The epistemology of (second order) cybernetics and of the Principia Cybernetica Project is constructivist. Ernst von Glasersfeld defines radical constructivism by the following two basic principles:

pucefakeAI and Robotics

Nov 30, 2013 (3 years and 8 months ago)

78 views

The epistemology of (second order) cybernetics and of the Principia
Cybernetica Project is constructivist.
Ernst von Glasersfeld

defines radical
constructivism by the following two basi
c principles:

1.

Knowledge is not passively received either through the senses or by
way of communication, but is actively built up by the cognising
subject.

2.

The function of cognition is adaptive and serves the subject's
organization of the experiential wor
ld, not the discovery of an
objective ontological reality.

The importance of constructivism is best understood by comparing it with
the opposite, more traditional, approach in epistemology or cognitive
science, which sees knowledge as a passive
reflection of the external,
objective reality
. This implies a
process

of "instruction": in order to get
such an image of reality, the subject must somehow rece
ive the information
from the environment, i.e. it must be "instructed". The naive view is that
our senses work like a camera that just projects an image of how the world
"really" is onto our brain, and use that image as a kind of map, an
encoding in a slig
htly different format of the objective structure "out
there". Such a view runs quickly into a host of conceptual problems, mainly
because it ignores the infinite complexity of the world. Moreover, detailed
observation reveals that in all practical cases, c
ognition does not work like
that. It rather turns out that the subject is actively generating plenty of
potential

model
s, and that the role of the outside world is merely limited to
reinforcing some of th
ese models while eliminating others (selection).

That construction serves in the first place selfish purposes: the subject
wants to get
control

over what it perceives, in order to eliminate any
deviati
ons or perturbations from its own preferred goal state. Control
requires a model of the thing to be controlled, but that model will only
include those aspects relevant to the subject's goals and actions. In a sense,
the subject does not care about the "thi
ng" to be controlled, only about
compensating the perturbations it senses from its goal, thus being able to
adapt to changed circumstances.

Constructivism has its roots in Kant's synthesis of rationalism and
empiricism (see
Epistemology: introduction
), where it is noted that the
subject has no direct access to external reality, and can only develop
knowledge by using fundamental in
-
built cognitive principles ("categories")
to organize experience. One o
f the first psychologists to develop
constructivism was Jean Piaget, who developed a theory ("genetic
epistemology") of the different cognitive stages through which a child
passes while building up a model of the world. In cybernetics,
constructivism has b
een elaborated by
Heinz Von Foerster
, who noted that
the nervous system cannot absolutely distinguish between a perception and
a hallucination, since both are merely patterns of neural exc
itation. The
implications of this neurophysiological view were further developed by
Maturana

and
Varela
, who see knowledge

as a necessary component of the
processes of autopoiesis ("self
-
production") characterizing living
organisms.

Constructivist mechanisms are not limited to higher level learning or
discovery of models, they pervade all
evolutionary processes
. The difference
between Lamarckian and Darwinian evolutionary theory is just that
Lamarck assumed that the environment somehow instructs an organism on
how to be adapted. Darwin's view emphasized that an organism has

to find
out for itself, by trial and error. A similar conceptual transition from
instruction to construction took place in the theories of immunity: the
organism is not instructed in any way how to produce the right antibodies
to stop the invaders, as was

initially believed, it needs to generate all
possible combinations by
trial
-
and
-
error

until it finds a type of antibody
that works. Once such an antibody is discovered, the "knowledge" about
how to fi
ght that particular infection remains, and the organism becomes
immune. The conceptual development from instructionism to selectionism
or constructivism is well
-
described in Gary Cziko's book
Without
Miracles: Universal Selection Theory and the Second Darwinian
Rev
olution
.

Since constructivism rejects any direct verification of knowledge by
comparing the constructed model with the outside world, its most
important issue is how the subject can choose between different
constructions to select the "right one". Withou
t such a
selection criterion
,
constructivism would lapse into absolute relativism: the assumption that
any model is as adequate as any other. The two most often used criteria are
coherence
, agreement between the different cognitive patterns within an
individual's brain, and consensus, agreement between the different
cognitive patterns of different individuals. The latter position lead
s to "
social constructi
vism
", which sees knowledge solely as the product of social
processes of communication and negotiation (the "social construction of
reality"). We reject these positions are unduly restrictive, and take a much
more pragmatic stance, where we note that the
adequacy of knowledge
depends on
many different criteria
, none of which has an absolute priority
over the others. People can very well use incoherent models, over which
there is no agreement with other
s, but which still are valuable for
adaptation to a complex world. These criteria will include at least subjective
coherence, intersubjective consensus, and (indirect) comparison with
"objective" environment.