the less than rigorous approach to the study of its use in the generation of mental
activity.
Closer approaches are for example the emulation theory of representation of
Grush Grush (1995) or the model-based sensory-motor integration theory of Wolpert
Wolpert et al. (1995). Grush proposed the similar idea that the brain represents
external-to-mind things, such as the body and the environment, by constructing,
maintaining, and using models of them. Wolpert addresses the hypothesis that the
central nervous system internally models and simulates the dynamic behavior of
the motor system in planning, control, and learning.
We think that we can go beyond using the concept of model-based-mind as metah-
phor or as de facto contingent realizations found in biological brains to the more
strong claim that minds are necessarily model-based and that evolutionary pressure
on them will necessarily lead to consciousness. This article is just one step in this
direction.
2.2 On models
This definition of cognition as model-based behavior many sound too strict to be
of general applicability; in particular it seems not fitting simple cognitive processes
(e.g. it seems that we can have a stimulus input without having a model of it). How-
ever, if we carefully analise these processes we will find isomorphisms between
information structures in the system’s processes —e.g. a sense— and the external
reality —the sensed— that are necessary for the process to be succesful.
These information structures may be explicit and directly identifiable in their
isomorphisms or may be exteemely difficult to tell apart. Models will have many
forms and in many cases they may even be fully integrated —collapsed— into the
very mechanisms that exploit them. The model information in this case is captured
in the very structure of the cognitive process. Reading an effective cognitive system
tells us a lot about its surounding reality.
The discussion of what is a the proper charaterisation of the concept of model is
also very old and plenty of clever insights as that one of George Box: ”Essentially,
all models are wrong but some are useful” Box and Draper (1987). Is this model
usefulness what gives adaptive value to cognition as demosntrated by Conant Co-
nant and Ashby (1970).
There are plenty of references on modelling theory, mostly centered in the do-
main of simulation Cellier (1991); Zeigler et al. (2000) but it is more relevant for
the vision defended here the perspective from the domains of systems theory Klir
(2001) and theoretical biology Rosen (1993, 1991).
This last gives us a definition of model in terms of a modelling relation that fits
the perspective defended in this article: a system A is in a modelling relation with
another system B —i.e. is a model of it— if the entailments in model A can be
ASLab.org / Principles for Consciousness / A-2007-011 v 1.0 Final