Programmer’s Guide to the Brain: With Examples In Python

By Rob Vermiller

In this fanciful book, the author imagines what it would be like to program the brain using the Python programming language. Along the way, he explores the nature of personality, emotion, free will, consciousness, and diversity, and suggests ways for us to prepare for the coming age of robots and intelligent automation.

• Language: English
• Publisher: Lulu Publishing Services
• Release date: Aug 23, 2019
• ISBN: 9781483400006

Book preview

VERMILLER

Copyright © 2019 Rob Vermiller.

All rights reserved. No part of this book may be reproduced, stored, or transmitted by any means—whether auditory, graphic, mechanical, or electronic—without written permission of the author, except in the case of brief excerpts used in critical articles and reviews. Unauthorized reproduction of any part of this work is illegal and is punishable by law.

This book is a work of non-fiction. Unless otherwise noted, the author and the publisher make no explicit guarantees as to the accuracy of the information contained in this book and in some cases, names of people and places have been altered to protect their privacy.

ISBN: 978-1-4834-0001-3 (sc)

ISBN: 978-1-4834-0000-6 (e)

Library of Congress Control Number: 2019909566

Because of the dynamic nature of the Internet, any web addresses or links contained in this book may have changed since publication and may no longer be valid. The views expressed in this work are solely those of the author and do not necessarily reflect the views of the publisher, and the publisher hereby disclaims any responsibility for them.

Any people depicted in stock imagery provided by Getty Images are models, and such images are being used for illustrative purposes only.

Certain stock imagery © Getty Images.

Lulu Publishing Services rev. date: 08/29/2019

INTRODUCTION

Imagine we could program the brain. How cool would that be! We could lift our mood, raise our ambition, cure anxiety, bolster our self-confidence, improve our leadership skills, and alter our worst impulses. We could also create new, artificially intelligent companions and transfer our minds to powerful robot bodies.

So what’s stopping us? First, we don’t really know how the brain works. I believe our current understanding of brain function needs a fundamental revision. We can’t program something we don’t understand. Second, current theories of artificial intelligence (AI) and neural networks—based on our misunderstanding of the mind—are also cast into doubt.

I’m a computer scientist, not a neuroscientist or academic. But I know enough to say that AI—as currently conceived—is not how the brain works. So, in this book, I will intrepidly offer my own, perhaps fanciful, hopefully thought-provoking, alternative—something we can program.

So what’s wrong with our current understanding of the brain?

We know the human brain contains around 100 billion neurons with many more supporting cells. Each neuron is connected to thousands of other neurons via synapses. Each neuron sends messages (electrical pulses) at a frequency up to 200 times per second to its direct neighbors. Scientists continue to devise clever lab experiments to study how human subjects think and behave. They poke and prod the brain and scan it using functional magnetic resonance imaging (fMRI) and other technologies to uncover its secrets (Le Bihan 2014).

Yet what have we learned? Do we know how memory works? No. Do we understand how traits like ambition, shyness, fear, or risk-taking are implemented in the brain? No. Do we know why happiness feels happy? No. Do we know why some people are narcissists or extroverts and others are not? No. Do we know the algorithms we use to identify a potential mate? What really happens in the brain of an ambitious person? How we make plans, learn faces, take risks, and experience awe? No, no, and no.

Instead, here’s how a typical scientist describes the workings of the brain:

The dopamine system is more or less obsessed with keeping us alive. It constantly scans the environment for new sources of food, shelter, mating opportunities, and other resources that will keep our DNA replicating … Dopamine yields not just desire but also domination. It gives us the ability to bend the environment and even other people to our will. (Lieberman and Long 2018)

I disagree. I think describing the brain in terms of molecules (e.g., neurotransmitters like dopamine) completely misses the point. A molecule is simply a bit of matter, a puff of smoke. It doesn’t know anything about food, shelter, or mating. A molecule is not an algorithm or set of instructions. If we gaze at a picture of a serotonin molecule, do we get motivated? No. Are we visually stimulated at the sight of a hormone molecule such as testosterone or estrogen? Not in the least!

You can’t program a molecule.

Molecules are simply messengers—smoke signals—helping to convey orders to an army of ready neurons in the brain. Like a general’s command to charge, the signal itself conveys very little information. Much more interesting is how the receivers of the signal—the soldiers—are trained, what maneuvers they can perform, and the history and usage of their weapons. A general’s order simply unleashes a complex process that’s already in place. A molecule can’t affect us unless our brains are prewired to be affected.

In addition to molecular explanations, neuroscientists also describe the brain in terms of its neural circuits and functional regions:

Pride, shame, and guilt all activate similar neural circuits, including the dorsomedial prefrontal cortex, amygdala, insula, and the nucleus accumbens. Interestingly, pride is the most powerful of these emotions at triggering activity in these regions—except in the nucleus accumbens, where guilt and shame win out. This explains why it can be so appealing to heap guilt and shame on ourselves—they’re activating the brain’s reward center. (Korb 2015)

Again, I think this explanation misses the mark.

Yes, it’s true that the brain is divided into specialized regions such as the prefrontal cortex, amygdala, and cerebellum. Planning seems to occur somewhere in the brain’s frontal lobes. Emotion appears to bubble up from the amygdala, a specialized region consisting of around 12 million neurons. Short-term memory is enabled by the hippocampus. The brain stem is responsible for basic life functions and respiration. The cerebellum helps with coordinated motor control. The cortex—or gray matter—comprises around 20 percent of the brain’s neurons and appears responsible for language, vision, and other higher-order capabilities.

But knowing which brain regions are more active when we engage in a specific activity doesn’t help us understand how the mind is implemented. Describing the brain in terms of its gross anatomy and specialized regions is no more helpful than explaining it in terms of neurotransmitters and hormones.

You can’t program a brain region.

Yet scientists continue to plunge ahead with their current approaches. The latest $100 million scientific research project called MICrONS (Cepelewicz 2016) endeavors to understand the brain by studying a cubic millimeter of a rat’s brain tissue—containing 100,000 neurons and one billion synapses—in the visual cortex, the part of the brain involved in sight. Best of luck to them, but I’m not holding my breath. A similar $1.3 billion Human Brain Project, launched by the European Union in 2013, collapsed after only two years (Theil 2015).

To truly understand how the brain and AI work, I propose that we focus on the activity of individual neurons. They don’t even have to be human neurons. The Open Worm project (openworm.org) studies a small nematode worm having only 300 neurons. Other scientists conduct research on large sea slugs—Aplysia californica—that have 20,000 central neurons in their nervous system, still a manageable number. Focusing on small worms and sea slugs is much more practical and has a much better chance of success in allowing us to understand how the brain works in general. Why? Because in small worms and sea slugs, many or all of the identified neurons have a unique function and carry out the same task across the species (Hoyle and Wiersma 1977).

Like sea slugs and nematodes, I speculate that human brains also contain identified neurons, each having a specific task to perform, although this is not scientifically proven. More radically, I think each neuron is responsible for the same function across the human species.

Think about it for a moment. If we all have the same set of identified neurons, it would explain a lot about how the brain works. Each neuron is accountable for performing a single, specific mental trait or task. Each neuron essentially becomes a computer on a network, able to run algorithms and programs, store information, and send messages to other neurons on the network. Neurons bootstrap with programs and algorithms written in our ancestral DNA. This is very similar to how the internet works today. Why not human brains?

Identified neurons give us something we can program!

When describing this theory of identified neurons, I’ll deliberately ignore the influence of other, commonly ascribed causes for human traits such as genetics, epigenetics, neurotransmitters, and hormones. I believe a neuron-centric view can replace them all.

Any complete theory of the mind should also address where feeling and consciousness come from. I propose that the activity of individual neurons and their algorithms, in the context of powerful emergent effects, is the only explanation needed. The brain’s simulation of reality, through the action of billions of individual neurons, begins to resonate in harmony, and feeling and consciousness are the result.

Many of these hypotheses may seem a little far-fetched, and in truth they are highly speculative and unproven. According to a critic:

Here is what we are not born with: information, data, rules, software, knowledge, lexicons, representations, algorithms, programs, models, memories, images, processors, subroutines, encoders, decoders, symbols, or buffers—design elements that allow digital computers to behave somewhat intelligently. Not only are we not born with such things, we also don’t develop them—ever. (R. Epstein 2016)

Clearly, I disagree. The ideas in this book are falsifiable, and that’s how science advances (Popper 1959). Science is a way of thinking, not a source of absolute truth. Will these ideas ultimately prove to be correct? Perhaps yes, perhaps no. But I think anything is better than the current quagmire of our understanding.

Finally, I’ll discuss the implications of this theory and how algorithms, trait diversity, and luck can answer persistent questions about free will, personal responsibility, and fairness. I’ll address the social issues arising from AI and how we can best prepare society for the changes to come.

It’s a lot to cover, so let’s get started!

PART 1—IDENTIFIED NEURONS

AI Is Not How the Brain Works

Artificial intelligence (AI) is a hot topic these days, offering the promise of self-driving cars and trucks, automated factories, and world-class chess play. Yet the enthusiasm is also laced with foreboding: Will a robot take my job?

Current AI techniques are known by many names, including neural networks and deep learning, a subset of machine learning (LeCun, Bengio and Hinton 2015). AI is used to mine big data sets to look for patterns, predict weather trends, diagnose complex diseases, monitor banking transactions for fraud, identify military targets, forecast stock market trends, convert speech to text, translate languages, enable factory quality control, understand handwriting, and—perhaps ominously—parse social media for consumer and political preferences.

AI can also be trained to recognize faces in photos and videos as we often see on TV police dramas. When a criminal culprit’s face is caught on camera, it doesn’t take long for law enforcement to identify him or her using AI facial recognition tools. AI is very good at finding patterns in big data, especially when millions or even billions of exemplars are available to train the neural network (Loy 2019).

However, as promising as AI might appear for these specific applications, dire warnings of an AI winter are mounting, as current approaches to AI don’t explain how the brain actually works.

To get a deep-learning system to recognize a