Brain-Mind Institute for Future Leaders of Brain-Mind Research

Home | About | Why BMI? | Why Me? | Programs | BMM | Press | Committees | Webinars | Classes | Founding | Login

On 10/9/11 3:28 AM, Juyang Weng wrote:

Dear colleagues,

You are invited since either you are a leader of a related discipline or your work is very related to the tentative mission of the new Brain-Mind Institute.   We expect that you would his or her one of the future leaders of brain-mind research, at least within your home discipline in your geographic region or internationally.   That is why we need your expertise in shaping this new institute. This is a starting group for the founding email meeting.  At any time, if you like to be dropped out of this email list, please simply send a private email to Weng. Please also suggest names you think that we should include.  All: please always use the latest updated email list when you send your emails to the evolving list.

 (1) The tentative mission of the BMI (all about BMI is open to change):

Collectively, the human race seems ready to unveil one of its last mysteries — how its brain-mind works at computational depth. From the knowledge required for a recently synthesized grand computational picture of the brain-mind and other partial computational pictures, it seems that the research community urgently needs a large number of leaders who have sufficient knowledge in at least six disciplines — Biology, Neuroscience, Psychology, Computer Science, Electrical Engineering, and Mathematics.

While increasingly more researchers are converging to this high-impact brain-mind subject, they face great challenges: Existing educational and research infrastructures, although some are multi-disciplinary, are not meant for this brain-mind scale (6 disciplines) at computational depth. The proposed Brain-Mind Institute (BMI) provides an integrated academic and research infrastructure for future leaders of brain-mind research, at least across the 6 disciplines. The Brain-Mind Institute (BMI) is a new kind of institute, not limited by boundaries of disciplines, organizations, and geographic locations.

(2) According to interests expressed through this email “meeting”, we plan to form the BMI Advisory Board, the BMI Program Committee, and the MSU Steering Committee before the first BMI Distinguished Speaker Series (DSS) talk with webinar which will be held in two weeks, Monday Oct. 24, 2011:

(3) Unless explicitly stated otherwise, your email contributions will be published at the “founding-mails” page at the BMI web site:
The email-address list itself will not be made publicly available.  Please provide your vision and input to shape this new kind of institute.

(4) Please give your expert opinions about how this Institute can be most useful to you, your advisees, your colleagues, your institute, and your discipline.   Please provide your answers to the following questions:

(a) Academy:  What (summer?) academic programs do you suggest?  What subjects? Do you like the credits of the BMI courses to be transferable to academic programs in your home institution? How many credits each course should have?  Is the laboratory component necessary?  What benefits should BMI provide to attract the best international instructors to teach the courses in the summer?  Do we need textbooks suited for anybody in a related discipline?  In addition to courses, what else should BMI do in terms of academy?

(b) Research and beyond:  What research environment should the BMI provide as part of the above summer academic programs, such as associated conferences (panels? keynote talks? paper sessions? special sessions? workshops?).  Do you think that the BMI should or can provide teaming opportunity for possible future research collaborations?  What BMI can do to facilitate communication and cross-fertilization while each of the 6 disciplines has very different research goals, research agenda, styles, terminology or even different meanings for the same terminology?

You do not need to address all.  Unconventional ideas are especially welcome.

Best regards,



On 10/9/11 7:51 PM, Michael Arbib wrote:

Dear John:

I share Alexei's concerns. Your Website does not have a clear statement of purpose on the home page.

Perhaps you can update the Website so that  the About page makes clear:

a) What the Institute will offer as an interdisciplinary graduate program at MSU. Would a student get an MS or PhD from the Institute? If not, what would the Institute offer in the way of seminars, courses, special access to cross-departmental faculty groups?

b) What the Institute will offer to people outside MSU.

c) At present, there are many places with excellent constellations of faculty in the areas you are targeting. What makes MSU special? Who are the excellent people at MSU? Who are the members from outside MSU who have already agreed to play an active role, and what is that role?

d) And there are a number of societies and journals dedicated to the same goals as your Institute. What will your relation, if any, be to these?

Once your Website can address all these points, perhaps it will be time to send out another mailing. And perhaps not. It is too early to tell.  At the moment, the whole thing seems very preliminary, and your email seems more like an attempt to get endorsements than a well-targeted attempt to get feedback on a well-formulated plan.

But best wishes for your work at MSU, whatever the outcome.



On 10/9/11 9:34 PM, Juyang Weng wrote:

Dear Michael,

Thank you for a great set of questions.  Although our BMI proposals have attempted to address almost all your excellent questions, we still do not feel that we have all answers to all your questions.  At this point, I provide only material that encourages discussions.

(1) You asked, would a student get an MS or PhD from the Institute? 

(a)  If the answer is no, why are your students motivated by learning 6 disciplines?  One may say, well, he or she will be at a unique position in today's competitive job market.  But, without a well recognized certificate of some sort,  does the institute serve our students sufficiently?   

(b) If the answer is yes, what is it?  We have seen various joint degrees in many institutions.  Is a joint degree from BMI considered a competition to existing programs (e.g., neuroscience program or cognitive science program)?   Further, why is a faculty member motivated by learning 6 disciplines?  

(2) You asked, what makes MSU special?  I quote from

Why Located at MSU with International Participation?

Like any academic institution, the BMI needs a stable, annually recurring educational infrastructure for its academic programs.

MSU played an instrumental role in initiating an integrated research paradigm for understanding brain-mind — known as
Autonomous Mental Development (AMD), natural and artificial. This paradigm now has a growing research community represented by the ICDL conference series, the AMD Newsletter and the IEEE Transactions on Autonomous Mental Development. MSU has the Cognitive Science Program, the Neuroscience Program, and a number of NSF funded centers (e.g., the Institute for Research on Mathematics and Science Education, the CVRC center, and the BEACON center), as well as the related Colleges and Departments. As the first land-grant university in US and a research-intensive university, Michigan State University (MSU) has the required international reputation, connections, and educational infrastructure that the new institute needs.

The institute will meet its full potential as an international resource by maintaining a stable support base for recurring annual operations. MSU has the necessary international and national relationships that have been built and nurtured over 15 years to meet the scholarship needs of the larger community. The MSU administration regards this education and research entity as an important component of its broad plan to enhance MSU's reputation amongst its peers and to meet the needs of its constituents.

(3) You asked: What will the Institute offer to people outside MSU?

Let us discuss your question (1) first, since this question depends on the answer to your question (3). Each reader of this email may ask himself or herself:  "Do I have sufficient infrastructure at my home institution for learning latest computational brain-mind related knowledge in all 6 disciplines?"

(4) You asked: What will be BMI's relations with a number of societies and journals?

We have societies like AAAI, IEEE CIS, INNS, Cognitive Science Society, Society for Neuroscience, and so on.  Shall the institute apply for technical sponsorship? We have even more computation oriented journals, IEEE TAMD, Adaptive Behavior, Neural Networks, Neural Computation, plus a long list of robotics journals, although not all of  them deal with brain-mind per se.

(5) You stated: At the moment, the whole thing seems very preliminary, and your email seems more like an attempt to get endorsements than a well-targeted attempt to get feedback on a well-formulated plan.

There was some preliminary plan on the BMI test web site.  Some members of the BMI Steering Committee asked to remove them totally so that we can start from "scratch".   We complied.  Since you asked, and we hope that the members of the BMI Steering Committee will feel better to provide some components of the plan back to the password protected BMI test site for us to discuss and shape.
username: xxx
password: xxx

Please give your inputs to re-shape these very preliminary components.




On 10/10/11 8:25 PM, Dorian Aur wrote:

Dear John,

Social interaction can play a fundamental role in the process of fast reshaping education determined by a constant rapid change in this new environment for science.  A new direction may seem to emerge and I feel that web tools, mobile technologies can be highly effective, low- cost modalities  that can provide  immense advantages since they  do not limit education  to a particular  geographic location.

Example: Many  previous   models have reinforced a digital like (action potential) based computation in the brain and have partially  neglected the ability of neurons to process information. Contrary to common belief these interactions that occur within neurons define a more powerful (non-Turing) model of computation present in the brain

An interactive dialogue between various scientists is required to reshape this field and BMI can have a distinct role to foster interaction between different international groups in an attempt to answer important questions in selected fields and provide the required education.

In addition, for several decades computer science has shaped the research in biology and neuroscience (even the action potential is digital) however, the time has come to build a different approach that “can change computing forever, even if most computer scientists don’t know it yet”.

Nothing can ever be perfect in the beginning, however  by using social media and web tools, the newly created BMI can encourage critical public interventions, can reshape itself over time to create a vibrant international forum required for education without any dogmatic preference for certain schools of thought.



On 10/11/11 12:01 PM, Juyang Weng wrote:

Dear Dorian,

You raised a related new direction.  You also have had some views about neuronal coding that is not familiar to many researchers doing work on neural networks and brain.  If BMI provides a social interaction platform for new ideas, e.g., through a conference during a weekend, what procedures can BMI put into place to guarantee adequate tolerance to unfamiliar ideas, ideas that reviewers do not subscribe to?

I noticed what has happened to Daniel Shechtman, the 2011 Nobel Chemistry prize laureate:

"For months he tried to persuade his colleagues of his find, but they refused to accept it. Finally he was asked to leave his research group. Shechtman returned to Israel, where he found one colleague prepared to work with him on an article describing the phenomenon. The article was at first rejected, but finally published in November 1984 — to uproar in the scientific world."




On 10/13/11 8:15 PM, Juyang Weng wrote:

Dear all,

I would like to thank those who sent individual emails or called us to
give their supports, suggestions, and ideas.  I refrained from sending
all the emails as they are not meant for all.   However, we are grateful
to those who provided inputs.

In US and other countries, there are various summer schools.   For
example, MBL, Biological Discovery in Woods Hole, founded 1888 as the
Marine Biological Laboratory, offer some impressive biological
courses.   A typical course example is a summer course of 8 weeks, with
tuition $4975,  not including room and board.  Jim Olds told me that
some Nobel laureates studied in MBL.

As another example, a 3-credit graduate course at MSU has a tuition of
$1595 for in-state applicants only.

We hope to pay instructor his/her home institution rate to attract the best
instructors from the world.   Therefore, the cost of instructor amounts to
a large part of the tuition.  How do you think about the following

3 weeks for each course: 3 credits (typical in US)
MSU rooms: $35/night for less than 30 days, $30/night for more than 30 days.

First 3-week session:
   Each person can take one of the three courses
    (Optionally, first two weeks can be through distance learning for
    those who like to reduce the cost of room and board.)

Weekend, 2-day conference for workshops, panels and posters (all the
course takers attend the conference)

Second 3-week session:
   Each person can take one of the three courses
   (The later two weeks can be through distance learning for those who
   like to reduce the cost of room and board.)

One-course registration with conference: $1145
Two-course registration with conference: $2000
Conference only registration: $195

This way, we hope that the cost of one-course registration with the conference
is not too much different from a one-week conference trip.  If we successfully
receive federal and industrial supports, some approved applicants can pay less.

There is a large neural network conference in the week of July 11, 2012.
MSU summer rooms are available till Aug. 4, 2012.  It seems that we can
fit the 2 3-week sessions into June 24, 2012 - August 4, 2012.
The 2-day conference may be held in the middle of the two sessions:
July 14 - July 15, 2012.

All the above ideas are very preliminary and tentative for discussion.

What do you think?  Please feel free to send your ideas to all.




On 10/15/11 2:02 PM, Juyang Weng wrote:

Dear colleagues,

I would like to thank George Stockman and Paul Spears who have set up
the bmi mailing list, so that any people can sign on and sign off
bmi mailing list as they wish.  Currently, the bmi list has over 2000 names.

If you have not received the following email but like to be included
in the bmi mailing list, go to

Best regards,


On 10/14/11 5:03 PM, wrote:

Brain-Mind Institute to launch with webinar by Distinguished Professor
Jay McClelland, 24 October.

MSU contacts:  George Stockman, Prof. Emeritus of MSU/CSE (517-449-0342)
                               John Weng, Prof. of Computer Science and

(The following release should be put on the BMI website and should be
released to appropriate channels in the MSU administration and
publicity. )

How does the brain work, and how does the mind emerge? Perhaps the
answers are at hand given the current progress in several scientific
fields? Researchers interested in brain-mind science are forming a
Brain-Mind Institute to foster better understanding of how the brain
works and ?creates the mind?.  Founding members of The Brain-Mind
Institute are internationally known experts and invite others to join
them in seeking a computational model of the brain consistent with
knowledge from biology, neuroscience, psychology, but empowered by
knowledge in computer science, electrical engineering, and
mathematics. The BMI proposes to develop new interdisciplinary summer
courses, hold summer workshops with panels, and prepare graduate
students and researchers to understand foundational work in six
fundamental disciplines ? biology , neuroscience, psychology, computer
science, electrical engineering, and mathematics. (The BMI
is in formation and has the goal of being a global institute and not a
local MSU unit.)

A Distinguished Speaker Series has been launched to begin BMI
activities and will be broadcast live via webinars to those who
register online. Webinars will begin October 24 with Prof. Jay
McClelland, Director of The Center for Mind, Brain and Computation of
Stanford University and following on October 31 with Mriganka Sur,
Paul E. Newton Professor of Neuroscience at Massachusetts Institute of
Technology. The live broadcast will be from Michigan State University
between 5 PM and 7 PM with moderated questions from text or email.
Presentations will be available for later viewing by those who cannot
participate in the live webinar.  To follow the BMI on the web and to
register for the webinars,  gets you to
the BMI web site with the current webinar schedule, talk abstracts,
upcoming activities, and organization plans for the future.

BMI mailing list

To subscribe or unsubscribe go to

and enter your e-mail address in the provided box and confirm your action by responding the the e-mail sent by listserv.


On 10/17/11 12:40 AM, Stan Franklin wrote:

Hi John,

I'm not sure what you're asking me re interest in the 6-discipline idea.

Do I think these six disciplines could benefit from cross fertilization? I certainly do. The best stuff is often found in the chinks.

Am I personally interested in research in these areas? I have forty published papers in mathematics and many more than that in computer science/artificial intelligence? I have joint papers with several psychologists and serve as editor of Topics in Cognitive Science. I have joint papers, both experimental and theoretical with a biologists. I have joint papers with one neuroscientist and am just beginning a collaboration with another. I've submitted grant proposals with computer engineers, am just now working on another such, and am currently guiding a doctoral student in biomedical engineering. My research has not intersected with physics unless you count an every increasing interest in non-linear dynamics. I guess it's clear that I'm interested in such interdisciplinary research.

Am I interested in the 6-discipline certificate program you're instituting? I'm certainly interested as an outside observer in seeing it succeed, because I think it has much to offer to science, particularly to AMD, to AGI, to BICA. And, you have a great track record in gathering good people and making their efforts succeed. Beyond being an outside observer? It depends on what. I'm a cognitive modeler (neither a symbolic nor a connectionist one by the way) with a active research program ongoing so as to have many irons in the fire, sometimes I think too many. Still, I'm always on the lookout for collaborative ways to forward my research agenda on how minds work. (I think of a mind as a control structure for an autonomous agent, be it human, animal or artificial.)

This message may be more of an answer than you wanted.

All the best,



On 10/18/11 10:52 AM, Yoonsuck Choe wrote:

Dear John,

Thanks again for the invitation to the BMI.

I have one suggestion. I think what distinguishes BMI from other existing programs (such as the Neuromorphic engineering workshop, and other neuroscience-oriented summer courses) could be found in the strength built up in the ICDL-EpiRobo community. Initially, it would be good to focus on this expertise to build momentum and recognition, and eventually pan out to the 6-discipline idea.



On 10/18/11 1:15 PM, Juyang Weng wrote:

Dear Yoonsuck,

Thank you for raising this good question.   The following is mainly from a few BMI proposals co-authored by multiple Co-PIs.

We know that each of the 6 disciplines runs its own summer schools on a regular basis.  For example, the Society for Neuroscience (SfN) and the International Brain Research Organization (IBRO) run summer schools in neuroscience every year, which mainly cover material in one of the 6 disciplines --- neuroscience.

There are several successful summer workshops, summer programs and summer schools that are multi-disciplinary.   The major related ones include: the Telluride Workshop and Summer School on Neuromorphic Engineering, the Summer Program organized by the RIKEN Brain Science Institute in Japan, and the Theoretical Neuroscience & Complex Systems Summer School run by the Frankfurt Institute for Advanced Studies (FIAS).

There have also been a few within-university programs with "brain" and "mind" in their name.    They include: The Zanvyl Krieger Mind/Brain Institute (MBI) a free standing institute at the Johns Hopkins University with strong connections to the Krieger School of Arts and Sciences and to the School of Medicine; The Kavli Institute for Brain and Mind (KIBM) at University of California at San Diego, which focuses on human cognition;  The Brain Mind Institute of the School of Life Sciences of at Ecole Polytechnique Federale de Lausanne, Switzerland.  There are various multidisciplinary graduate programs at different universities, e.g., Carnegie Mellon Univ., Columbia Univ., CalTech, MIT, MSU, Stanford, UCSD, UIUC, Univ. of Michigan, Wayne State Univ. that address their own specific multi-disciplinary goals.

The BMI program does not compete with those existing programs, as the goals are different.  There are some major differences between the above programs and the proposed BMI:

- Prerequisites: Many existing summer institutes offer tutorial style seminars on selected topics suited for domain experts.  The material is difficult to understand by senior members and graduate students in all the 6 disciplines.   Many courses in existing multi-disciplinary within-university programs (e.g., Cognitive Science and Neuroscience Programs) have similar prerequisites. The proposed BMI summer school would offer a unified 6-discipline program  suited for all qualified applicants --- courses that do not require prerequisites other than a bachelor degree.

- Scope: The above summer programs and within-university programs do not offer the entire scope of the 6 disciplines.  Not all researchers have realized the necessity of this 6-discipline scope yet.  The need for the breadth and depth in all the 6 disciplines seem urgent when sufficient knowledge are already available for humans to resolve this grand brain-mind puzzle.

- Teaching material: The existing summer programs and within-university programs do not provide teaching materials in the 6 disciplines for such a wide variety of researchers. The teaching material to be produced from the proposed BMI summer institute will be unique, specially tailored to break such disciplinary barriers and incorporating latest advances in brain-mind understanding and modeling, with an emphasis on computational integration of brain and mind.

- Research environment:  BMI would provide a research environment that is truly in the scope of the 6 disciplines.  The BMI summer conferences that run with the BMI summer courses are part of such a 6-discipline research environment.   Brain-storming, teaming activities, and other activities would be proposed by BMI summer conference participants.

A major academic goal for the proposed summer institute is that the proposed BMI 6-discipline certificate for brain-mind is achievable for every applicant with a bachelor degree, regardless if the degree is in social science, natural science, or engineering, in a time frame of about 2-3 years. The planned first summer at the BMI summer institute is 2012.  Depending on individual availability in the summers, we expect that a researcher with a bachelor degree in one of the 6 disciplines needs an average of 2-3 summers to complete the BMI required breadth and depth in all 6 disciplines.  Those who take less than 2 courses in each summer may take slightly longer time to receive a BMI 6-discipline certificate.  As part of a 6-discipline certificate, disciplines that an applicant is familiar with can be passed through BMI standard exams, which opens doors for all international applicants whose bachelor degrees may be from very different programs in each country (BMI does not have the resource to check the quality of every bachelor degree itself).

However, this line of ideas is unconventional, still preliminary, being shaped by this BMI discussion group and the BMI committees currently taking shape.  Many people are working together, openly and behind the scenes.  Let us all thank them.



On 10/20/11 2:07 PM, Juyang Weng wrote:

Dear all:

After talking to some of my colleagues, we here kick of a BMI debate via this email on
Many of you on this anonymous list told me that they are interested and want to be posted.  However, we will use this
anonymous list sparely.   If you want to keep posted about this debate and other BMI activities, sign on bmi mailing list
at or simply Google it with key words like "BMI mailing list MSU".
Once you receive email from the mailing list, you can post simply via reply.   BMI mailing list is a moderated list to avoid
unrelated emails.  If there are sufficient interest, BMI might host a live web debate in a few weeks.  Post your views!

The following email I sent to Dave Touretzky is the kick-off for the BMI debates.  I will provide some interesting examples soon.

On 10/20/11 12:59 PM, Juyang Weng wrote:
Hi Dave,

I read some of your papers about hippocampus, which are very interesting.  Let me inject some basic but probably very controversial ideas you probably will reject.  If you do not mind, I will post this discussion to the BMI mailing list.   The main purpose is to attract more talented researchers to this important brain-mind subject.  

How about looking at the brain from a top system point of view?  I believe that top (but detailed) theory is powerful, since the brain basically does signal processing (not in the traditional sense).   Maybe with this view, our future design of experiments could be  more productive?  Let me start from one example:

One of your papers is "Synaptic Learning Models of Map Separation in the Hippocampus", Neurocomputing, 32:379-384, 2000.   The co-authors wrote: "If the perforant path projection to CA3 functions as a pattern completion mechanism, and the DG projection via the mossy fibers performs pattern separation (O'Reilly and McClelland, 1994), then ..."

My new perspectives about the brain benefited from such local views, but I think that such local views can also benefit from the entire brain-mind point of view, in the sense of a giant Finite Automaton (FA).   This brain FA is not handcrafted, but rather developed, since all phenotypes emerge from a single cell (zygote).   So, I model such a developmental FA as the Developmental Network (DN).  Then, the Hippocampus is simply a very small part of a giant DN.  According to how the DN works, I predict the following:  If we focus on a small part (e.g., Hippocampus) of this DN, we definitely will get hopelessly lost, like a hiker in a forest without a global map.   He can see some local phenomena from where he stands, but he did not see the entire forest. 

Focused, per-phenomenon discoveries have been prevailing in the brain science literature in the modern science, with few exceptions (Charles Darwin is one).  This is probably because only such papers can be accepted and funded in the modern time.  Although those phenomena are useful, they are piece meals.  Now, there seem to have enough pieces to put the grand puzzle together.  I have established what a DN can do in real time, by modeling the brain-mind from the entire FA (DN) point of view.  Since all pieces of DN seem to fit what we know about the brain science, the brain should not be less efficient than a DN.

You can say that this is just fantasy, but I have a series of rigorous proofs.

Daniel M. Wolpert said at SfN 2009 that the over 1400-page long volume of "Principles of Neural Science" by Kandel et al. could be much condensed if we could model the entire brain in computational theory.   I hope that the DN theory can help that condensing process.

A major infrastructure problem is that what I talked about above spans at least 6 disciplines.   Meaningful conversations are extremely difficult.  If you feel angry or insulted by my above text, I feel that it is partially because of this huge divide.

I am giving a CC to Jay, as his work was cited.

Best regards,




On 10/20/11 3:02 PM, Ali Minai wrote:


I think that a developmental perspective is crucial if we're ever going to understand how the mind emerges from the brain, or how the brain-body system works. In fact, I would say that we have to include not only development but also evolution - not only how the zygote develops into a functional animal, but also how simple animals evolve into animals with more complex functionality by using the same modules in myriad ways. I have argued (and am writing a book chapter on this) that the "evo-devo" approach needs to be extended into the third dimension of mental function - asking "what systematic evolutionary and developmental processes allow the emergence of a system capable of mental function. Just as we have the idea of "evolvability", so there must be an equivalent idea of "mentability" (or some such word) that distinguishes systems capable of mental function from those incapable of this. This should then be connected to development and evolution.

All this said, I think that these types of global theoretical approaches complement rather than replace the focused study of specific subsystems like the hippocampus. Of course, I say this as someone who has worked on such systems (including the hippocampus, where Dave's work has been a major influence for me). Both global and parcellated investigations contribute to our understanding. To insist on one or the other would just be an ideological choice.




On 10/22/11 10:15 AM, leonid wrote:

Thank you for the good words.
When you were at MIT we were developing similar techniques. Now you rely on logic. The problem with logic is: How it emerges from illogical neural firings?
We would be glad to get you involved.


On 10/22/11 4:32 PM, Juyang Weng wrote:


Good question:  How logic emerges from illogical neural firings?

Many neural net researchers have a background in electrical engineering or physics, but not much in computer science.  To understand how a new category of neural networks (DN) can do abstraction, one needs to be familiar with the
automata theory typically taught in computer science.  Mathematical logic (proposition logic, first order logic, second order logic, etc.) is useful for understanding what formal logic means. But mathematical logic is not sufficient to explain brain functions. 
Brain functions are not based on mathematical logic. 



On 10/22/11 10:59 PM, leonid wrote:

Thank you for Paul slides. Actually I participated in the same CLION meeting, there were many our friends. Quite productive meeting.
Why brain functions are not based on mathematical logic? - It is a good question. It is possible to show that exponential complexity of symbolic algorithms is related to logic. The proof follows closely Godel's proof of logic inconsistency. In case of an infinite system the result is inconsistency, in case of a finite system (say, a computer, or brain) the result is coombinatorial or exponential complexity.
Logic appears in brain as a result of dynamic processes that start with vague states and converge to near logical states. This process overcomes combinatorial complexity. I published several mathematical modeling papers on this topic. Recently it was proven experimentally in Harvard Brain Imaging Center that this process "from vague to crisp"  is a good model for actual neural processes in visual system during perception.


On 10/23/11 12:43 AM, Juyang Weng wrote:


Yes, I agree with you about the exponential complexity of symbolic inputs and symbolic states.   Here is a more intuitive way for those who are not familiar with discrete complexity theory:

- Two different symbols are simply different.  There is no natural distance between them.  Some methods, e.g. Soar, assign each symbol with a set of handcrafted features, so that any two symbols have distance measured in terms of the handcrafted features.
This leads to well known high brittleness since such a set of handcrafted features is always not sufficient for an open-ended world.

- In contrast, the brain users emergent sensory images and emergent muscle images.  Objects or actions in such images are "continuous" since they arise from the natural world and natural actions.  If our DN model is correct, the brain interpolates between an exponential number of sensory subimages and action subimages, not by mathematical logic, but by associations (spatial statistics).

Many domain experts will laugh or at least doubt when they read the above, but the papers cited at the BMI site have more detail.  Those who have vision to go through the BMI 6-Disciplinary Program will learn rich evidence that support the above explanation.



On 10/23/11 8:07 AM, leonid wrote:

I agree with your descriptions. We might have significant commonality in our approaches. It would be great to find a grad-student or post-doc intersted in studying and combining both.
Do you think it is possible?


On 10/24/11 9:24 AM, Paul Werbos wrote:
> Good mor ing!
> I apologize if my message to John was not so coherent. It was not intended
> as something for the group.
> So maybe I owe you a bit of clarification.
> There are many worthwhile goals out there related to the brain, all the way from
> routine clinical practice to basic science.
> In basic science, I would want to focus energy especially on two "grand challenges":
> (1) The ability to understand and replicate the ability of the basic mammal brain ("the mouse")
> to do what NFS Engineering has called "cognitive optimization and prediction."
> Here I am referring to what you would see if you searched on "COPN" at
> This is a relatively crisp goal, which nonetheless requires some greater cooperation between
> a number of disciplines and approaches.
> (2) "Human potential."
> It is easier in many ways to cooperate towards the first, focused goal. 
> My paper in Neural Networks 2009 basically gives my view of how we could get there,
> and it also stresses the two "subchallenges" -- the cognitive optimization part and the cognitive prediction part.
> I view the image processing work recently discussed by Leonid, by LeCun, by Schmidhuber
> and even by Grossberg as basically part of the "cognitive prediction" stream of this...
> The second one basically tries to go beyond the individual mouse brain ... to the human mind in
> its fullest expression, trying to use what we learn from the mouse as a foundation for better addressing
> larger issues which humans have grappled with for a long time. This is NOT such a crisp challenge,
> but I view it as extremely important, and I often feel bad about how much we have neglected
> many aspects of it. I view issues related to language as one part of the second stream.
> I am not saying which of the two is "the right one." They are both right, maybe for different people in different mixes
> at different times. And so too are a few other important basic challenges which may USE neural networks but are not
> ABOUT neural networks... such as energy challenges, where several new technologies need to work together foir
> best results.
> Best of luck,
>      Paul

On 10/24/11 6:19 PM, Thomas Shultz, Dr. wrote:

Ali and John,

I agree with Ali that we ignore evolution at our peril. It is critically important that we come to understand how evolution shapes the adaptability of organisms and how it interacts with learning and development. All three of these processes concern problems of adaptation, and they share interesting algorithmic ideas and can even compensate for each other. They are not in natural opposition as so many psychologists assume.

I also agree with John, and much of biology, that development serves as a bridge between genotype and phenotype.

Even if one’s primary interest is in building human-inspired robots, one should figure out what to build in, what to give over to development, and what to leave to learning. Learning methods themselves are products of evolution. We need to understand all three adaptive processes, which occur on different and nested time scales, and how they interact.

Natural evolution can indeed take a long time, but we can also study its principles in computational models, much like we do with cognition, development, and learning.




On 10/24/11 6:27 PM, James A. Bednar wrote:
> |  From: Juyang Weng
> |  Date: Oct 23 20:13:47 2011 -0400
> | 
> |     Yes, from a scientific point of view, both development and evolution
> |     are important to study.   From an engineering point of view, however,
> |     the cost of evolution to reach a human-level performance is extremely
> |     high.  Primates have a history of at least 65 millions of years.
> |     This perspective does not rule out any possible benefits of evolution
> |     in engineering studies.  Partial evolution based on development is
> |     still worth studying.
> Even apart from engineering, I think that evolution has a very
> different scientific status than neural development.  Development of a
> lifeform with very impressive brain/mind capabilities (such as a cat)
> can occur over a few weeks or months.  That timescale makes it
> theoretically possible for a scientist to observe and tinker with the
> entire process in order to come to a complete understanding of it as a
> physical mechanism (eventually).  Where brains are concerned, we're
> still very far from such an understanding, but the potential to
> understand it is there, and thus development is fully available as an
> object of scientific inquiry.
> We can also understand many aspects of evolution, and of course
> explaining development beyond the mere mechanics of it will require
> understanding evolvability.  But evolution of advanced life forms
> requires so many generations that it is not something that a scientist
> can either observe or substantially alter in his/her own lifetime.
> Thus many of our hypotheses about the evolution of higher animals are
> untestable.
> Being untestable doesn't mean the hypotheses aren't true, but it does
> mean that evolution can't be put on quite the same scientific footing
> as development.  Whereas I do firmly believe that the only hope that
> we'll ever understand brains is via development.  The adult human
> brain is astonishingly complex in its details, but is nonetheless
> constructed from a relatively compact specification in the genome.  If
> we can start mapping out the smaller rather than the larger end of
> this process, we might have some chance to make headway in our
> lifetimes.
> |     The major problem in many evolutionary models is the absence of
> |     development --- genome is mistakenly taken to be task specific,
> |     corresponding to intelligence directly.   As I understand, the main
> |     purpose of the genome is to regulate development, not
> |     to directly generate behaviors or intelligence.
> Heartily agreed!  But there are at least a few computational
> evolutionary approaches that take development seriously.
> Jim


On 10/29/11 4:42 PM, Juyang Weng wrote:

Ali, Tom and others,

I agree with James Bednar's comments.  Let me provide some more intuitive reasons here so that more people on this list can understand and see the point in the above title.

Evolution is extremely important for the development of human body, which is called morphogenesis at a lower level and only in a more restricted sense (just shape).  This is because the human body must be developed (emerge) from a single cell (zygote).  Different organs in the body deal with very different functions.  Limbs move; the stomach digests; and the heart pumps blood.  All these organs and their functions must emerge from the single cell!   The genome in every cell beautifully regulates this process of body development.  This developmental process is largely still unknown, although a lot of detail has been known in biology. 

The genome does not accomplish this miracle through micro-management" (not just rigid "unfolding" organ structure).  Instead it regulates:  The genome in each cell enables every cell to work autonomously so that all emerged phenotypes successfully assist the emergence of additional phenotypes through every cell's experience!  Designing the developmental program (DP) of the human body is extremely difficult due to the high diversity of body organs and their functions.   There are some studies in theoretical biology, physics and mathematics that discuss this process of morphogenesis (just shape, not function though).  In summary, the DP for the body (e.g., mammals, primates, and human) are extremely complex.  Evolution for a body is interesting. However, since humans can design a body and organs, as many engineered robots and organs have shown).  Evolution for a body is not comparable to engineered bodies, as far as engineering is concerned.

However, the same seems very different with the brain.  A human cannot handcraft a complex brain.  One may argue:  Similar to the high diversity and complexity of body organs, the diversity and complexity of the human brain are higher!  David Touretzky along many readers on this list would probably agree with this argument.   However, I think that this argument is not wrong, but superficial.  Why?  Unlike a body which deals with various chemical molecules (e.g., your food), a brain deals with mainly electric signals!  Your eye ball generates spikes; your ears generates spikes, your skin generates spikes again.  The effectors of the brain (muscles and glands) are driven by spikes too.  This is indeed a beauty of nature.  Therefore, the known knowledge in electrical system modeling, such as those in mathematics, electrical engineering (EE), and computer science (CS), are effective tools for us to understand the brain-mind.   A lot of theories, principles and tools have been developed in the three disciplines that are essential for us to understand the brain-mind.   I am not saying that the current knowledge in these three disciplines are sufficient, but I am saying that the basic knowledge in these three disciplines are necessary if one wants to scientifically understand brain-mind. 

From the above perspective, one can see why I can prove that the new Developmental Network (DN) can abstract as well as any complex Finite Automaton (FA) which is the basis model for all modern symbolic models in artificial intelligence (AI).  My recent article titled "Why Have We Passed `Neural Networks Do not Reason Well'?" will appear in the new INNS Society Magazine, Natural Intelligence.  It should be readable by everybody on this list.  This paper bridges the well known divide between symbolic networks (traditional AI) and emergent networks (neural networks). 

Asim, it gives a detailed solution (with mathematical proofs under a journal review) which implies that there is no brain neuron that purely corresponds to an extra-body concept (e.g., the "Jennifer Aniston" cell, the "Mother Teresa" cell, and the Pythagorean theorem a^2 + b^2 = c^2 cell that Christof Koch talked about in Scientific American 2011).  The first theorem in my above paper indicates that the brain network (modeled by GDN) does not need such concept cells to learn any human society's knowledge modeled by a complex FA, incrementally, immediately, and error-free.  The other two theorems indicate that GDN learning is optimal for noisy data.

I guess that all on this list should be able to understand my above paper, but not all will be able to understand my proofs without
a solid background in CS, EE, and Mathematics.  One cannot understand what the above three theorems imply to brain-mind if he or she has not learned biology, neuroscience, and psychology.   Many researchers doubt and said: "This is YOUR model".  I asked back: "What if it tells how the brain-mind works?"  I did not ask: "Can you afford to miss this opportunity by not starting to learn all the 6 discipline?"


On 12/16/11 2:12 AM, Juyang Weng wrote:

Dear colleagues,

I guess that many of you are now close to the end of the final exam week. It appears that it is time for us to continue to consider the academic scope that the BMI could get involved in. In the last several exchanges of emails, we discussed very interesting issues for a single brain-mind.

Now, I respectfully raise to you the multiple brain-minds issue --- group intelligence.

In the BMI MSU Steering Committee, you can notice that we have some respected experts in laws, economy, psychology, and artificial intelligence (AI).  I do not mean that these disciplines are sufficient for such a complex issue. All I intend to say is that we do have people who are experts in group intelligence.  Please feel free to forward this email to your colleagues who are interested in group intelligence, natural or artificial.

I do not want to conceal any possible disagreements against raising this issue.  Some of my colleagues argued that this subject is not our expertise and a discussion on this broader issue will make us be ridiculed. I understand that such views are well intended.  But, let us be ridiculed, as long as within us there are a few who are interested in such issues and are not afraid of being ridiculed.  I have been ridiculed many times in the last 10 years by our respected proposal review panels and paper reviewers.

The following is a relatively new example. My only paper submitted to ICDL 2011 got the lowest review score possible: All 3 respected reviewers judged that my paper belongs to "definitely reject" category, the lowest category possible.  Consequently, I did not have any paper in ICDL 2011. My ICDL 2011 paper submission was a conference version of the paper
"Symbolic Models and Emergent Models: A review" accepted by IEEE TAMD after many rounds of reviews by the time of submission:

I bet that many people on this mailing list are against the views expressed in my paper.  How can the brain's internal representation be emergent, without any module corresponding to an extra-body concept and without any boundaries of extra-body concept?

Some of you asked about the basic principles based on which the brain-mind works.  We hope that such principles can be useful for us to envision the scientific principles for developing group intelligence, including a company, a country and this troubled world. It is unlikely that group intelligence should work exactly like the biological brain-mind. However, they seem to face the same set of problems.

Some similarities between a brain and a brain group:

A brain:  Each cell does not have the brain intelligence.  Each cell in the brain is autonomous while it interacts with its environment (other cells in its environment and the physical world).  The brain needs to organize rich information from many receptors and many effectors which interact with the brain's external environment (outside the skull) and the internal environment (within the skull).   The collection of cells must learn important events quickly and effectively so that the group of neurons responds intelligently.

A brain group:  Each human does not have the group intelligence.  Each human in the group is autonomous while he interacts with its brain's external environment (other humans in his environment and the physical world). The brain group needs to organize information from the "sensors" and "effectors" of the group which interact with the group's external environment (outside the group) and the internal environment (within the group).   The collection of brains must learn important events quickly and effectively so that the group of brains responds intelligently.

Some differences between a brain and a brain group:

The basic unit of a brain is a cell whose operation is governed by biology (also chemistry, physics and mathematics).  The basic unit of a brain group is a human, whose brain is governed by the brain science.  However, a cell is not as sophisticated as a human brain.  A brain displays the "group intelligence" of many cells.

What does the brain science tell us?

From my latest understanding, the brain does not seems to have a central controller.   Interactions between biological cells are
regulated by the same genome in the nucleus of each cell.   Each cell can be removed from the brain leaving the brain working almost as well.

What does AI tell us?

In AI, if the human designer is the central controller in designing the representation of an AI system, this system is highly brittle and cannot deal with new tasks even if it can learn.   We have recently developed a brain-mind model DN, which is fundamentally different from the conventional wisdom of AI but can also abstract.

What does the US constitution tell us?

Unlike what our politician superficially put, "human rights", "value", "freedom", and "dignity", the US constitution is an example of group intelligence: checks and balances of power.  Separation of church and state is an example --- not letting a static ideology control the operation of US. Looking at China, it still lets a static ideology "socialism" control the operation of China, but the true story is that "a few powerful people" use the name "socialism" to control the entire China.

What does the US anti-trust law tell us?  It is harmful for a group of brains in a powerful company to control the entire sector of the economy.

What does the group psychology tell us?  Some of us only hopes a "wise" leader (well, simple), but others are trying to refine the existing systems to make different people to communicate better and collaborate better (well, complex).

If you like, please provide information about your position and your work on these issues by simply reply to this bmi mailing list.   Please reply to all so that the entire mailing list can learn from your information.

Best regards,


On 12/16/11 4:58 AM, Marvin Minsky wrote:

You would probably win your bet -- except that many people cannot access your paper -- because you didn't include the text. So I have no idea what you are talking about, and I don't want to pay a journal for it. I put my papers on my home page, and do not give exclusive copyrights to journals.

Do you know many cases where groups get better ideas than its individuals? For which kinds of problems does that happen?

On 12/16/11 6:47 AM, Juyang Weng wrote:

Thank you, Marvin, for this suggestion.  I typically do as you.   I guess that this pre-publication case one should be fine too as IEEE TAMD has already put pre-publication versions on its IEEE Explore web.   The new link is below:

J. Weng, "Symbolic Models and Emergent Models: A Review," IEEE Transactions on Autonomous Mental Development, accepted and to appear in vol. 4, 2012. PDF file.

More detail about how network emergent representations are without any rigid module for extra-body concepts but can abstract well is available at:

J. Weng, "Why Have We Passed `Neural Networks Do not Abstract Well'?'', Natural Intelligence: the INNS Magazine, vol. 1, no.1, pp. 13-22, 2011. PDF file

Please air your objections to this list, as they are very important for many others to critically think.  

Yes, I have many many cases where groups get better ideas than its individuals.  Laws for human groups and programs for AI agents all have a long way to go compared with the beautiful ways that the brain generates group intelligence from
100 billion autonomous cells!

For human groups:

(1) China: the first 30 years after 1949.  The nation's learning program (Developmental Program (DP), or constitution first, then laws, and next regulations) was so badly designed that group intelligence in China was among the worst in the world, even worse than the current North Korea.  The Chinese constitution that time was like symbolic representations in AI.  It stipulates a static structure of intelligence --- socialism was the doctrine of the nation, static!   It stipulated that Zhedong Mao is the leader, a central controller of the country's "brain".  Mao had very limited knowledge about scientific ways to run a country.  He did not even know about market economy!

(2) China: the second 30 years after 1949. I used the same country as my example so that other factors are more or less similar.  The current DP of China, since Xiaoping Deng came to power after Zhedong Mao's death in the middle 1970's, was also very badly designed.  Group intelligence in China was still among the worst in the world.  The current Chinese Constitution still stipulates that a single party is the leading party.  It is still like symbolic representation, static in terms of a small module in power.  But within the party, used to be one person who made all decisions, now a few people in political bureau make all the major decisions.  Only this slight change of DP, mainly made by Xiaoping Deng, introduced some dynamics into DP.  Due to this little emergent representation about group intelligence, China moved its Per Capita PPP (Purchasing Power Parity) GDP to around 99 in the world in 2010.  With its large population, this bad number moved China to No. 2 in total GDP.

(3) US: I leave this subject to free discussion, because I will meet more objections.  All the major problems in US, as far as I can see, are due to the systematic rigidity of the US DP currently in place.   I wish that US Congress and Senate would come to the Brain-Mind Institute from the summer 2012 to learn how the group intelligence works in the brain.  But I understand this is hard until US public can see that their coming will fundamentally solve all the major problems in US and in the world.  Subprime crisis in 2008?  Occupying the Wall Street in 2011?  An obvious lack of knowledge about group intelligence.

For machine groups:

(1) Each machine individual:  We need brain-like DP to run our developmental robots. This is a new era that few can see now.  I humbly estimate that all major bottleneck problems of AI have been solved by DN in an integrated way, at least in theory with some exciting experimental results.  Demonstration of human-level performance by machines will take a little more time, as it requires money.  Is there any chance for a symbolic machine to solve major bottleneck AI problems?  Almost none.  Sorry to say.  However, it is easier for symbolic representations to get funded.

(2) Group intelligence in AI:   The future societies consisting of wetware humans and hardware robots will need to go through
the similar periods that the human societies went through in the last 250 years.   However, the speed of progress will be much faster, since the 6 disciplines have accumulated much more facts and knowledge about group intelligence than 250 years ago.
The current delay in due progress is mainly due to the cost of communication (what we are doing now), instead of due to a lack of facts and knowledge.  The cost of communication in China is even higher, as the Brain-Mind Workshop that we are trying to run now at Fudan University in Shanghai has shown.   Researchers in China do not believe that brain works this way computationally, almost like researchers in developed countries as indicated by the respected reviewers who reviewed my ICDL 2011 submission.  

By the way, my proof of DN-FA relationship was submitted to the journal you knew, but it was rejected even after I mentioned your name with a CC to you. The first rejection came without any review; the second rejection came with only a single-person 1-paragraph review.


On 12/16/11 10:33 PM, Paul Werbos wrote:

This subjective of collective intelligence is certainly an important one. Sometimes it works, sometimes it doesn't.
Last year, MANY proposals were rejected here when people said: "We will solve the problem by using a multiagent system.
We all know that the whole is greater than the sum of the parts. We can even prove that our system as a whole will
converge to a Nash equilibrium, so it gives the best result."

The problem is that Nash equilibria are not in general Pareto optima. When each agent maximizes its own thing,
treating other agents like mindless physical objects, without taking advantage of what might be gained by making
deals... the result is often so far suboptimal that it breaks things. The design of games and markets SO AS TO MAKE
the Nash equilibrium BECOME closer to a Pareto optimum is one of the key ongoing challenges, especially in areas
like energy. One of the possible approaches is to design distributed systems which are physically distributed 9made
up of many agents) but mathematically designed to be one integrated sparse system.

Of course, analyzing human brains and societies adds many other dimensions of complexity, beyond the scope of a quick
email. Certainly human groupthink screws up very badly at times, but equally certain there are types of corporate
culture which generate very useful dialogue. (Like my next panels, if I can find the right people...)

Best of luck,


On 12/16/11 11:32 PM, Carlos Gershenson wrote:

> Do you know many cases where groups get better ideas than its individuals?  For which kinds of
> problems does that happen?

See  In this paper they offer some overview of
collective intelligence (attached below): Evidence for a Collective Intelligence Factor in the Performance
of Human Groups Anita Williams Woolley, Christopher F. Chabris, Alex Pentland, Nada Hashmi, and Thomas
W. Malone Science 29 October 2010: 330 (6004), 686-688.
Their results show that groups of people work better not depending on the intelligence of individuals,
but on how efficiently they interact. There is an interesting 6 min related TED talk at

I believe that Prof. Weng generalized the question: any cognitive system can be divided into components,
usually the properties of the system are different than those of its components (e.g. neurons+molecules+energy),
but we usually do not refer to properties of a brain as "group intelligence", even when it is indeed product
of a collection of neurons, etc. It is just a convention.

Best wishes,


On 12/16/11 5:38 PM, hans kuijper wrote:

See also Keith Sawyer, Group Genius: The Creative Power of Collaboration,  Basic Books, 2007.  

Hans Kuijper Joliotplaats 5 3069 JJ Rotterdam 

On 12/17/11 12:37 AM, Vojo wrote:

Group i.e., collective i.e., swarm i.e., community, i.e., ..., intelligence is good
Individual ingenuity is better!

Each one, needs another one.
The group's intelligence is based on what each individual knows, and it is advanced on the differences in each individual knowledge.

At some points some person(s) (member of the swarm, group, collective, community) creates (makes) a crucial difference, The Difference, and the whole group is raised to the another level.
A group, being a swarm, keeps  doing what it has always been doing - it keeps improving its knowledge.
The process of improving may be labeled i.e., called the intelligence, and so the new cycle starts by repeating and improving itself at the higher level again, until, again, the (new or same) individual  makes a crucial difference again.

In short, all advances are always a result of a group-composed-of-individuals intelligence.

It is a subjective matter where one draws a splitting line between the group and the individuals.


Vojislav Kecman

On 12/17/11 10:41 AM, Alex ~`Sandy' Pentland wrote: 

you might be interested in this follow-on to my Science paper.  We found that 50% of the `collective intelligence' (objective performance) of a group can be predicted by the pattern of interaction alone.  Similar results hold in real-world situations (companies, etc) that we have analyzed. 

On 12/17/11 11:18 PM, Juyang Weng wrote:


group intelligence seems not uniformly true across subject matters:  from little-known domains (e.g., science) to common-knowledge (e.g., raising tax).

Laymen and peers are almost always falling behind with new discoveries of science.   Here are just a few examples among a long list:
- Nicolaus Copernicus and Galileo Galilei's  Heliocentrism,
- Charles Darwin's theory of evolution,
- Albert Einstein's theory of relativity,
- peer reviews for paradigm-shift papers and proposals.

What subject matters did your study address?


On 12/20/11 11:02 PM, Rob Goldstone wrote:

Dear Sandy et al.,

Another compelling example of groups getting better ideas than its individuals acting in isolation is the Polymath Project: .  It was created in 2009 by Timothy Gowers.  Not a mathematical slouch even when on his own (he won the Fields Medal in 1998), he wanted a way to make faster progress on difficult combinatorics problems than he could achieve acting solo.  The first major success of this experiment in massively collaborative mathematics problem solving was a "simple" combinatorial proof of the Hales-Jewett Theorem ( in only 6-7 weeks.  It's a nice example because nobody can claim that the problems that the group is collectively solving are trivial, or that any of the individuals would likely have been able to come up with the solutions by themselves.

Of course, one could argue that what Gowers implemented is just a higher-bandwidth and accelerated version of what typically goes on in academic communities, where researchers build on and extend each others' published results.


On 12/23/11 5:02 PM, Irving Engelson wrote:

Carlos -- I know of cases where groups get worse ideas than many of its individual e.g., the US Congress.

On 12/24/11 12:49 AM, Vojo wrote:

Yep, quite correct statement,

but the basic assumption while talking about group's both action and performance is that the group has the same cost (fitness, objective, loss, merit, etc) function. In other words, that it works under same norm. (In science, the norm used is typically to get better results, meaning the result closer to the 'truth', or closer to the true state of the nature as perceived by the group. Important is that each individual works under same norm, which is often the case in 'science').

If, on the other side, subgroups of the group (and there may be 2,3, ..., up to |group| )   have different objective functions, group will rarely be better than many individuals. Example given below is a great one!

On 12/24/11 9:21 AM, Kimball Williams wrote:

Groups with better ideas then their individual members: 
For an interesting perspective, plan to attend IEEE the Southeastern Michigan Section Conference
in April to hear Dr. Jane Prey discuss "Women in Computing: Why are There so Few of Us?".  Among her
research conclusions is the strong evidence that a key factor in successful inovations is group gendere
If you don't know Jane, she recently retired from Microsoft  where she led the Tablet Technologies
in Higher Education Initiative. 
The SEM Section Conference web site is not yet open for registrations, but may be viewed at:
Kimball Williams
NOTE:  Please direct all your e-mail transmissions to me to my IEEE Alias shown
above.  This will ensure that I receive your message. 
Ph: 248-372-8074
IEEE SEM Section Chair
IEEE EMC-Society Past President

On 12/25/11 7:17 AM, Juyang Weng wrote:

Some people suggested that a mailing list is better than direct mailing based on one's research interest, as it allows one to subscribe and unsubscribe as often as one likes. As indicated at the end of every bmi email, to unsubscribe you can simply send an email from the correct email address to  If you have changed your email address recently, go to and use your previous email address.   That site is also used for subscription. 


On 2/11/12 11:21 AM, Juyang Weng wrote:

Dear all,

It is time for us to suggest names for keynote speakers for the International Conference on Brain-Mind (ICBM), Sat. July 14, 2012 - Sun. July 15, 2012.  
Thank those who have provided great inputs to the summer 2012 program!   The preliminary program for BMI 2012 is at:

According to the above preliminary program, we plan to have a 3-week course session right before July 14 and a 3-week course session right after July 15.   ICBM is meant as an integrated and important event of BMI.  ICBM is open to all latest, creative, tested or controversial ideas and results.  

The ICBM keynote speakers are leaders in any discipline whose research results are closely related to brain-mind research.  By default, a part of the ICBM registration fees will be used for keynote speakers.   Other ways are planned for covering the cost of keynote speakers, including federal and industrial funds.

Please suggest names for keynote speakers by sending to and   

We hope to list confirmed invitees as soon as possible, well before the paper submission deadline March 1, 2012.

Important dates: 
Panel proposals: by Feb. 15, 2012
Full papers: by March 1, 2012
Abstracts: by March 10, 2012
Advance registration: March 15, 2012
Instructor applications: April 1, 2012

Best regards,


On 2/19/12 10:19 AM, Kimball Williams wrote:

I am trying to determine if this conference is an IEEE related event.  It appears that it directly relates to several areas of IEEE interest, and could be supported by and in conjunction with several IEEE Societies.  However, I don't see mention of any related IEEE linkages, and I do not find it mentioned on the IEEE SEM Calendar. 
If I have missed the reference, please let me know.  If IEEE has not been involved in this conference, please let me know why. 
Thank You.
Kimball Williams
IEEE SEM Section Chair

On 3/7/12 1:54 PM, Juyang Weng wrote:

Dear friend,

- When I was a graduate student at UIUC in the mid 1980s.  Narendra
Ahuja was teaching David Marr's vision theory and Gestalt psychology in
his Computer Vision course, I raised my hand and commented stupidly:
"Why do we have to learn this kind of material?  All we need to do is
computer vision."  Narendra, still remember?

- In the early 2000s, a research buddy working on computer vision said
to me:  "Do not do development.  It is a myth."

- An AI society recently wrote to us: "ICBM does not fall within the
general parameters of conferences that we typically cooperate with, and
that are closely related to the artificial intelligence field."  Well, I
guess they meant that brain-mind is not closely related to artificial

- A close friend who has already done impressive work on computational
development for several years wrote to me: "we do not even aim to
reproduce what happens in the brain."

Indeed, the above is the way many many researchers are still thinking now.

However, research is not "do not even aim to ...", while one does not
know what "..." is.

You will be greatly surprised that "..." has already solved your current
central research problems beautifully, without doing pseudo-inverse,
without using SVM (or your another favorite tool), without doing batch
learning, without facing exponential explosion, without ...

Dear friend, I will be happy to explain to you in detail when you and I
have taken some courses in BMI together.  Now, unfortunately I cannot
explain to you because you would not simply believe what I said.

You may say "I am actually strongly collaborating with several
neuroscience labs."

If you think that collaborating with a few neuroscientists will enable
you to understand brain-mind.  Wrong.  That is exactly what I complained
about many multidisciplinary projects:  Insufficient.   This world has
many neuroscientists and their publications.  But, this world badly
lacks researchers who have the 6-discipline knowledge.   I wrote to my
friend: "If I were you, there is no other commitment that has a higher
priority than attending BMI."   If you still do not understand why, you
might want to read
Let us know if you have better arguments to refute or support the
arguments there.  Let us communicate.  I have only 2 cents of worth.

Best regards,


On 3/7/12 7:33 PM, Narendra Ahuja wrote:


Yes, I remember your preoccupation with math, and a successful one in a conventional sense. But of course brain is more than that.
Mind I think is way beyond that. I like the contrast and comparison brought out by the article you have pointed to.


On 3/7/12 4:45 PM, Wlodzislaw Duch wrote:

Juyang,   of course we do not aim at reproducing but need to get hints and inspirations that go beyond threshold 
logic or perceptron elements. This is why I wrote "Neurocognitive Informatics Manifesto"  

Summary: Informatics studies all aspects of the structure of natural and artificial information systems. Theoretical 
and abstract approaches to information have made great advances, but human information processing is still unmatched 
in many areas, including information management, representation and understanding. Neurocognitive informatics is a 
new, emerging field that should help to improve the matching of artificial and natural systems, and inspire better 
computational algorithms to solve problems that are still beyond the reach of machines. In this position paper 
examples of neurocognitive inspirations and promising directions in this area are given.  This is just one step, but 
I think an important link to AI community objection.  


On 3/8/12 5:49 PM, Juyang Weng wrote:

Dear (name omitted),

Taking some courses elsewhere?   How many years do you plan to spend on
learning neuroscience before you can understand how the brain-mind
works?  How many years do you have before you are too old to do
research?  Do you know how many years the well respected senior author
of the attached paper (and other well known researchers for that matter)
has learned neuroscience-and-psychology and communicate extensively with
neuroscientists and psychologists?   Longer than I did, as far as I can
see.  The attached paper is for you to exam.  It is a piece of work
after so many years of learning neuroscience and psychology and modeling
with well respected skills and intuition in mathematics.

Unfortunately, the brain is a much larger, deeper, and darker trap, much
more than artificial intelligence as a trap.   If you do not want to get
stuck into "local extrema" for many years and then regret, I suggest
that you immediately take advantage of what has happened here in MSU,
both research wise and education wise.

Sorry, you would not be able to understand what I meant here until you
have taken the highly integrated, known-brain-solution directed, 6
disciplinary program at BMI. I cannot guarantee you can largely
understand how the brain-mind works only by taking multidisciplinary
courses elsewhere, since the senior author appears to have had.  I
estimate that a researcher needs a minimum of four BMI courses to go
through the BMI 6-Discipline Program if he takes two subject tests for
the remaining 2 disciplines (e.g., math and EE).  By that time, he
should largely understand how the brain-mind works (not all the details
which will greatly inspire him to work on).

Best regards,


On 3/17/12 3:50 PM, Juyang Weng wrote:

Dear friend,

Please proceed with your intellectual drive to know, and discover further, one of the most exciting discoveries of our era --- how the brain-mind works at computational depth!  The BMI 6-Discipline Certificate is a "ticket" to get into the door of such exciting picture of knowledge.

I have applied for BMI 811 (Biology for Brain-Mind Research) and BMI 821 (Neuroscience for Brain-Mind Research). I look forward to being a student next to you! I cannot apply for BMI 871 (Computational Brain-Mind), a distance-learning course, since I plan to co-teach it.

The deadline for course applications is this Sunday.  You need to get admitted into BMI first before you can register for the courses at a later day (April 15, 2012 if you want to get a early rate).

The number of applications for a particular course will affect whether BMI will offer the course.  I courage you to apply now to reserve your seat in the courses you like to take.

Note: BMI871 Computational Brain-Mind (distance learning) will enable you to get the first glimpse at the beautiful integrated picture of the computational brain-minds! Normally, it requires all 6 other BMI courses as prerequisites.  But since this is the first year, the instructor will consider approving those who take two BMI courses this summer plus a background check, since BMI 871 immediately follows them.

Also tell your colleagues and advisees about the Sunday deadline.

Some senior researchers were concerned about the exam.  The BMI Exam results are always private and confidential.  You are not required to take the exam for a BMI course.  However, in the future, BMI course certificates will be a highlight on your resume that people respect!  The BMI 6-Discipline Certificate would be more respected!

Please contact me if you have questions.  More detail is available at:

Humbly yours.

Juyang Weng



Last updated: January 2, 2014