Wednesday, June 13, 2018

Robotics, Automation and the Cerebellum

image
(Drake et al. 2010. Gray’s Anatomy for Students 2nd edn)
Abstract

The cerebellum can learn complex sensorimotor tasks using a simple technique called imitation. If you are a roboticist or an automation expert, you will find the powerful supervised learning technique I describe below of special interest because it could potentially simplify your work. What makes this technique so powerful is its sheer simplicity and its ability to a learn complex tasks very fast.

The First, the Last and Everything in Between

Sensors
There are two kinds of sensors in the brain. One kind (poor) is used by the neocortex and the other (rich) is used by the cerebellum. Poor sensors come in complementary pairs, stimulus onset and offset. For example, we may have a sensor (A) that fires a single pulse when the amplitude of an audio frequency climbs above a particular level. The complementary sensor (B) would fire when the amplitude falls below the same level. It so happens that there is a train of pulses between A and B but the neocortex does not care about what happens between them. What matters to it is the precise timing of the first and last pulses. Of course, for every type of stimulus, the brain uses many sensors to handle multiple levels or amplitudes.

Unlike the neocortex, the cerebellum is a hungry beast because it wants it all: the first, the last and everything in between. Thus every sensory input going into the cerebellum is a train of pulses. This might seem like a total waste of pulses but it is actually essential to the learning method used by the cerebellum. Again, for emphasis, I differentiate between the two types of sensors by referring to cerebellum sensors as rich sensors. Single pulse sensors (first and last) are poor sensors.

Cerebellar Neurons


Cerebellar cortical neuronal circuits. Mossy fibers from pontine nuclei etc., send excitatory synaptic outputs to granule cells. A granule cell forms one or a few excitatory glutamatergic synapses on a Purkinje cell, where LTD occurs depending on the activity of the granule cell and a climbing fiber. Molecular layer interneurons (stellate and basket cells) receive excitatory synaptic inputs from granule cells and inhibit Purkinje cells. At inhibitory GABAergic synapses between a stellate cell and a Purkinje cell, rebound potentiation (RP) is induced by climbing fiber activity.
Tomoo Hirano and Shin-ya Kawaguchi
Regulation and functional roles of rebound potentiation at cerebellar stellate cell—Purkinje cell synapses

The main neuron in the cerebellum is the Purkinje cell (PC) which was named after its discoverer, Czech physiologist Jan Evangelista PurkynÄ›. There are approximately 15 million PCs in the human brain. Each PC emits pulses that are used to control a motor effector. They are arranged in tight formations like a forest with lots of parallel fibers running through the dendrites like telephone wires. Each PC can receive signals from as many as 200,000 parallel fibers. Each parallel fiber is a long bifurcated axon of a granule cell, an intermediary neuron that conducts sensory signals arriving on mossy fibers. However, not all of the input signals arriving on mossy fibers have sensory origins. Some are control signals that are used to inhibit the PCs when necessary. These fibers are likely used for task control. They do so via so-called Stellate and Basket cells which make inhibitory synaptic connections with the PCs.

Supervised Learning in the Cerebellum

The second most important entity in the cerebellum is the climbing fiber (CF). There is one CF for every PC. The CF carries training input signals to the PC. Those signals originate from the inferior olivary nucleus in the medulla oblongata which relays motor signals from motor effectors in the spinal cord to the cerebellum.

In order to understand how the cerebellum is trained to perform a sensorimotor task, it is important to know how motor effectors work. An effector is the opposite of a sensor. It, too, has a first (start) and last (stop) pulse and pulses in between. It is attached to a muscle and generates a train of pulses that contracts the muscle for as long as the pulses keep coming. The cerebellum accomplishes motor control via the use of a mix of excitatory neurons, inhibitory neurons and tonic neurons. The latter are neurons that continually generate pulses unless they are inhibited. The exact circuit details are not important and is implemented differently in various animals. What matters are the principles.

Learning in the cerebellum consists of finding parallel fiber inputs to Purkinje cells that activate and deactivate motor effectors at the correct time. The training occurs while the neocortex is going through a given sensorimotor task. The cerebellum learns to faithfully imitate the task. Remember that parallel fibers carry pulse trains from rich sensors. These fibers try to make synaptic connections with as many PCs as possible. To train a PC, the training mechanism only needs to send corrective signals to the PC via the climbing fiber whenever the associated motor effector stops firing. The CF signal will suppress and disconnect any parallel fiber connection that is still receiving sensory pulses. The end result is that only parallel fibers that cause the PC to fire and stop firing at the right time will remain connected.

Once the cerebellum has fully learned a task, the neocortex can just turn it on or off whenever it needs to in order to focus on other important matters.

Applications

This training system can be put to good use in all sorts of applications that require automation. Notice that there is no need for either pattern detectors or a conventional multi-layered neural network. Lots of simple rich sensors will do the trick. Sensors are essentially connected directly to motor effectors. Potential applications can range from self-driving trains, cars and buses to self-flying aircrafts and self-navigating ships. The learning system simply learns by imitating human operators.

Robots might be a little harder to train. It would require a human trainer to wear a harness fitted with special sensors that can record precise movements. These could then be used as training signals for the robotic cerebellum. I expect training to be extremely fast.

Coming Soon

In an upcoming article, I will describe how I got my understanding of the cerebellum. Stay tuned.

Thursday, June 7, 2018

I'm Rather Busy But Cerebellum Post Is Coming Soon

"Behold, I stand at the door and knock"

"If anyone hears my voice and opens the door, I will come in to him and will dine with him, and he with me." Believe it or not, this metaphor from the Book of Revelation is the essence of supervised sensorimotor learning in the cerebellum. It is as simple as it is powerful. If you are into robotics, you will not want to miss this. Stay tuned.

Monday, May 28, 2018

"I Will Spew Thee Out of My Mouth" or Why the Cerebellum Cannot Speak

Abstract

I changed my mind about writing an article on the purpose of cortical columns. Too risky in my opinion. Instead, I decided to write about the cerebellum, an important but a much less disruptive part of the brain. In this article, I argue against the hypothesis promoted by some in the neuroscience community that the cerebellum contributes to speech generation. Caveat: This article is for believers only.

The Zombie in the Back of Our Head

The cerebellum means little brain in Latin. It is smaller than the neocortex but don't let this fool you. It contains more neurons than the rest of the brain. It is an unconscious supervised neural network that handles a large number of routine but important sensorimotor tasks while the conscious cortex is busy with other matters. Examples are walking, running, balancing, maintaining posture, etc. The cerebellum makes it possible for the brain to multitask.

It is probably best to think of the cerebellum as an automaton, a robotic assistant to the cerebrum, the conscious or volitional part of the brain. The cerebellum can neither learn nor initiate a task on its own. It is entirely subservient to the neocortex and fully dependent on it for its training. It is a mindless zombie in that it does what it is told to do without question. It will make mistakes because it cannot handle new situations that it was not programmed to handle. This is why it's not a good idea to drive or even walk while texting.

No Speech For You

Mainstream neuroscientists incorrectly attribute speech production capability to the cerebellum. They do so for two reasons. First, they do not understand the purpose and function of the cerebellum. Second, they misinterpret clinical data showing speech impairments in patients with cerebellar lesions. The truth is that speech production is a fully conscious phenomenon that requires no input from the cerebellum. Speech difficulties arise only because the conscious cortex cannot focus on more than one thing at a time and must attend to important tasks that it would normally rely on the cerebellum to handle automatically.

People with cerebellar disorders must find ways to compensate for the deficiency. Depending on the severity of the problem, some stop talking altogether (a condition called mutism) because they are forced to focus their attention almost exclusively on sensorimotor tasks (e.g, walking, maintaining posture and balance, etc.) Others learn to speak in a staccato voice (often accompanied by a trembling posture) during which they rapidly switch their attention between speech generation and sensorimotor tasks. It is a form of imperfect multitasking.

Interestingly enough (and this supports my claim that the cerebellum does not produce speech), people with speech impairment caused by cerebellar disorders can sometimes speak normally. Their voices can return to normal if they lie down in a relaxed position which relieves them of the necessity to attend to other tasks.

I Will Spew Thee Out of My Mouth

I initially became interested in the cerebellum after deciphering the occult message to the Church in Laodicea in the Book of Revelation.
14 “And unto the angel of the church of the Laodiceans write: ‘These things saith the Amen, the faithful and true witness, the beginning of the creation of God:
15 I know thy works, that thou art neither cold nor hot; I would thou wert cold or hot.
16 So then because thou art lukewarm, and neither cold nor hot, I will spew thee out of My mouth.
17 Because thou sayest, “I am rich and increased with goods and have need of nothing,” and knowest not that thou art wretched and miserable, and poor and blind and naked,
18 I counsel thee to buy from Me gold tried in the fire, that thou mayest be rich, and white raiment, that thou mayest be clothed and that the shame of thy nakedness may not appear, and anoint thine eyes with eye salve, that thou mayest see.
19 As many as I love, I rebuke and chasten: be zealous therefore, and repent.
20 Behold, I stand at the door and knock. If any man hear My voice and open the door, I will come in to him, and will sup with him, and he with Me.
21 To him that overcometh, will I grant to sit with Me on My throne, even as I also overcame and am set down with My Father on His throne.
22 He that hath an ear, let him hear what the Spirit saith unto the churches!’”
It did not take me long to figure out that the Laodicea message was a description of the cerebellum and that verses 15 and 16 obviously meant that the cerebellum was not involved in activating muscles used by the mouth and tongue. But why? We are told that it is because the Church of Laodicea is "neither cold nor hot." What does this mean? It means that the cerebellum is not controlled by emotions. It has neither likes nor dislikes, i.e., no motivation or goals. It is a zombie. Speech production must always be a volitional and conscious process because it requires intent.

Coming Soon

The cerebellum is a fascinating neural network consisting of many highly specialized sub-networks. As always, the occult texts can pack a lot of amazing information in just a few short verses. This is the power of metaphors. The Book of Revelation gives a detailed description of its organization and function. I'm a little busy right now but I plan to write an article to explain how the cerebellum is organized, what type of sensory signals it receives, how it learns directly from the motor cortex and how sensorimotor programs are activated. Stay tuned.

See Also:

The Yin-Yang Brain: Even Faster Learning Using Spike Timing

Monday, May 21, 2018

The Yin-Yang Brain: Even Faster Learning Using Spike Timing

The First and the Last

In September of last year, I wrote an article to describe a fast method used by the brain to learn elementary patterns using spike timing. I just want to show in this quick post that pattern learning can be much faster if we take advantage of the Yin-Yang or complementary nature of the brain. The reason is that, if everything in the brain comes in complementary-opposite pairs, learning one component of a pair automatically learns its complement. The trick is to always create and connect everything in pairs. Thus every sensor or pattern detector belongs to a dual entity. For example, a sensor that detects the onset (first signal) of a stimulus is paired with another that detects the offset (last signal) of the same stimulus. Likewise, a pattern neuron that detects the movement of an edge in one direction is paired with another pattern neuron that detects the opposite movement.

He Who Overcomes Will Not Be Hurt by the Second Death

The title of this paragraph comes from the occult book of Revelation where I get a huge part of my understanding of intelligence. It is a good idea to think of paired entities as a single object or complementary unit (CU). This way, instead of connecting a sensor to a pattern neuron, we can connect a sensory CU to a pattern CU, thereby killing two birds with one stone, so to speak. This is where it gets interesting. During learning, both connections can fail the concurrency test but if one of them passes, there is no need to continue testing the other connection. One successful test is enough to certify both connections. In other words, two failures (two deaths) are needed to disconnect the CU but a single success is not nullified by the failure of the other connection. I could be wrong but this is my current interpretation of the metaphorical message to the Church in Smyrna.

A similar approach is used in the cortex where all columns are organized in complementary pairs. The end result is that, since not all connections have to be tested, pattern learning in both the thalamus and the cortical columns is much faster than it would be otherwise.
Complementary organization of cortical columns in the visual cortex
Note: I have been thinking of writing an article to explain the true purpose of the cortex and its 100 million cortical columns. I am currently weighing the pros and cons. Hang in there.

See Also:

Fast Unsupervised Pattern Learning Using Spike Timing
Message to The Church in Smyrna
Solving the Mysteries of Reciprocal Corticothalamic Feedback and Cortical Learning
Fast Cortical Learning Using Spike Timing
The Yin-Yang Brain Revisited: Stephen Grossberg's Work

Saturday, May 12, 2018

The Yin-Yang Brain Revisited: Stephen Grossberg's Work

A Funny Thing Happened this Morning

In December of last year, I wrote a two-part article on how I came to understand the Yin-Yang organization of the brain. I have known for a long time that the Yin-Yang principle was the basis of reality and I came to understand that complementarity was absolutely essential to the organization of the brain. Thanks to my research in deciphering the meaning of certain occult texts, I discovered that each hemisphere of the brain consists of two separate but complementary hierarchies. These are symbolized by two olive trees.

I had assumed (wrongly, as it turned out) that I was the only person to have arrived at this understanding. This morning, one of my readers (Spent Death) left a comment on my blog to recommend the work of cognitive neuroscientist Stephen Grossberg. After a quick search on Google, I was blown away by what I found. In 2000, Grossberg published a paper titled, THE COMPLEMENTARY BRAIN Unifying Brain Dynamics and Modularity (pdf) in which he describes a revolutionary model of the brain based on complementarity. This post is not intended to be a review or a critique of Grossberg's work. I merely wish to point out the commonality between his views and mine. Here is what he wrote in the paper's abstract (emphasis added):
How are our brains functionally organized to achieve adaptive behavior in a changing world? This article presents one alternative to the computer metaphor suggesting that brains are organized into independent modules. Evidence is reviewed that brains are organized into parallel processing streams with complementary properties. Hierarchical interactions within each stream and parallel interactions between streams create coherent behavioral representations that overcome the complementary deficiencies of each stream and support unitary conscious experiences. This perspective suggests how brain design reflects the organization of the physical world with which brains interact. Examples from perception, learning, cognition, and action are described, and theoretical concepts and mechanisms by which complementarity is accomplished are presented.
In the same paper, Grossberg offers a hypothesis to explain how the brain handles sensory uncertainty using parallel streams and multiple stages or levels in the hierarchy. I propose a somewhat similar solution which also uses multiple levels and parallel streams but is implemented via feedback pathways in the cortex and the thalamus. If you have any interest in how the brain works, I heartily recommend that you read Grossberg's work on complementarity.

Needless to say, nobody in mainstream AI is thinking along these lines even though some of them claim to base their research on neuroscience. They are lost in a lost world. AGI will not come from the mainstream.

See Also:

Solving the Mysteries of Reciprocal Corticothalamic Feedback and Cortical Learning
Fast Cortical Learning Using Spike Timing
The Two Olive Trees and the Yin-Yang Brain: How My Understanding of the Cortex Evolved Over the Years

Wednesday, April 25, 2018

Occult Knowledge: Thalamo-Cortical Error Correction During REM Sleep

I Know Many Things About the Brain That Brain Experts Do Not Know

In a recent article on cortico-thalamic feedback, I wrote the following regarding the method used by the brain to eliminate bad connections:
The error correction method is straightforward. Every time a minicolumn receives a feedback signal, it strengthens every input connection that just received a strong enough signal. Bad input connections that do not fire on time rarely get strengthened and so remain weak. However, these bad connections are not severed immediately. This happens at night during REM sleep.
Some of you may be wondering how I figured out that bad connections were severed during REM sleep. I mean, I am neither a neuroscientist nor a neurobiologist. Furthermore, nobody in neurobiology knows what really happens to neural circuitry during REM sleep. Well, the answer will surprise those of you who are not familiar with my work. I did not figure it out. I found it. Or rather, I decoded it.

My Source of Knowledge

I am making claims about the brain that will eventually be experimentally corroborated in the lab. Unfortunately, I don't have the resources for that kind of work at this time. Hopefully, some enterprising university research students somewhere in the world might want to jump on the opportunity and make a name for themselves. I get almost all my knowledge and understanding of the brain by deciphering certain ancient metaphorical (occult) texts in the Bible. Here is the passage from the book of Zechariah that pertains specifically to destroying bad connections during REM sleep:
“I will make it go forth,” declares the Lord of hosts, “and it will enter the house of the thief and the house of the one who swears falsely by My name; and it will spend the night within that house and consume it with its timber and stones.” (The flying scroll metaphor, Zechariah 5:4)
The brain has an error correction mechanism (symbolized by the flying scroll) that only works during sleep. This is indicated by the "spend the night" metaphor. The mechanism scans (flies) through memory (symbolized by the whole land elsewhere in the same chapter) and corrects or eliminates two types of bad connections, thieves and liars. A thief is a redundant connection in pattern memory which is located in the thalamus. A liar is a bad connection to a minicolumn inside a cortical column. It is called a liar because its signals rarely agree with the other connections on the same minicolumn.

More to Come

The metaphorical books of Zechariah and Revelation are a treasure trove of amazing knowledge about the brain. It is the kind of knowledge that will transform the world as we know it when it is fully released. One only needs to figure out what the metaphors mean. It is much easier said than done, however. One's own misconceptions about how things should work get in the way. I have been researching it for almost two decades, on and off, and I am still ignorant about many things. The work continues.

See Also:

Solving the Mysteries of Reciprocal Corticothalamic Feedback and Cortical Learning
Fast Cortical Learning Using Spike Timing

Tuesday, April 24, 2018

Thou Sufferest that Woman Jezebel, Which Calleth Herself a Prophetess

Battling Vampires

Quoting symbolic Biblical scriptures is my way of keeping the enemy at bay. It does wonders, like pouring holy water on a vampire, laughing out loud. I have been using this method successfully for years.

The Secret of Motor Learning Hidden in Plain Sight for Centuries
Notwithstanding I have a few things against thee, because thou sufferest that woman Jezebel, which calleth herself a prophetess, to teach and to seduce my servants to commit fornication, and to eat things sacrificed unto idols. And I gave her space to repent of her fornication; and she repented not. Behold, I will cast her into a bed, and them that commit adultery with her into great tribulation, except they repent of their deeds. And I will kill her children with death; and all the churches shall know that I am he which searcheth the reins and hearts: and I will give unto every one of you according to your works. Message to Thyatira, Revelation 2:18-28.
I'm still laughing. Hang in there.