|Feature Article - March 2021|
|by Do-While Jones|
Even a superficial analysis gets complicated.
Last month, for Valentine’s Day, we saw what a new book about Darwin’s Descent of Man said about sexual selection. It not only exposed the conflict between sexual selection and natural selection, it led us off on a tangent about how political bias affects the theory of evolution.
This month we will look at another chapter in that book. Chapter 2, “Remarkable but not Extraordinary: The Evolution of the Human Brain,” written by Suzana Herculano-Houzel, made us ponder linear control systems, basketball, neural networks, and pseudorandom sequences, because they all show our brains are more extraordinary than she thinks.
And if it turned out that those neuronal loops and circuits in the brain were organized in similar ways across species, then the inevitable conclusion would be that the minds that those brains can generate must not differ terribly in kind, just as Darwin hypothesized. As it turns out, vertebrate brains are much more similar to than they are different from one another in how their neuronal loops are organized into circuitry. 1
This instantly attracted my attention because of my engineering background. I spent decades designing circuitry which implemented control loops. I recognize you probably don’t have this same background, and aren’t instantly interested in the topic, so I’m going to share some interesting information with you as simply as possible which I hope will help you appreciate just how complex our brains are.
But first, let’s look at what Suzana Herculano-Houzel wrote about the subject. She said there is no fundamental difference between human brains and other mammal brains.
Curiously, the idea that the human brain must be fundamentally different from others is contrary to Darwin’s reasoning and resonates instead with expectations of his contemporary Alfred Russel Wallace. But so far, no striking differences have been found that would cause radical changes of quality rather than gradual differences of degree. The list includes synaptic densities that are slightly higher in human brains than in mouse brains, a particular type of glial cell that is larger in the cortex of the former, and myriad genes that have been shown to impact how many neurons are generated in the brain during development. It is unlikely that any single difference can be pinpointed as the difference that distinguishes the human species from any other chosen as a reference. Rather, what seems to matter most is the degree to which differences accumulate and modify the result. 2
Darwin thought that man is somehow special. That’s a thought generally shared by Christians. The secular, evolutionary opinion is that humans are not fundamentally different from animals. The only difference is that our brains work better than animal brains do because they have evolved more. It is a matter of degree, not a matter of design. All brains are basically the same—but some are just smarter than others.
Darwin didn’t know as much about brains as modern scientists do. He thought there was a big difference between a man’s brain and a cow’s brain; somewhat less difference between a man’s brain and an ape’s brain; and less (but still noticeable) difference between a European brain and an African brain. Darwin was wrong.
Indeed, ever since Darwin, nonhuman primates, then non-primate mammals, and more recently birds have been awarded higher cognitive status than even Darwin himself might have suspected. He did consider that similar emotions were expressed across species, and we now know that the circuits underlying anger, fear, joy, and pleasure are very much the same. But shared across humans and other mammals are also maternal care, deceit, self-recognition, planning for the future, playing, learning by imitation, using and making tools, cooperating in problem solving, having a sense of beauty, and even appreciating the taste of food touched by fire. 3
What made our brains evolve? Evolutionists don’t agree. Some think walking upright freed our hands for using and making tools which, in turn, sharpened our cognitive abilities. Others think it was cooperating for problem solving in a social environment. Still others, like Herculano-Houzel, think it was cooking and eating meat. She says,
Once Homo species cooked their food, they overcame the energetic constraints that otherwise apply to larger apes, and the possibility of a larger primate brain opened up to them in a trend that continued until the largest Homo brains had nearly tripled in size. 4
Of course, evolutionists can’t figure out how an ape-like brain evolved into a human brain because it didn’t happen.
The words that triggered this essay were “loops” and “circuits” in the first quoted passage.
And if it turned out that those neuronal loops and circuits in the brain were organized in similar ways across species … 5
As an electronic engineer, I spent decades working with loops and circuits. They were the key components of the smart weapons I helped design.
The two kinds of control systems are “open-loop” and “closed-loop.” They are as different as a urinal and a toilet.
A urinal is an open-loop system. A man pulls a handle and water flows for a predetermined amount of time. There is no feedback. With a little luck, enough water flows to get the job done without wasting too much.
A toilet tank is a closed-loop system. When the lever is pushed, the tank empties. The flapper valve closes and the refill valve opens. As the tank fills, a float measures the water level. When the proper level is reached, the refill valve closes. Luck has nothing to do with it.
In a closed-loop system, a command goes out and a response is fed back to the controller. An open-loop system isn’t really a loop at all. Information flows in one direction out from the controller, and nothing comes back to tell the controller if the desired effect has been achieved or not.
Consider the cruise control on a car. If the car is going too slow, it speeds up. If the car is going too fast, it slows down. Linear closed-loop control systems are notoriously hard to design.
If the gain of the system is too low, the speed won’t be controlled very well. If set for 65 miles per hour, it might drop down to 60 miles per hour going up hills.
If the gain of the system is too high, the system will oscillate violently. If set for 65 miles per hour, it will go full throttle when the speed drops to 64.9 mph, and slam on the brakes at 65.1 mph.
Once upon a time, designing a good closed-loop system was an art only a few could master. Once engineers learned about gain margin, phase margin, poles in the complex plane, lead compensation, lag compensation, lead-lag compensation, and other parameters, linear control system design changed from an art to a precise science which is difficult to master.
Herculano-Houzel’s casual observation that the brain employs neuronal loops made me think of basketball. Shooting a free throw is not as easy as she must think it is. It requires the brain to make precise geometrical calculations.
Basketball is an American sport; but it has been in the summer Olympic games since 1948, so presumably all of our international readers are familiar with it, and know what it means to shoot a free throw. If you tried to shoot a free throw blindfolded (that is, with an open loop) it would be like playing pin-the-tail-on-the-donkey. You need the feedback of a closed loop to make the shot.
Let’s do some math. The free throw line is 15 feet (about 5 meters) from the basket. (International readers, please excuse us for doing all the rest of the math in inches.) 15 feet equals 180 inches. Human eyes are about 3 inches apart (1.5 inches to the left and right of the nose). When the player is looking straight at the basket, his right eye is looking just slightly to the left of center, and his left eye is looking just slightly to the right of center. From the free throw line, “just slightly” equals 0.477 degrees (the arctangent of 180/1.5).
That means a player’s brain must realize that looking 0.477 degrees cross-eyed means the hoop is 15 feet away. That is amazing. But wait—there’s more!
The hoop is 18 inches in diameter, so 9 inches less than a perfect shot would cause the ball to hit the front rim. Since the ball is 9.4 inches in diameter, if the shot was just 4.7 inches shorter than that (a total of 13.7 inches short) the ball wouldn’t even make it to the rim. A 166.3-inch shot would be an “air ball.” If the player’s brain thought his eyes were 0.517 degrees cross-eyed (instead of 0.477 degrees) he would think the basket was 166.3 inches away (instead of 180 inches). Just a 0.04 degree angular error would cause the player to aim 13.7 inches short.
But wait—there’s more! Just knowing the distance to the basket is only part of the problem. The player has to be able to throw the ball along a trajectory that will go that distance. Knowing the weight of the ball and the force of gravity, the player must pick a launch angle and his brain must compute the proper muzzle velocity for the ball to take the proper ballistic trajectory to the basket—in a vacuum.
But since he is shooting the ball in the air, his brain needs to compute the effective drag on the ball. Drag is proportional to the square of velocity, but the velocity of the ball constantly changes (getting slower on the way up, but then speeding up on the way down). Therefore, drag depends upon the velocity profile; but the velocity profile depends upon drag, so iterative computations would have to be made to converge on a solution.
If that sounds complicated and confusing, good! I’ve made the point without having to show more calculations than necessary. The point is that the brain has to interpret precise angular measurements from the eyes to compute the distance to the basket, and then has to compute the direction and force necessary to make the ball go through the hoop. That’s just for a free throw from a fixed location. Often the player has to shoot from different places while moving.
It seems unlikely that the brain would make all those calculations each time. It seems more likely that the brain remembers data from previous practice shots, and uses remembered successful directions and forces for various distances through visualization. Perhaps the brain imagines the ball going through the air into the basket, and remembers how to match the visualization. Even if calculations aren’t involved, the data storage, data retrieval, and pattern matching is mind boggling.
This brings us to the neural networks our brains use to store, retrieve, and process all the data necessary to shoot a free throw. Scientists have tried to create artificial neural networks for years.
Artificial neural networks (ANNs) are comprised of node layers, containing an input layer, one or more hidden layers, and an output layer. Each node, or artificial neuron, connects to another and has an associated weight and threshold. If the output of any individual node is above the specified threshold value, that node is activated, sending data to the next layer of the network. Otherwise, no data is passed along to the next layer of the network.
Neural networks rely on training data to learn and improve their accuracy over time. However, once these learning algorithms are fine-tuned for accuracy, they are powerful tools in computer science and artificial intelligence, allowing us to classify and cluster data at a high velocity. Tasks in speech recognition or image recognition can take minutes versus hours when compared to the manual identification by human experts. One of the most well-known neural networks is Google’s search algorithm. 6
Scientists have been trying to build artificial neural networks since 1958.
Frank Rosenblatt is credited with the development of the perceptron, documented in his research, “The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain.” 7
In that paper, Rosenblatt said,
The theory has been developed for a hypothetical nervous system, or machine, called a perceptron. The perceptron is designed to illustrate some of the fundamental properties of intelligent systems in general, without becoming too deeply enmeshed in the special, and frequently unknown, conditions which hold for particular biological organisms. The analogy between the perceptron and biological systems should be readily apparent to the reader.
At birth, the construction of the most important networks is largely random, subject to a minimum number of genetic constraints. 8
Rosenblatt recognized that brains contain apparently randomly connected layers of neurons which are trained by experience.
The apparently random connection of neurons might appear to be evidence of chance—but those apparently random connections are actually evidence of design. Pseudorandom sequences are used in sophisticated designs. I used them in two different weapons systems which I obviously can’t discuss in detail—but I can explain the concept at the unclassified level using the TREE-SORT.
The apparently random connections of neurons in the brain might not actually be random—they might be pseudorandom. A pseudorandom sequence is a sequence satisfying one or more statistical tests for randomness, but is produced by a definite mathematical procedure. It looks random—but it isn’t. Pseudorandom sequences are used in communication systems where there is a very low signal-to-noise ratio.
I hope you will find the following discussion of how pseudorandom sequences work interesting; but if you don’t, all you need to know is that pseudorandom sequences are real, and they work in systems designed by humans. Therefore, it would not be surprising to find them in biological systems; but it would be surprising if they arose by accident.
When trying to extract a weak signal from a very noisy background (like when you are trying to listen to a distant radio station) you are actually trying to sort information from static. The kind of information doesn’t really matter.
Thinking is the process of sorting through facts to separate information from static. One common sorting algorithm is the tree sort.
A pseudorandom sequence can be used to improve a tree sort. A tree sort is a method of reaching a conclusion through a series of choices. It gets its name from the notion of following a path from the base of a tree out to a particular leaf on the tree. You start moving up the trunk and come to a point where the tree branches. You either take the left branch or the right branch. Then you follow that branch until it branches again, and again decide to take the left or right branch. You keep going up the tree, taking left or right branches, until you finally arrive at a leaf. The conclusion leaf is uniquely determined by the sequence of left and right choices made.
In 1961, Joseph E. Forrester sold a punched-card system called TREE-SORT which he claimed was “The FASTEST and EASIEST WAY To IDENTIFY and STUDY 260 native and exotic trees of the United States and Canada.” 9 I swear, I am not making this up. A guy named Forrester really did sell a set of punched cards based on the tree-sort algorithm to sort trees. I still have mine. Here is a picture of it:
It came with these instructions:
You just stick a knitting needle through a hole corresponding to a fact, and if the fact is true, the card falls out of the deck. You take the stack of cards that fell out and stick the needle through a different true fact, and more cards fall out. You keep doing this until there is only one card left. The example in the instructions told how to find the Eastern white pine tree card.
This is the back of the Eastern white pine tree card.
Here’s the front of the card.
Ignoring the holes on the right side (which are used for sorting by name), I’ve numbered the holes as if they were bits in a 43-bit binary number. A 43-bit binary number has 17,592,186,044,416 different values. Punched at bits 35, 32, 24, and 0, the Eastern white pine card = 0000000100100000001000000000000000000000001, which corresponds to the decimal number 38,671,482,881. There are 259 other cards in the deck, each of which is punched differently (depending upon number of leaves or needles, seeds, et cetera), representing 259 other 43-bit binary numbers.
The brute-force method of identifying a particular tree would be to search a table with 17,592,186,044,416 entries. Table entry number 38,671,482,881 contains the phrase, “Eastern white pine,” 259 other entries contain the names of other trees, and 17,592,186,044,260 entries are blank. Clearly, that’s not a very efficient use of memory.
On the other hand, a 64-bit computer could easily compare the 43-bit number representing the characteristics of a tree to ten 43-bit pseudorandom numbers. Then it could uniquely identify the tree using a 10 x 260 matrix table containing tree names. The 2600 entries in that matrix require much less memory than the more than 17 trillion entries in the table that doesn’t use pseudorandom numbers.
If you knew how, you could build an artificial neural net with 43 inputs and just ten internal stages and you could get 260 correct outputs using pseudorandom connections—but you would have to design the network and then train it.
Here’s what all this has to do with evolution. The seemingly random connections of neurons in the brain didn’t arise by chance. They are a highly-efficient pseudorandom decision algorithm which serves a purpose. One might argue that, since pseudorandom numbers are virtually indistinguishable from truly random numbers, truly random connections might work as well as pseudorandom connections. That could be true—but just having ten pseudorandom (or ten truly random) connections won’t sort trees all by themselves. Random or not, the neurons must be purposefully connected and trained.
Darwin didn’t know about closed-loop control systems. He didn’t understand what visual processing the brain needs to do to control where to throw a basketball. Not only that, he didn’t realize that it takes a closed-loop control system for the player to run without falling over. The brain takes data from the ear and uses it to control leg, back, and arm muscles to maintain balance. Nor did Darwin realize there are internal closed-loop systems controlling blood sugar levels and responses to infections. The human body has many closed-loop systems which must function correctly to maintain life.
Darwin didn’t know how neural networks in the brain allow the basketball player to visualize the trajectory that would take the ball from his hand through the hoop, and Darwin didn’t know how apparently random connections could make that neural network sort through all the possible trajectories almost instantly.
Modern engineers know how to design closed-loop control systems and neural networks. It is hard to convince engineers who have ever built them that closed-loop control systems and neural networks arose by chance.
|Quick links to|
|Science Against Evolution
|Back issues of
of the Month
A Most Interesting Problem (p. 51). Princeton University Press. Kindle Edition.
2 ibid., (p. 54)
3 ibid., (p. 59)
4 ibid., (p. 57)
5 ibid., (p. 51)