Conscious Mind, Resonant Brain: How Each Brain Makes a Mind

The book Conscious Mind, Resonant Brain: How Each Brain Makes a Mind by Stephen Grossberg was recently released. I’ve read it and it was an eye opening read. Many of the issues that come up on this forum are discussed in that book. Some of the highlights for me:

  • Algorithms for implementing active prediction
  • Algortihms for implementing hierarchy
  • Algorithms dealing with repetition in temporal sequences
  • Algorithms for lifelong, continuous, real time, learning in machines
  • Explanations of learning in place cells, grid cell modules, navigation, and cognitive maps
  • Biological microcircuits that connect algorithms to cortex structures
  • Biological macrocircuits that connect algorithms to brain structures and pathways
  • Computational architectures for goal direction behavior and value/reward learning
  • Support for macrocolumns in the sensory cortex
  • Formal mathematical models of neural network dynamics
  • Computational architectures that support multiple types of learning: unsupervised, supervised, semi-supervised, self-supervised, reinforcement learning, serial learning

There is a strong claim that adaptive resonance theory (ART) has explained and predicted more psychological and neurobiological data than other available theories. I believe that is correct. Just the knowledge about human vision and how that maps to biologically plausible computational algorithms is mind boggling. When you realize Grossberg also has modal architectures for audition, cognition, emotion, and motor control, it can be quite overwhelming! (hence the size of the book)

So much of Grossberg’s work is answering questions that were raised in Jeff’s book On Intelligence that I must admit to some frustration that I’ve only found this work now.

Grossberg is one of the founders of computational neuroscience, connectionist cognitive science, and neuromorphic technology. He has been researching AI for over 60 years. He has more than 80,000 academic citations.

I’m continuing to study Grossberg’s work and if others are also interested in collaborating on that, then please let me know.


Grossberg has been doing various versions of ART for decades, each exploring different aspects of the neural hardware. Some of his terms and definitions were established before the “neural network industry” settled on different models and terminology. This does make it a little hard to read his older works.
That said, he has had decades to work out how much of this stuff works and this book is the culmination of this body of research. Expect that reading & understanding it will be a project.
Very worthwhile, but still a project.
My biggest takeaways from years of reading his work is “resonance with inputs as pattern completion” and the value of habituation in computation.


Grossberg is 81, so this book must be his swan song. A good Sunday afternoon read, like @Bitking noted, it will take a bit of transliteration.


You are right the term ART (adaptive resonance theory) has been used for many things and has expanded in many directions. We can think of it as a theory for explaining how autonomous adaptive intelligence occurs in both humans and future machines. That leads to design principles that enable the behavior of individuals, or machines, to adapt autonomously in real time to unexpected environmental challenges.

There is a radical over-simplification that people can make about Grossberg’s work: it is far more than pattern recognition, we can see that in the list of issues it addresses in my initial post.

To put it in HTM terms, the resonance is how ART resolves the integration of input pattern matching and learning with active prediction. This is basically solving the hierarchy problem that was an open issue for me in HTM. Nobody has been able to provide an implementation of active prediction in HTM but this is central to Jeff’s vision. It is very nice to see how active prediction and hierarchy are synergistic.

The work goes far beyond this too. For example it implements active inference in vision - explaining how we go from retina to 3D object recognition with a sensory-motor system.


I’m delighted that my new book called Conscious MIND, Resonant BRAIN: How Each Brain Makes a Mind has been getting discussed! I did want to call your attention to why the word “conscious” and “resonant” are in its title.

I realize that the word “consciousness” is often used in a loose way. I was lucky to notice, after years of developing Adaptive Resonance Theory with many gifted colleagues, that its processes of Learning, Expectation, Attention, Resonance, and Synchrony always led to detailed representations of individual conscious experiences. Hence Consciousness was always in the mix, leading to the idea that ART clarifies the CLEARS processes.

With this background, I should note that the book explains how humans consciously see, hear, feel, and know things about the world and shows how these conscious representations enable us to plan and act to realize valued goals;

classifies 6 distinct resonances in conscious perception or recognition, how they work, where in the brain they occur, and WHY evolution may have been driven to discover conscious states;

identifies resonances that cannot become conscious and explains why;

identifies brain processes that never become resonant and explains why;

and supports all of these proposals with principled explanations of large psychological and neurobiological databases.

Testable predictions are also provided, many of which have been supported by subsequent psychological and neurobiological experiments.

The book also summarizes a thought, or gedanken, experiment about how any system can autonomously correct errors in a changing world that is filled with unexpected events using hypotheses that are familiar facts from daily life. The result is Adaptive Resonance Theory.


Any thoughts on Graziano’s Attention Schema Theory?

I prefer to let each person form his or her own opinions about work by other people that they may read. I am glad to respond to comments and questions about work that I have done.


Fair enough, I just thought you may have looked at it since it bears similarities to ART.

1 Like


I still remember the first time (decades ago) I saw how ART searched through competing patterns using habituation.
Considering the relative simplicity of the model, the behavior was incredible. This opened up my eyes to the possibilities of neural networks and showed that it is really possible to have patterns “overlap” in the same space. Your work really turned me on to neural networks and away from “symbolic AI” processing.
Thank you for starting me on this path.


Dear Prof. Grossberg, I was going to reread your book a third time before emailing you about this, but since you’re here anyway …

I spent quite a while trying to understand what’s going on with dopamine, RPEs, amygdala, mPFC, BG, and so on. I wound up in a similar place as MOTIVATOR in many respects, especially on the key issue of how RPEs are calculated. But there were other areas where I departed from MOTIVATOR.

So, here’s the blog post setting out what I currently believe:

And here’s the “supplementary information” post, the second half is specifically discussing the differences between what I believe and MOTIVATOR:

If you have time and interest, I would be very interested in critical feedback, and sorry in advance if I described MOTIVATOR wrong or disagreed with it for dumb reasons. :slight_smile: Thanks in advance, Steve Byrnes


Dear Steve Byrnes,

You write a great deal about your own modeling efforts in your attachments. Congratulations on this serious endeavor!

And thanks very much for buying and thoroughly reading my new book

My own work with multiple colleagues on reinforcement learning, motivated behavior, and related affective neuroscience topics is, however, just lightly sketched in my book, if only because one could devote several books to these topics alone.

Below are listed some articles that can be downloaded from my web page about these topics, ordered in chronological order.

I also provide their urls below for your convenience. The articles are all self-contained and explain different sets of data.

I am not aware of any other models that can provide principled and unifying explanations and predictions about this range of psychological and neurobiological data about cognitive-emotional dynamics in both normal subjects and clinical patients.

If yours can, please let me know what data you can explain that have not already received a unified explanation in the articles below that I will list in several subsequent emails so that HTM Forum will hopefully post them.


Hi Professor Grossberg,

Thanks for writing your book. I am slowly working through it. It describes the complexity so well, but to follow your points that means jumping back and forth, which is difficult in such a heavy book. I wonder if you have considered doing a web variant of it to help people move easily from the point to the detail, to the examples etc.?

Best regards


1 Like

There is a Kindle version and we can run the Kindle app on a computer/tablet.

I think Dr. Grossberg’s links may have been removed due to him being a new poster on the forum.

@moderators Is there anyway to recover them and/or promote him to a status that would allow him to try again?

1 Like

I bumped Dr. Grossberg’s permissions up to member.


There was recent work suggesting that out of pattern completion prediction could emerge Scott free

Is Prediction Nothing More than Multi-Scale Pattern Completion of the Future?

After reading on intelligence and other works, I do believe that pattern completion is fundamental. Intelligence is akin to navigation in conceptual space, or symbolic space. Via pattern completion a map like hierarchical network is formed optimally arranging the information to which the mind is exposed to. This internal ‘database’ or web of information can be used to rapidly navigate the landscape of patterns, concepts, or ideas, towards goal states, it is navigation through the abstract world of information(information being the basis of existence).

I will give an account of what I think is going on. The 4k HDR image set, is vast yet finite not infinite, but it contains all images of all math journals past present future of all possible minds in all possible worlds, so too for biology, physics, computational science. Is there order, we know from the fraction of the whole we have that amidst the noise order exists. And akin to the library of babel, there is the index of babel.

I believe that there is likely fractal, or hologram like structure wherein fractions or portions or units of the whole network contain aspects of the whole network. The brain’s algorithms likely function as such an optimal information processing and storage set of rules, creating an optimal data structure for navigating through patterns, or symbolic structures, towards desired end states(goals provided).

Will definitely buy and read this latest work. I believe consciousness intrinsic to information, but whether it is spatiotemporal or merely spatial information is a good question. It appears this book comments on that. I take time to be an illusion, and believe in blocktime as I hear most physicists do. I do believe consciousness and the nature of qualia is explainable.