NonAlgorithmic Intelligence

Penrose was right. Human intelligence is nonAlgorithmic. But that doesn’t mean that our physical brains produce consciousness and intelligence using quantum mechanics (although all matter is what it is because of quantum mechanics). The parts (even small ones like neurotubules) contain so much mass that their associated wavefunction is too small to exhibit quantum mechanical effects. Here Penrose got roped in by Kauffman thinking that neurotubules were the carriers of the quantum mechanical indeterminacy. They aren’t, they are just too big. The dimer of alpha and beta tubulin contains 900 amino acids — a mass of around 90,000 Daltons (or 90,000 hydrogen atoms — which are small enough to show quantum mechanical effects).

So why was Penrose right? Because neural nets which are inherently nonAlgorithmic are showing intelligent behavior. AlphaGo which beat the world champion is the most recent example, but others include facial recognition and image classification [ Nature vol. 529 pp. 484 – 489 ’16 ].

Nets are trained on real world images and told whether they are right or wrong. I suppose this is programming of a sort, but it is certainly nonAlgorithmic. As the net learns from experience it adjusts the strength of the connections between its neurons (synapses if you will).

So it should be a simple matter to find out just how AlphaGo did it — just get a list of the neurons it contains, and the number and strengths of the synapses between them. I can’t find out just how many neurons and connections there are, but I do know that thousands of CPUs and graphics processors were used. I doubt that there were 80 billion neurons or a trillion connections between them (which is what our brains are currently thought to have).

Just print out the above list (assuming you have enough paper) and look at it. Will you understand how AlphaGo won? I seriously doubt it. You will understand it less well than looking at a list of the positions and momenta of 80 billion gas molecules will tell you its pressure and temperature. Why? Because in statistical mechanics you assume that the particles making up an ideal gas are featureless, identical and do not interact with each other. This isn’t true for neural nets.

It also isn’t true for the brain. Efforts are underway to find a wiring diagram of a small area of the cerebral cortex. The following will get you started — https://www.quantamagazine.org/20160406-brain-maps-micron-program-iarpa/

Here’s a quote from the article to whet your appetite.

“By the end of the five-year IARPA project, dubbed Machine Intelligence from Cortical Networks (Microns), researchers aim to map a cubic millimeter of cortex. That tiny portion houses about 100,000 neurons, 3 to 15 million neuronal connections, or synapses, and enough neural wiring to span the width of Manhattan, were it all untangled and laid end-to-end.”

I don’t think this will help us understand how the brain works any more than the above list of neurons and connections from AlphaGo. There are even more problems with such a list. Connections (synapses) between neurons come and go (and they increase and decrease in strength as in the neural net). Some connections turn on the receiving neuron, some turn it off. I don’t think there is a good way to tell what a given connection is doing just by looking a a slice of it under the electron microscope. Lastly, some of our most human attributes (emotion) are due not to connections between neurons but due to release of neurotransmitters generally into the brain, not at the very localized synapse, so it won’t show up on a wiring diagram. This is called volume neurotransmission, and the transmitters are serotonin, norepinephrine and dopamine. Not convinced? Among agents modifying volume neurotransmission are cocaine, amphetamine, antidepressants, antipsychotics. Fairly important.

So I don’t think we’ll ever truly understand how the neural net inside our head does what it does.

The uses of disorder

There was a lot of shock and awe about a report showing how seemingly minor changes in an aliphatic group on benzene led to markedly different conformations in its protein target (lysozyme from bacteriophage T4) http://pipeline.corante.com/archives/2015/06/18/tiny_and_not_so_tiny_changes.php.

Our noses are being rubbed in just how floppy proteins are, in contrast to the first glimpses of protein structure obtained by Xray crystallography. Back then we knew so little about proteins, that seeing all the atoms laid out in alpha helices and beta sheets was incredibly compelling. We talked about the structure of a protein rather than a structure. Even back then, with hemoglobin (one of the first solved proteins) it was obvious that proteins had to have more than one structure. The porphyrin ring in heme that oxygen binds to is buried deep in hemoglobin, and the initial structure had to move in some way to allow oxygen to find its way in (because the initial structure showed no obvious channel for oxygen). So hemoglobin had to breathe.

We now know that many proteins have intrinsically disordered segments. Amazingly, the most recent estimate I could find in my notes (or in Wikipedia) is this — It is estimated that over 30% of eukaryotic proteins have stretches of over 30 amino acids that are intrinsically disordered [ J. Mol. Biol. vol. 337 pp. 635 – 645 ’04 ]. Does anyone out there know of more recent data?

We’re a lot smarter now — here’s a comment on Derek’s post — “I have always thought crystal structures of proteins/enzymes are more a guide than actually useful. You are crystallizing a protein first-proteins don’t pack like that in vivo. Then you are settling on the conformation that freezes out- is this the lowest energy form? Then you are ignoring hte fact that these are highly dynamic structures that are constantly moving, sliding, shaking, adjusting. Then if you put a ligand in there you get the lowest energy form-which is what it would look like after reaction and before ligand dissociation- this is quite different from what it can look like at other stages of the reaction.”

Here is an interesting example of the uses of protein disorder going on right now in just about every neuron in your body. Most neurons have long processes, far too long for diffusion to move a needed protein to their ends. For that purpose we have microtubules (aka neurotubules in neurons) stretching the length of the processes, onto which two types of motors attach (dyneins which moves things to negative end of the microtubule and kinesins which move things to the positive end).

The microtubule is built from a heterodimer of two proteins (alpha and beta tubulin). Each contains about 450 amino acids and forms a globule 40 Angstroms (4 nanoMeters) in diameter. The heterodimers pack end to end to form a protofilament. 13 protofilaments line up side by side to form the microtubule, a hollow structure about 250 Angstroms in diameter. In cells microtubules are 1 to 10 microns long, but in nerve process they can be ‘up to’ 100 microns in length. Even at 1 micron (1,000 nanoMeters) that’s 13 * 250 heterodimers in a microtubule.

Any protein structure this important has a lot of modifications imposed on it to alter structure and function. Examples include phosphorylation and the addition of glutamic acid chains (polyglutamylation). The carboxy terminal tails of alpha and beta tubulin are flexible and stick out from the tubulin rod (which is why they aren’t seen on Xray crystallography). The carboxy terminal tail is the site of post-translational glutamylation. The enzyme polyglutamylating the carboxy terminal tail of beta tubular is TTLL7 (you don’t want to know what the acronym stands for). It binds to the alpha/beta tubular heterodimer by an intrinsically disordered region of its own (becoming structured in the process), then it binds to the intrinsically disordered carboxyl terminal tails, structuring them and modifying them. It’s basically a mating dance. There is a precedent for this — see https://luysii.wordpress.com/2013/12/29/the-mating-dance-of-a-promiscuous-protein/

So disordered regions of proteins although structureless are far from functionless