Introductory Posts
Particular Anatomical Structures
Organizational Principles
Related Sequences
In 1969, H. Chandler Elliott said
Every brain system grows logically from the tube.
He was right. Today, we’ll learn why.
Primary Neurulation
A blastocyst has no brain. To correct this unfortunate situation, every vertebrate genome contains instruction for constructing a neural tube. This structure emerges via folding:
This process is called primary neurulation. The resultant embryonic structures become the following systems in adults:
In the human embryo, primary neurulation is completed in the fourth week. It turns out this developmental milestone is relatively more prone to error than others – 1 in 500 embryos suffer from neural tube defects.
Vesicle Differentiation
The neural tube does not stay uniformly shaped for long: soon, three bumps (vesicles) emerge. These vesicles appear at three weeks – even before primary neurulation is complete! Informally, these bumps comprise forebrain, midbrain, and hindbrain. Two weeks later, the human neural tube has differentiated into five vesicles.
“Encephalon” means brain.
Vesicle-Structure Maps
Developmentally, the neural tube can be divided into five regions: tel-, di-, mes-, met-, and myel- encephalon.
Structurally, the central nervous system can be divided into three structures: cerebrum, cerebellum, and spinal cord.
We can usefully combine these perspectives:
The brainstem and cerebellum are highly conserved across species. Most biological innovation is driven through the cerebrum (telencephalon).
Here is an anatomical view of the same relationships:
Finally, here is a topological view of these same vesicle-structure maps:
You may notice how this graphic coheres well with my post on cranial nerves.
From this level of abstraction, we are better equipped to create structure-function maps, also known as neural architectures. To be continued.
The brain receives information from three distinct sources:
This infographic explores the third source of information.
Loosely, cranial nerves exist because sensory information coming from the head doesn’t make much sense coming through the spinal cord.
Note that PSNS stands for Para-Sympathetic Nervous System (a branch of the autonomic nervous system).
The autobiographical origin of this post is rather involved:
At the end of this adventure, the following concepts stand out to me as particularly significant:
A few general musings to close:
To be continued. This was fun!
Part Of: [Mental Architecture v1.5] sequence
Overview
This data collection effort will become larger over the coming months. Eventually, I hope to break this up into a more formal knowledge representation system. It would be nice to create (inter-referencing, version-controlled) tables & software models containing:
Implementation Notes
This document is not close to representative of my entire research project, but it is better to have at least something written down.
Finally, please note that Schema Memory, Semantic Memory, and Episodic Memory are not represented in what follows. These memory systems each comprise their own representational format, but following the principle that “modules tend to save their work near where they are computed”.
Without further ado, here is my December 2015 mapping between my mental architecture, and the brain.
Modular Localizations
The following are cognitive networks (or streams, or circuits) which have been fully or partially localized. Here lie 17 distinct networks.
The following are cognitive networks which I have not yet localized. Here lie 4 distinct networks.
The following are cognitive modules which have been fully or partially localized. Here lie 14 distinct modules.
The following are cognitive modules which I have not localized. Here lie 9 distinct modules.
Part Of: [Mental Architecture v1.5] sequence
An summary image like this is, of course, highly tentative & deserves a writeup summary & a summary of extant comparative anatomy data. Maybe in v1.6!
Part Of: Neuroanatomy sequence
Followup To: The Thalamocortical Plasma Globe
Content Summary: 1100 words, 11 min read
Cortical Area & The Obstetric Dilemma
Last time, we learned that the brain is organized like a plasma globe: a sphere within a sphere. Today, we’ll be exploring a technique for reasoning about the cerebral cortex, or “outer sphere”. A few things you should know about this organ:
Does your neocortex have the most neurons? No, that title goes to the cerebellum, whose 100 billion neurons coordinates complex movements. But your brain must do a lot more than motor fine-tuning. Your brain perceives its environment, identifies objects, sets goals, makes decisions, feels emotions, and experiences consciousness. Where does your brain perform these tasks? Primarily in the neocortex! Loosely speaking, the outer sphere does much of the “heavy lifting” for your brain.
Why is the human brain so wrinkled? After all, not all species have brains with this shape:
Consider the following evolutionary pressures, together known as the obstetric dilemma.
As any microchip designer can tell you, wrinkles are a way to increase surface area, while holding volume constant!
Flattening The Lobes
And now, a short story. Cognitive neuroscience is rife with Latin terms for cortical areas. There are hundreds of them: “anterior cingulate cortex”, “fusiform gyrus”, “temporo-parietal junction”, etc. Over the past few years, as I consumed more of the field, I had slowly acclimated to hypotheses regarding the functions and interrelations for such areas. But given my lack of robust anatomical intuitions, these Latin names were just linguistic markers; I lacked an appreciation for geography.
While 3D models of different locations were mildly helpful, the pieces really fell into place when I discovered cortical flat maps.
While the neocortex is like a sheet, in some ways it is more accurate to imagine two sheets. Most people know that the brain has two hemispheres, but fewer know that these hemisphere’s are not (directly) connected? The two halves of your brain do talk to one another, but via a subterranean tunnel known as the corpus callosum.
Okay, here it is: a flat map of the human brain.
The brain is often divided into four to six different sections, or lobes. My map colors each lobe, illustrating the relationships between 3D brain and 2D map.
Here are my five lobes:
Flat maps are an underappreciated resource. But I’ll return to this point some other day. In the meantime, I should mention that these flat maps are imprecise elaborations of the flat maps created by the Gallant Lab. That is, my creations strike me a bit like this:
I am planning to ultimately construct more precise flat maps using tools like Freesurfer. But in the meantime, I’ll stick with my quick-and-easy 16th century cartographic approach. 🙂
Primary vs Association Cortex
Let’s turn now to perception. In Tunneling Into The Soup, we discussed how sense organs (e.g., the rods and cones in your eye) translate the physical reality outside your body into neuron-compatible signals that your brain consumes. To make this intuitively compelling, I employed the following metaphor:
Your eye transmits data to your brain via information highways known as the optic nerves. The traffic of these highways – sense data – captures only a subset of physical reality; call this subset an umwelt. Such highways drain their contents onto landing sites on the cortex known as primary areas.
Aristotle is wrong: you have more than five senses. Balance and body alignment are senses in just the same way as sight, hearing, touch, taste, and smell. I am currently aware of nine different types of sense, or sensory modalities. But of these, three sense modalities are particularly important (and consume more space!) in the human brain. They are:
These three primary areas of your cortex are marked in dark green:
Non-primary areas of the cortex (areas that are not dumping grounds) are called association areas. Association cortex is associated with two distinct functions:
Evolutionary Considerations
How does the human brain compare to that of other species? Consider again the mouse brain. It is obviously smaller than the human variety, but is the ratio of primary-vs-association cortex preserved?
It turns out that the answer is no. Because primary sensory processing is more immediately useful to survival, the neocortex of the mouse is actually dominated (>50%) by primary areas.
This biological fact reminds me of the (amusing) Expanding Earth conspiracy theory. In contrast to plate tectonics, the Expanding Earth claim is that continent size has remained constant, whereas the oceans have been expanding. Despite being complete bunk, it does provide a colorful metaphor to our genetic distinctions from the mouse. Our species has invested more heavily in association cortex (oceans) than primary cortex (continents).
Have humans invested equally in every part of the association cortex? No: the hominid line shows pronounced (6x) gains in the prefrontal cortex, and comparatively restrained growth elsewhere. We thus have reason to believe that most uniquely human behaviors must be supported by the prefrontal cortex. To be continued.
Takeaways
Part Of: [Cognitive Architecture] sequence
Followup To: [Mental Architecture v1.4]
Let’s review our theoretical trajectory.
The Autonomic Mind: Belief, Motivation, and Decision Making
Our first architectures (1.0 → 1.2) explored the autonomic mind, which comprises our most fundamental mental capacities.
The Algorithmic Mind: Attention, Consciousness, and Intelligence
Subsequent architectures (1.3 → 1.4) complemented this understanding with the algorithmic mind, by weaving together three theories:
There is a close link between intelligence and consciousness, as evidenced by working memory’s strong correlation with both. Generally fluid intelligence (IQ) is essentially a measure of the precision of your attentional streams.
The Reflective Mind: Metacognition, Control, and Culture
The novel innovation of this architecture (1.5) could be entitled Prefrontal Cortex: The Final Frontier. It integrates two theories into the base corpus:
These five theories together constitute the foundation stones of my mental architecture. Let me call this synthesis the Attention-Modulated Tripartite Mind theory.
Putting Clothes On My Theory
A theory is a house, and the above merely represents its foundation. On this base, I will add details: the following posts are planned:
Part Of: Neuroanatomy sequence
Content Summary: 400 words, 4 min read
Today, we embark on a (very) brief tour inside your head!
How Neurons Work
What is a brain? A brain is a collection of 120 billion neurons.
What is a neuron? A neuron is simply a cell. Since you are a eukaryote, all cells in your body – neuron cells, skin cells, etc – have a lot in common, including:
In addition to these shared features, specialized tendrils (dendrites and axons) set neurons apart, giving them their web-like shape.
Neurons are useless individually. But when chained together, they may transmit pulses of electrochemical energy known as spike trains. input dendrites receive such signals and – sometimes! – fire, pushing the outgoing signal through the output axon. The output tendrils of a neuron connect with the input tendrils of other neurons; these connections are called synapses.
In this way, neurons mediate a relationship between input and output signals. What other kinds of things do this? Mathematical devices known as functions, and electrical devices known as transistors.
Gray Matter vs White Matter
Let’s zoom out to consider the entire brain. A brain has two symmetric halves (the hemispheres) connected by a bridge (the corpus callosum). Here’s what it looks like from the inside:
Certain parts of the brain are gray, and others are white. Why? What happens when you put these under a microscope?
If white matter doesn’t process information, why is there so much of it?
The Neural Plasma Globe
To answer this question, we turn to the gray matter. Where does it live? In only two places: the wrinkly surface of the brain (the cerebral cortex), and in an evolutionarily ancient part of the brain (the thalamus, and neighboring midbrain structures). Think: spheres within spheres.
We can see now why there is so much white matter. White matter comprises information highways which transport information to and from different areas of the cerebral cortex. Some of these highways directly connect cortical regions, but much travels through the thalamic “central hub”. We will call such highways thalamocortical radiations.
With tractography (a sister technology to the MRI scan), we can now directly visualize such radiations. There is something almost poetic about these highways… you could say that the brain is like a plasma globe.
Next time, we’ll explore some implications of this metaphor.
Part Of: [Mental Architecture] project
Followup To: [Mental Architecture v1.3]
Less of a jump than v1.3, but still a worthwhile place to “save my work”. Significant new features include:
Consolidation Of Working Memory
Adopts the Alternative View, contra the “Standard Model”, as put forward by Postle (2006) Working Memory As An Emergent Property of the Mind and Brain. Abolishes the Visuo-spatial Sketchpad and the Phonological Loop and replaces it w/ the contents of perceptual memory (and semantic memory, etc).
Improved Sensory Specificity
Reintroduces some of the sensory details already present in Mental Architecture v1.0, and adds a few organizing distinctions (e.g., exteroceptive vs interoceptive)
A Home For Emotional Processing
For months, my Emotion Generator module has lacked a comfortable home. Now, following Damasio’s gesture at homeostasis, I finally have a home for the “big six” emotions (Anger, Disgust, Fear, Joy, Sadness, Surprise). Specifically, I conjecture that these emotions are produced in topographic map located in the middle of a processing stream, whose data flows from interoceptive (e.g., hormonal) sense organs.
Explicit Brain Architecture
Administrivia
Module Changelog
Open Tasks
Open Questions
Frustratingly Unaddressed Human Explananda
