism affects Thom’s models insofar as
elementary catastrophes correspond to the geometrical
groups of regular polygons, Platonic solids, etc. (cf.
Slodowy, 1995). Therefore, one may assume that the
elementary catastrophes are structures stripped off their
(semantic) information and that via deformation,
breaking of symmetry and combination of the
catastrophe schemata the information may be recovered
or reconstructed.
16
A new perspective on Thom‘s
proposals
 This puts a new interpretation on Thom’s semiophysics
but the major concern is now, not how to find (or
construct) the archetypes, but how to describe the
modes of accumulating meaning, starting from these
meaningless forms.
 The process of “diffusion de prégnance” is now rather a
process of meaning creation than of meaning distribution
from an original source full of (pregnant with) meaning.
 This puts Thom’s semiophysics from its head to its feet
and at the same time it diminishes the role of basic
releasers
17
The propagation of “prégnance” and the
dynamics of language
 All basic meanings are “theoretical entities”, i.e., they
are difficult or impossible to observe, but must be
assumed in order to understand the observable effects.
 We may observe basic behavioral categories in babies,
register the semantic structures in one-word or twoword utterances and finally describe the lexicon and
syntax of the adult language.
S0
before birth
> S1
new born child,
prelinguistic
> S2
linguistic start
>S3
adult competence
The theoretical starting level S0 must evolve regularly given a human genome
and normal conditions of maturation (before birth).
18
 At birth only some auditory capacities already developed
in the womb exist; the first period of development
concerns our other sensory organs and basic motor
programs (cf. Piaget’s senso-motor level of cognitive
development).
 The genome cannot code for specific sensory forms or
motor-patterns, i.e., it can only fix some gradients that
allow the detection of relevant input in order to trigger the
process of elaboration (intrinsic) and learning (extrinsic).
 Parallel to these basic gradient-fields the sensory
capacities and corresponding categorical and memory
capacities evolve. For simplicity sake, I assume only one
basic (inborn) gradient (a correlate could be found in the
genetic code) called P (prégnance).
19
 A first division of P may be due to the multiplicity of Si. If
the different sense organs evolve in a consecutive order,
e.g., Sau(ditive)  Svi(sual), the types of P invested by Sau
may condition the structure of P invested by Svi. This
leads to a cascade of Pk invested by Si:
P invested by Sau

PK(Sau)
P invested by Svi

PK(Svi)
The same thing may occur with all sensory fields
(whose number is open, traditionally five, maximally 30).
If the first investments prefigure the later, we obtain a
more complicated net:
P
P
PK
(Sau)
PK (Svi)
20
Evolutionary dynamics of language
 A new level is reached in alarm calls, i.e.,
perception is coupled to a behavior which is able
to redistribute the perception and triggered
action to a community (independent of
consciousness, free will or intentionality).
 In this case, the category of interpretant begins
to surface in behavior (not yet in
consciousness). If the propagation of P is the
base line which links perception and action, than
the alarm is caused by the stability of the link
between both which causes thirdness (the
interpretant).
21
I2
The alarm-call is
perceived and causes
action (in adult
animals who have
learned the rule) and
we can proceed
further in the
hierarchy of semiosis.
Rule 2
R2
Alarm call I1
Action
O2
Rule 1
R1
O1
Interpretant I
alarm call
Rule
Perception
R
Action
O
The circular process of semiosis
begins.
22
Selforganization of complex
systems
 If call-repertoires beyond the number of two, three, four
evolve; this may be assumed as soon as bodily group
communication (e.g., lousing) is replaced by call patterns
(cf. the work of Dunbar).
 A new type of phenomenon occurs which may be called
the denotational and connotational fields. The set of
external and social meanings is now organized by
criteria of parsimony and distance in a semantic space.
 For the denotational space, processes of metonymical
and metaphorical generalization have been shown to be
crucial; cf. Lakoff and Johnson (1980).
 In the case of social meanings Osgood was able to show
that an abstract space E (Evaluation), P (Potency) and A
(Activity) may be assumed.
23
Propagation of “prégnance” P into more and
more segmented attractor fields
 The simple propagation of “prégnance” typically produces a cascade
of attractors with diminishing energy,
24
Valence patterns
In the case of valence we must assume two stages:
 A bifurcation into two categorically opposed fields; e.g.,
subject–predicate (or topic–comment; figure–ground;
trajectory–landmark).
 In the next stage a bifurcation of one nominal role (NP)
into two or three (in extreme cases four).
 The valence pattern is described by a conflict of
“prégnances” in Thom (1978c: 76). If these conflicts are
stripped off their specific intentional and real-life content,
a formal topologico-dynamic pattern is left, which can be
matched against the hierarchy of elementary
catastrophes in Thom (1972).
25
Is there a „pragmatic“ „prégnance“?
 The fact that full linguistic competence is neither
achieved under extreme social depravation as in
Kaspar-Hauser cases suggests the necessary
maturation of a specifically human „social
competence“.
 If autism has genetic sources, there may even
be a genetic basis for this maturation; i.e. genes
and social context must be guaranteed in order
to allow for language acquisition.
26
 This type of „imprinting“ may be specific for
humans, although precursors may exist in other
social mammals, and more specifically primates
(cf. bonobos). It could have evolved out either of
more primitive sexual and/or child rearing
instincts.
 A neural precondition were probably the “mirror
neurons” which allow for motor imitation and
sympathy effects with the other
 The ethic principle “love your next “ is perhaps
the best compression of this “prégnance”
feature.
27
Linguistic/cognitive consequences
 Multi-agent systems and conflicts/alliances between
agents presuppose a social schema which goes beyond
perceptual “saillance” (center/periphery and prototype
effects may generalize based on “saillance”).
 The agent/patient opposition makes use of the basic
consciousness of self/other.
 Causal cognition elaborates the effect Ego has/intends
to have on others and reinterprets the interaction with
objects in this agent/patient light.
 The instrumental relation is further elaborated in the
context of early tool industries or it gives rise to these
industries and thus to modern technology and science.
28
Conjecture on the source of
valence patterns
 The complexity of valence patterns and
basically already the possibility of
predication ask for something beyond
animal instincts (hunger, thirst, sex) and
sensory categorization and make up the
basic human nature of natural languages.
 The genetic disposition for this faculty is
still open, but a morphosemiosis without
such a factor remains incomplete.
29
Conclusions
If we remember the skeptical remarks by Cassirer concerning
the psychology of perception, we can say that the
mathematically derived archetypes are only an abstract set
of possible schemata, which may find applications in rather
divergent fields.
The specific choice, elaboration and filling is an empirical task
in semiotics.
The empirical specification may be found at different levels:
 classificatory procedures of descriptive linguistics, which
specify semantic roles and frames (valence patterns),
 experimental psycholinguistics and developmental results
concerning the formation and interpretation of sentences,
 neurolinguistics and neurodynamics of sentence production
and understanding,
 evolutionary anthropology and genetics (which have to
include a social factor).
30