Silicon Brains: Less gushy and maybe just as powerful

Recently we reviewed the interesting work of FACETS, a large European collaboration developing hardware-level designs for computer circuits that mimic the architecture of our brain. Another group here in the United States at Stanford University is taking an alternate hard-wiring approach to designing a brain in silico.

They hope to create a computer that works nearly as powerfully as the human brain–and be “affordable” at the same time. In addition, they also anticipate that not only will their work take a step forward to a deeper understanding of human brain function, but it will also provide the computational power to help other neuroscientists better analyze and simulate neural activity to advance their own research.

The research team, lead by Kwabena Boahen, is developing a neuromorphic chip: a computer that is not based on the classic transistor developed in 1947, but instead is composed of individual mini-circuits designed like a human neuron, developed some 250,000 or more years ago. More specifically, the ion-flow regulated in the neuron’s membrane is replicated by electron flow in the silicon device. And with quite a bit of clever foresight, the interconnections between the “silicon neurons” are not permanently hardwired on the circuit. Instead, each silicon neuron is identified by a memory address, like in a typical RAM chip, and their electrical activity is referenced by the controlling software. This allows for the same chip to be soft-wired to model the interconnectivity of any sort of neural network that is desired to be used for a particular computational application.

Read more about the specific details of how the neuromorphic chips are designed, fabricated, and tested at the Brains in Silicon group’s website. [ VISIT ]

Recommended article:
K Boahen, “Neuromorphic Microchips,” Scientific American, vol 292, no 5, pp 56-63, May 2005. [ READ (pdf) ]

DARPA’s Path to The Singularity

On May 5, 2009 DARPA (Defense Advanced Research Projects Agency) announced that it is preparing to begin an exciting new research program that may be the most ambitious and direct effort by the United States Government to to push human technology closer to the edge of the awaiting Singularity. The program is referred to as Physical Intelligence, and DARPA is currently soliciting interested research groups to develop project proposals for submission. The ultimate goal of the effort will be to fundamentally understand the physical phenomenon of intelligence and to then demonstrate the characteristic in a man-made electronic or chemical system.

Although you might have considered taking on this problem yourself this weekend, it’s understandable if a week’s worth of yard work and Mother’s Day preparations took a critical priority. Leaving this project to large governmental agencies and massive academic and industrial collaborations may be the best idea for your personal work-load at this time.

The funding levels for the Physical Intelligence program have not yet been set, as they will be later determined depending on the details of winning proposals. This could be an effective blank check from the Federal Government supporting a potentially mammoth project that would do nothing less than transform humanity. Why go back to th Moon when we could instead solve one of the most fundamental questions of our species. In the meantime, America could certainly regain our stature of being the primary scientific center on Planet Earth.

What is particularly interesting about this solicitation is that DARPA has explicitly limited the theoretical framework from which researchers may pursue the solution to understanding Physical Intelligence. They make the bold claim that the phenomenon of intelligence emerges directly from thermodynamic processes in the human brain or an engineered machine. Any proposal that contains alternate viewpoints will automatically be rejected from consideration for funding.

At first, it may seem that starting with thermodynamics is too limiting for theoretical progress in modeling intelligent behavior. As a basic starting point, the science of thermodynamics looks at characteristics that emerge from a system composed of effectively infinite parts. For example, the measured temperature of your steak flaming on the grill is just the collective measurement of the motion of trillions of meat atoms and molecules. At other levels, the theory models the transfer of energy between systems and measures the slightly odd variable of entropy, which essentially characterizes how messed up the observed system is. In other words, the shattered glass just knocked to the floor by your coordination-lacking infant son has a higher entropy than it did moments before while sitting peacefully on the dinning room table.

But, we aren’t just talking about heat engines that convert a hot flame into mechanical motion and the phase transition we experience every day while boiling water into steam over a hot stove. Thermodynamics and the broader field of statical mechanics represent the fundamental physics that underlie all of the relatively new ideas of self-organization, complex systems, network architecture and many other concepts that are driving the latest in brain science. Maybe DARPA really is on to something theoretical and, even if they don’t know the answers to life’s biggest questions just yet, they certainly know how to keep their funding solicitations general enough to allow for a broad range of scientific collaborators to jump on board … if they are only brave enough.

The Physical Intelligence program is organized around three levels of critical milestones. The first step is to develop a mathematical theory of the thermodynamics of intelligence and then to represent this theory in a producible system. Second, the aforementioned engineered system must be built and successfully demonstrate intelligence. Third, and finally, additional tools must be developed and designed to further analyze and monitor the created intelligent systems.

The other key limitation to this solicitation is that proposers must be able to submit plans that cover not just a portion of these three milestones, but they must be prepared to take the project all the way to home plate. This is Nobel Prize territory, folks, and anyone who is prepared to tackle human species-altering projects must be ready for the ride of a lifetime.

The boldness of the program is nothing less than what would be expected from proud United States scientists, and the American society is certainly ready for another “One small step for man… one giant leap for mankind.” It certainly is an exciting moment to see the interest, dedication, and–of course, most importantly–financial backing of the Federal Government be honed onto the advancement of machines that match, or even exceed, the level of human intelligence that we effortlessly demonstrate every day.

Build a Brain Workshop in the EU

Our brain is like a uniquely powerful computer. It’s in a class of technology that no typical serial or parallel processor today can replicate. Many scientists have tried to develop computer code that attempts to mimic through simulations, such as the current Blue Brain Project, but the computing power for these approaches are becoming immense.

Alternatively, a large European Union collaboration called FACETS has been working on the design, fabrication, and implementation of a new kind of transistor-based computer chip that structurally mimics the neural networks of the human brain.

FACETS project (

The goal of the project is to create a unique computing architecture that uses what we already know of the structure of the human brain as a foundational design concept. The anticipation is that by creating this new hardware, we might gain a significant advancement in computational technology that might keep us moving upward along the classical Moore’s Law path even after traditional transistor-based architectures reach their lowest physical size limit.

The current scale of these brain-like computer chips are far from the level of the interconnectivity of the human brain. At this time, they have developed chips with around 200,000 transistor-styled “neurons” utilizing 50 million mimicking “synapses”. This is a far cry from the human brain’s nearly 100 billion neurons and countless synapses.

This project is not necessarily trying to build a silicon brain… but is wisely trying to take the structural concept of the human brain and apply it to a new hard-wired approach to develop the next-next-next generation of desktop computers.

“Building a Brain on a Silicon Chip” :: Technology Review Published by MIT :: March 25, 2009 :: [ READ ]

Learn more about FACETS (Fast Analog Computing with Emergent Transient States) [ VISIT ]

FACETS Project Presentation (pdf) [ VIEW ]

Brute force engineering may not be the best path to The Singularity

Humanity may undergo an evolutionary phase transition within this century at the very moment our computational creations develop a level of intelligence that surpasses that of the typical human being. This predicted development is referred to as The Singularity. This transitional event might be considered a true evolutionary development for our species because our primary evolutionary advantage over the rest of the natural kingdom–high-level tool-making–is driving our engineering of ever-more powerful robots and computing machines.

However, we do often forge ahead in our technological developments with pure brute force in an attempt to make as much progress as possible in as short a time as possible. Notably, the classic Moore’s Law, an idea suggested by the co-founder of Intel, has accurately predicted the advancement of computing power over many decades. This exponential “law” is still expected to continue, and is a key predictive element for the coming Singularity Event.

However, bottlenecks for technical advancements in silicon-chip development that conforms with Moore’s Law have been foreseen in the past–and have been overcome. Making transistors ever smaller has been the primary brute-force method of increasing computational power, but there must be an ultimate limit: the scale of a single atom.

So, if silicon chips with transistors still larger than a single atom do not provide enough computing power to bring about The Singularity, what fork in the road of this bottleneck might we diverge onto? The human brain clearly does not have a circuitry that mimics the traditional structure of the computer chip. Even massively parallel computing systems do not come close to replicating the network structure of our brain. And it is the morphological structure of our networked neurons that ultimately gives rise to the emergent computational power of the mind… it’s just that we don’t yet understand this complex network structure.

A brief commentary on this potential bottleneck in reaching The Singularity with a call to consider alternate approaches was presented recently in a New York Times guest column. How will we finally reach The Singularity? A new technological approach may be necessary; a new philosophical approach may be necessary; a new, more complete understanding of the structure of our own brains will certainly be necessary.

“Computers vs. Brains” :: Guest Column from The Wild Side, The New York Times:: March 31, 2009 :: [ READ ]

The Mind’s Free Will is more Complicated than the Brain’s Free Will

The notion of Free Will has been debated at least since the days of Aristotle, and the proper identification of what this human sensation really is or how it works remains far from our grasp.

However, a recent fascinating study from Angela Sirigu at CNRS Cognitive Neuroscience Centre in Bron, France was published in Science that discovers a possible bread crumb as to how our brain processes what we sense as Free Will. The research uses direct cortical stimulation in awake patients undergoing surgery to identify areas in the brain that seem to directly link to one’s “desire” to move an arm or a tongue and to the actual sensation of movement… even when no actual movement of a limb occurred.

Pulling a direct connection from this work to the observation of Free Will is like pulling a magic rabbit out of a top hat. Free Will as we personally sense it is so much more than a causal relationship between one neural network in the brain telling another neural network to do something else. In fact, it seems that this very description of direct causality is the antithesis of what Free Will might be.

Free Will is more like … well, it’s more like … Of course, if I could complete this sentence then I would be considered more brilliant than 2300+ years of human thinkers. But, it is certainly a real sensation than human beings have, which is why we’ve been talking about it for so long. It’s a complicated sensation and one that can only emerge from a complicated computational network like our brain.

“Possible site of free will found in brain” :: NewScientist :: May 7, 2009 :: [ READ ]

A little background on Free Will … [ here ] and [ here ]

What do you think?

The Human Brain’s Light Bulb Might be on a Dimmer Switch

Sometimes it just comes to us. Sometimes we have to ponder so hard that we break a sweat. But, whatever sort of conscious considering we feel we are performing while trying to solve a problem, it might be the case that all of the real thinking work is happening elsewhere in our brain… just slightly out of reach.

Recent research from Goldsmiths’ College in London and the University of Houston is trying to electrically monitor and predict the moment in a human brain when the “light bulb” turns on. And, it turns out that the light bulb might be flickering on moments before we even consciously know about it.

Our sub-conscious neural networks are working non-stop. You’re likely breathing steadily while reading and you don’t even realize it. (Don’t think about it too hard, or you might breath out of sync!) The research suggests that this sub-conscious also works hard while we are in the process of trying to solve a complex problem, and it might be the one to figure it out before we are aware of the solution.

It almost makes it seem like our consciousness is working hard not at trying to solve the problem at hand, but rather at trying to access the solution from the depths of our brain. Or, maybe the conscious brain is indirectly guiding or monitoring the unconscious efforts while it’s busy cranking away at the numbers… like a nosy boss who is pushing the office assistant to finish the daily reports before tee-time.

However, experimental evidence that might possibly point to this notion of a top-level system controlling a lower-level “ghost” does bring back those old-age ideas–now considered to be bollocks–of the homunculus of the mind where there is a separate entity or function that is the real brain behind the brain. So, how many personalities do you have, now?

“Conscious and unconscious thought” :: Incognito from The Economist:: April 16, 2009 [ READ ]

Last updated May 25, 2020