School of Computer Science THE UNIVERSITY OF BIRMINGHAM CoSy project CogX project

The deep, barely noticed, consequences of embodiment.
(Ignored by most embodiment theorists)
Invited talk for PT-AI Conference, Thessaloniki, 3 & 4 October 2011
Philosophy and Theory of Artificial Intelligence

Aaron Sloman
School of Computer Science, University of Birmingham.
(Philosopher in a Computer Science department)

Installed: 18 Sep 2011
Last updated: 18 Sep 2011; 19 Sep 2011; 21 Sep 2011; 27 Sep 2011; 28 Sep 2011; 30 Nov 2011
This file is
A PDF version may be added later.
A partial index of discussion notes is in

The deep, barely noticed, consequences of embodiment.
(Ignored by most embodiment theorists)

In just over four decades of thinking about relationships between AI,
Philosophy, Biology and other disciplines I have found that there are a number
of requirements for progress in our understanding that are often not noticed, or
ignored. In particular, our explanatory theories need to take account of a
number of facts about the world and things that live in it, which together have
deep implications for theories of mind, whether natural (evolved) or artificial.

NOTE: Different things are ignored by different researchers.

For example, researchers who emphasis the use of logical forms of
representation, rule-based systems, structure-matching, topological
relationships, or linguistic interaction often ignore the requirement for
perceptual systems (including visual systems) to perceive processes of various
kinds. Some of them may assume that once perception of static structures has
been sorted out, perception of processes can be handled in terms of sequences of
static states.

On the other hand, researchers who emphasise online interaction with
the immediate environment tend to ignore the need to represent and
reason about structures and relationships that are not currently
sensed or acted on, but which might be relevant to explaining things
that have happened, making plans, choosing goals, cooperating with
others, designing new shelters or machines, etc. For instance, they
ignore states and processes referring to what happened or did not
happen in the past, why something happened or did not happen, what
would have happened if something had occurred, what exists in
distant places, what could happen in the future, what's impossible
in a situation and why it's impossible, what would become possible,
or impossible, if some future possibility were realised. And so on.

Some of those who emphasise the production of behaviours during
online interactions ignore all requirements for representing the
contents and processes in the environment in ways that are
independent of how they are sensed or acted on.

Those who assume intelligent agents have to know about the contents of the
environment may ignore the point (emphasised by James Gibson) that some of the
contents of the environment can most usefully be represented in terms of how
they affect possible actions (Gibson's affordances).

Many of those who study the perception of static and changing physical
environments ignore the diversity of other functions of visual perception, e.g.
understanding written communications, How-To diagrams, seeing intentions, likes
dislikes, kinds of effort, difficulty in the actions that others are performing.
(Some of them studied by G. Johansson, using moving point lights.)

Can we come up with an approach to studying perception, action, learning,
reasoning, communicating, and other aspects of intelligence in a way that helps
to prevent excessively narrow focus?

Here are some thoughts:
1. The universe contains matter, energy and information. (For an answer to 'What
is information?' see [1]). Life is intimately connected with informed control.
Almost all processes involving living things, including metabolism, use
information to select among options provided by configurations of matter and
energy, whereas most inanimate matter behaves in accordance with resultants of
physical and chemical forces and constraints.

2. The types of information, the types of control, and the types of problem
for which informed control is required, are very varied, and changed
dramatically in many different ways between the earliest life-forms and modern
ecosystems including humans and their socio-economic superstructures. We need
to understand that (enormous) diversity in order to understand the varieties
of natural intelligence and in order to understand requirements for modelling
or replication in artificial systems.

3. The earliest and most obvious uses of information are in "on-line" control of
discrete or continuous forms of behaviour triggered or guided by sensory
information -- and this may suffice for microbes in constantly changing chemical
soups. Some researchers seem to think that's all brains are for, and some
roboticists aim for little more than that in their robot designs.

4. New, more diverse, and more complex, challenges and opportunities were
presented by changes in physical environments, physical bodies, and types of
behaviours of prey, predators, conspecifics, and inanimate but changing features
of the environment (e.g. rivers, winds, waves, storms, diurnal and seasonal
cycles, earth-quakes, avalanches, etc.).

    Evolutionary transitions
  [A crude reminder of evolutionary transitions where changing environments and changing morphology
  require changing information processing mechanisms, forms of representation, and architectures.]
  For more details see the presentations here:

5. As a result, the types of behaviour, sensory-motor morphologies, forms of
control, types of information, and forms of information-processing became more
and more complex, especially in organisms near the peaks of food-pyramids with
r/K trade-offs favouring K strategies (few, but complex, offspring [2]).

6. In particular, for some species, the relative importance of on-line control
of interaction with the immediate environment declined, in some situations, in
comparison with abilities to store and use information about the past, about
remote locations and their contents, about possible futures, and about the
information processing done by other individuals (e.g. infants, mates, competing
and collaborating conspecifics, prey, predators etc) and by themselves
(self-monitoring, self-debugging, selection between conflicting motives,
preferences, hypotheses, etc.)

7. One consequence of all this was the increasing importance of informed control
of information processing, as contrasted with informed control of actions in the
physical environment. The need to be able to acquire, store, analyse, interpret,
construct, derive, transform, combine and use many different types of
information, including information about information, led to development (in
evolution, in epigenesis and later in social-cultural evolution) of new forms of
encoding of information (new forms of representation) new information-processing
mechanisms and new self-constructing and self-modifying architectures for
combining multiple information processing subsystems, including, but not
restricted to, sensory motor sub-systems [10].

8. The ability to think about possibilities, past and future and out of sight,
touch and hearing, as opposed to merely perceiving and acting on what is actual
became especially important for some species. Some robot developers understand
this, for instance those who work on SLAM [8].

9. The requirements for such mechanisms are closely related to the development
of mathematical capabilities in humans. For a partial analysis of the
requirements see [3]. For links with development of mathematical competences
in children and other animals see [4].

10. Because it is very hard to think about all of these issues, and how
interdependent they are, most researchers (in philosophy, AI, robotics,
psychology, neuroscience, biology, control engineering) understandably focus
their research on a small subset. Unfortunately some of them write as if there
is nothing else of importance, and that has been an unfortunate feature of many
recent waves of fashion in AI, including the fashion for emphasising only
aspects of embodiment concerned with on-line interaction with the immediate

11. That fashion ignores information-processing requirements concerned with
being located in an extended, rich, diverse, partly intelligible universe of
which the immediate environment is a tiny fragment and in which not only what
actually exists is important but also what might happen and constraints on what
might happen[5], along with the invisible intangible insides of visible
and tangible things, and their microscopic and sub-microscopic components.

12. A good antidote for some of this myopia is the work of Karmiloff-Smith on
transitions in understanding micro-domains [6].

13. When we have absorbed all that, perhaps we can attend to the requirement for
much of the information processing to make use of virtual machinery as has
increasingly been required in artificial information processing systems over the
last six decades, including self-monitoring directed at virtual machine
operations, not physical processes -- providing the roots of a scientific theory
of qualia and the like, with causal powers. But first we have to understand the
(mostly unobvious) requirements that drove it all, discussed in [7].
We can describe this as research on Meta-Morphogenesis, the morphogenesis, over
various time scales and evolutionary, developmental, and social/cultural
transitions, of forms of morphogenesis [9].
[1] 'What's information, for an organism or intelligent machine?
How can a machine or organism mean?' In Information and Computation
Eds G. Dodig-Crnkovic and M. Burgin, pp. 393--438, 2011.


[3] Requirements for a Fully Deliberative Architecture (Or component of an architecture)

[4] If learning maths requires a teacher, where did the first teachers come from?

[5] Actual Possibilities, in Principles of Knowledge Representation and Reasoning
Eds L.C. Aiello and S.C. Shapiro, 1996, pp 627--638}.

[6] Annette Karmiloff-Smith, Beyond Modularity: A Developmental Perspective on Cognitive Science, MIT Press, 1992,
(Discussed in

[7] Evolution of mind as a feat of computer systems engineering:
Lessons from decades of development of self-monitoring virtual machinery.

[8] Some exceptions to my strictures

Not all research in AI and Robotics exhibits the kinds of myopic focus on online
interaction with the immediate environment criticised above. Examples include
work on SLAM (Simultaneous Localisation and Mapping), Planning, Mathematical
reasoning, Game playing, and various applications of AI requiring reasoning
about complex, structured, systems.

E.g. find out about SLAM here

[9] Virtual Machinery and Evolution of Mind (Part 3)
Meta-Morphogenesis: Evolution of Information-Processing Machinery
(To be published in a collection of papers on Turing's work.)

[10] There is much work on architectures in AI and Cognitive Science, and many different architectures are proposed
either because people ignore previous work or because different researchers focus on different subsets of
requirements. A partial survey of architectures is available at
My own work (with colleagues at Birmingham) on requirements not just for one architecture, but for a space
of biological architectures of many kinds (the CogAff project) can be found here
There are strong connections with Marvin Minsky's work, which focuses on the special case of a human architecture,
The Emotion Machine. See

For more on all this see

Maintained by Aaron Sloman
School of Computer Science
The University of Birmingham