School of Computer Science THE UNIVERSITY OF BIRMINGHAM CogX project

Requirements for Working Minds: The Meta-Brain Project (MBP)
Is a human brain too complex to be understood by something as simple as a human brain?
(Incomplete draft.)
Aaron Sloman

Created: 18 May 2011
Modified: 22 May 2011
I have been asked to join (provisionally) the EU "flagship" project, the "Human brain
project" (HBP) which has begun to attract some notoriety, in part because of
incompetence and or prejudices of journalists.

Expect the amount of publicity to grow.
Here's a tiny sample thrown up by google on 22 May 2011, following recent announcements
and a conference presentation.
A few internet links: expect the numbers to grow dramatically

Proposal for a parallel, closely related project:
The Meta-Brain Project (MBP), more appropriately thought of as Meta-(Brain-Project)

In proposing the MBP I am not seeking funding. This is just a continuation of what I
have been doing for about 40 years and what I intend to go on doing, as presented in
various online materials, e.g.

It mostly requires analytical thinking, as well as a lot of reading and talking to
people with diverse interests and expertise.

A project about a project

This is a project whose subject matter is not (primarily) human brains, but
    research attempting to understand human brains.

The project attempts to clarify the aims of such research, along with analysing the
criteria by which progress can be evaluated (including the evidence base, the
conceptual clarity, the explanatory depth and breadth, the precision, and the
practical applicability).

In particular, the requirements for adequacy of theories about how brains work are
far from obvious to neuroscientists, psychologists, psychiatrists, etc. There are at
least two different reasons why formulating the criteria for evaluating the HBP will
be difficult:

    (a) human brains are not the only kind. They are primarily a product of millions of
    years of biological evolution during which time very many types of information-based
    control system were produced. Understanding how any one of those systems works
    requires understanding the varieties of problem that were encountered and (to some
    extent solved) during our evolution, and in in part that requires understanding
    similarities and differences between different evolved systems.

    (b) the conceptual frameworks for thinking about what the functions of brains might
    be and how those functions can be met by implemented systems may still be too
    primitive, just as in the mid-20th Century, the understanding of what computers might
    be able to do, and what sorts of technology might enable them to do it, were so
    limited that most of what computers do now was not thought about, and could not be
    thought about.

In short: we don't know enough about (a) what needs to be explained and (b) the
variety of conceptual and other tools required to provide the explanations. The HBP
is one way of reducing these gaps in our concepts and understanding.

The "meta-" project associated with HBP that I am proposing, the MBP, needs to be
highly interdisciplinary.

For example, the MBP requires a high level of philosophical competence, in addition
to deep understanding of the relevant science and technology, including advances in
computer science and computer systems engineering. (Unfortunately most philosophers
have ignored most of this science and technology in their philosophical work, even
though they use the results every day in their private lives and their work.)

Some features (and non-features) of the MBP project

I am not asking for any money: only information and collaboration.

This is a low-budget, informal, collaboration between people in HBP and the rest of
the world.

It will not have a leader -- though I am willing to collaborate with others in driving
the work forward.

In the folowing, meta-goals (and meta-X) refer to goals (and X) of the MBP, whereas other
goals (and X) refer to the goals (and X) of the HBP and its pillars.

Initially MBP and HBP will be inherently opposed insofar as MBP looks for flaws,
gaps, etc. in the work done in HBP, but over time each will inform the other

The MBP will have a very important, objectively based, steering function --
objectively based because all criticisms and suggestions will be based on empirical
(mostly biological), logical, mathematical, and computational results not on personal
preferences and assumptions.

The task will be to collect and attempt to organise in a systematic way:

1. A collection of open-access web-sites of various sorts (e.g. on the HBP web site)
supplemented by freely available online documents, together presenting:

  o changing aims of the Meta-Brain project
    (e.g. the aim of specifying criteria for progress),

  o evidence against which progress in HBP can be evaluated
    (including: videos of children and other animals, reports on experiments, research
    summaries, evolutionary considerations, developmental trajectories, cross-species


    I think Anette-Karmiloff Smith's 1992 book
        Beyond Modularity: A Developmental Perspective on Cognitive Science
    presents many examples of a feature of human and animal behaviour that has mostly
    gone unnoticed by researchers in recent AI, Robotics, Developmental psychology, and
    philosophy, especially recent philosophy emphasising embodied cognition. It is a
    feature that overlaps with what I have been investigating for some time, namely the
    biological competences that made the development of mathematics possible as a product
    of human cognition.

    The key idea is that in addition to Behavioural Mastery in many domains (the
    book presents several domains), humans and some other animals can move on to achieve
    several additional kinds of mastery related to each of those domains (also closely
    related to Immanuel Kant's philosophy of mathematics and to Piaget's research
    published in his last two books).

    These include reflective capabilities -- being able to think about what you have
    done, what you have not done, why you did not do it, what would have happened if you
    had done it, what you could do in future, what the consequences would be if you were
    to do something, including what else it would then make possible or impossible.

    They also include communicative abilities -- both being able to report on such
    considerations to others, and also being able to understand their reports and to
    engage in questions and answers about what is or is not possible, why, and how things
    might be enabled or prevented, or explained.

    AK-S describes the development of these additional competences beyond behavioural
    mastery as "Representational redescription". Whether her explanatory theories are
    correct or not, or perhaps useful first stepts towards good explanations, her
    observations are sharp and important, and should provide part of the theoretical
    background to the task of specifying criteria for evaluating progres in HBP, namely
    can it explain the development and the deployment of these additional comptences.

    For more on this book and why it is important see this personal (still incomplete)
    analytical review/discussion:

  o debates and discussions
    (e.g. about the criteria developed in MBP and how well the methods, results,
    formalisms, mechanisms, etc. of HBP conform to the criteria -- and sometimes failure
    to conform will lead to changes in the HBP, sometimes to changes in the criteria
    defended by MBP)

  o MBP provisional conclusions (about requirements to be met in HBP
    and about the extent to which partial results in HBP suggest requirements are being or
    will be met, or are failing.)

  o criticisms of MBP provsional conclusions, and revisions

  o attempts to apply the evaluation criteria, as they develop, to research in the HBP
    pillars, and work done elsewhere. Often this will require substantial projects -- e.g.
    doing empirical research to find out more about what needs to be explained, or
    building rival systems to point out benefits or inadequacies of systems built in HBP
    pillars. Such work may depend on successful applications for new funding. Work funded
    by HBP itself may be too biased (e.g. because of a wish to get more funding).

  o regular summaries and evaluations of progress in HBP with conclusions regarding future
    plans for HBP

  o reports on newly identified research gaps -- things researchers should be aiming for
    which so far nobody is aiming for (in HBP and similar projects).

  o On going self-criticism of MBP and revision of MBP plans, theories, criticisms, etc.

  o A cumulative repository of specifications of short term, medium term and long term MBP
    criteria for evaluating progress in HBP and other related projects

MEMBER LIST: to be created

Aaron Sloman and ...
    (I can think of people to invite, ... but probably there are many people I do
    not know who could contribute a great deal.)


There's an enormous amount of literature on requirements analysis and evaluation of
various kinds of engineering project. I suspect most of it will not be directly relevant
because the MBP work will address something unusual: requirements and criticisms of
systems that are primarily (a) information-processing systems and (b) simultaneously need
to have a scientific explanatory function and also a useful practical applications. Also a
lot of philosophy of science literature.
    A First Draft Analysis of Some Meta-Requirements
    for Cognitive Systems in Robots
    Aaron Sloman and David Vernon
    What's a Research Roadmap For? Why do we need one? How can we produce one?
    Aaron Sloman, Presentation at euCognition Workshop, Munich 2007