School of Computer Science THE UNIVERSITY OF BIRMINGHAM The machine in the ghost Machines for intelligent ghosts?

Message to the Philosophy List about Gilbert Ryle
and his notion of 'Logical Geography'.
Aaron Sloman
Last updated: 22 Feb 2007


This message is in the plain text format I use for email. I may change the format later.

Message posted Thu Jan 4 23:10:21 GMT 2007
Subject: logical geography vs logical topography

I've just joined this list (despite having been a philosopher for
at least 45 years) so I am not sure what sorts of postings are
regarded as acceptable. This is a request for comments and
criticisms. This message is also publicly available here:

For many years I thought I understood what Ryle meant by 'logical
geography' e.g. when he wrote in The Concept of Mind (page 7):

    'The philosophical arguments which constitute this book are
    intended not to increase what we know about minds, but to
    rectify the logical geography of the knowledge which we
    already possess.'

I thought I was doing the same sort of thing in much of my own
work, though using different conceptual tools, as explained below.

However I have recently realised that there are two different
kinds of analyses and I think Ryle was talking about one and I
about another, for which I have recently begun to use the phrase
'logical topography'.

The difference can be compared with describing how some terrain
is divided up into regions for some social or political purpose
(logical geography) as opposed to describing the features of the
terrain that exist independently of that division and which, for
some purposes, may subserve or subvert particular divisions
(logical topography).

An example of analysis of logical geography might be an analysis
of a collection of colloquial concepts used to describe different
kinds of stuff (iron, carbon, water, fire, etc.) in some
linguistic community.

A contrasting analysis of logical topography would start from a
scientific theory of what matter is which explains the different
forms it can take, from which we can generate a survey of
varieties of matter, and then consider the various naturally
grouped subsets and fracture-lines which support different ways
of classifying kinds of stuff, e.g. on the basis of their
atomic/molecular constitution, possibly combined with resulting
physical and chemical dispositions.

The use of the atomic theory of matter to generate the periodic
table of the elements would be an example.

(The logical topography does not uniquely identify some 'correct'
logical geography. Whether isotopes should all have different
names, or the same name with different numerical subscripts, e.g.
carbon12, carbon14, is probably a matter of convenience. The
current nomenclature simultaneously reflects the similarities and
the differences between the isotopes.)

Another example is the use of evolutionary history and unobvious
physical and behavioural features rather than superficial
physical appearance and behaviour as basis for classifying
species of animals. This led some things once called fish to be
re-labelled mammals.

That kind of scientific advance can cause us to 'rectify' the
logical geography in a manner that is different from Ryle's way.
His way depended simply on revealing inconsistencies and
confusions in the current system without introducing any
new explanatory theory.

The particular kind of logical topography I have been
investigating over many years is, in contrast, based on an
attempt to develop a (generative) theory of (virtual machine)
information processing architectures underlying different sorts
of minds and showing how different architectures support
different sets of properties, states, events, processes and
causal interactions. (As if we had multiple periodic tables for
matter, not just one.)

Thus for each architecture (an ant's mind, a bat's mind, cat's
mind, a newborn infant's mind, a toddler's mind, an adult
philosopher's mind (various sorts)) there will be a different
logical topography, each supporting different kinds of logical
geographies used by different linguistic communities, depending
on what the people in those communities know or think they know
about the realities of those minds.

(A logical geography  can be used that includes fantasy elements
such as certain romantic or theological notions of 'free will' --
unlike the everyday notion and the legal notion of 'free will'.)

I don't know if all that suffices to make clear what I think the
difference is between logical topography and logical geography
(it's my first draft sketchy specification of some logical
geography of philosophical analysis, based on hunches about the
logical topography of the terrain common to science and

I suspect Ryle never thought about this difference. But I also
suspect that if he had, he might have enjoyed broadening his own
adventures and explorations to include study of logical

Perhaps the idea of some reality underlying the logical geography
was already implicit in his determination to 'rectify' logical
geography, for that implied that something could make a
particular system of logical geography correct or incorrect.

My questions for this philosophy list are:

(a) does this distinction make sense, and if not what's wrong
with it?

(b) is the notion of 'logical topography' just my name for
something that's already well known under a different name?
(My interdisciplinary interests have prevented me keeping up
with all relevant philosophical literature.)

(c) does anyone who knows/knew Ryle better than I do know whether
I am right in saying his focus was on logical geography rather
than logical topography? Or did he perhaps accommodate both
under his label?

I suspect the work he did in later life attempting to develop a
theory of hierarchies of dispositions that could account for some
of the rich and diverse mental phenomena he was studying might be
construed as an attempt to construct an explanatory theory that
would generate the logical topography of the terrain. In this, he
nearly (re-)invented Artificial Intelligence and the ideas of
cognitive virtual machines implemented in physical machines. I
heard him give a couple of talks on that stuff, but I did not
then understand what he was doing, perhaps because he needed
better conceptual tools for the task.

I have an online draft discussion paper on all this here:

(or just give 'logical topography geography' to google).
Comments and criticisms welcome.

I hope I've not spoilt anyone's new year.

Aaron Sloman

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