The brain as a connected map
A word about this chapter, before we start
I want to begin with an unusual confession: this is the hardest chapter in the book to organize, and I have come to think the difficulty is real rather than a failure of effort. It is worth a paragraph to say why, because understanding the problem is half of understanding the solution.
Every other chapter has a natural spine. The trouble with introducing the anatomy is that anatomy and function pull against each other. If I march you through the structures one by one — “here is the medulla, here is the pons, here is the thalamus” — without telling you what any of them is for, I have handed you a list of names, which is the least useful and least memorable thing a textbook can do. (You have suffered enough lists in your education; I will not add to the pile.) But if I leap straight to function — “the basal ganglia select actions” — I have made a claim about a place you cannot yet locate, and a claim about a place you cannot find is just noise. We have already been doing this to you, I confess: in the last two chapters I have said “hippocampus” and “amygdala” and “thalamus” and “nucleus of the solitary tract” without ever telling you where they are. I have been spending anatomical vocabulary on credit, and this chapter is where the bill comes due.
So neither pure order works. Some textbooks solve this by exiling the anatomy to an appendix you are meant to flip back to — an honest admission that it does not sequence. I am going to try something else, and I want you to know the strategy up front so you can hold me to it. I am going to give you a map first — just enough to locate things — and then I am going to spend the rest of the chapter on the thing that actually matters, which is not the places but the connections between them. The anatomy is the language of the brain; you genuinely do need it, and you will keep meeting it for the rest of the course. But you learn a language by speaking it, not by memorizing the dictionary first. I will teach you the minimum grammar and vocabulary to make true sentences, and then we will make sentences. The detailed anatomy of each system — the full roster of hypothalamic nuclei, exactly which receptors do what — will come in the units where each system is the main event. This chapter builds the framework; the later units fill it in. When you see me deliberately leave something unopened, that is not an omission. It is a placeholder, and I will tell you so.
One more promise, and then we begin. I am going to lean on a single metaphor — the brain as a map of places connected by routes — and like all metaphors it will eventually break. I will use it while it earns its keep and tell you plainly when it stops. With that, let us draw the map.
The map grows out of the five bumps
In Chapter 2 we watched the neural tube swell into a series of bumps: first three primary vesicles, then five. I promised then that this development mattered. Here is where I keep that promise, because the five bumps are not a developmental curiosity to admire and move past — they are the top level of the map we are about to use for the entire rest of this book. Every structure we discuss is a descendant of one of those five swellings. If you remember the five, you have the skeleton; everything else hangs off it.
Let me recall the derivation briefly, because a map whose shape is explained is a map you can reconstruct, whereas a map you simply memorize is a map you will forget. The neural tube’s front end (the prosencephalon, or forebrain) divides into the telencephalon and the diencephalon. The middle (the mesencephalon, or midbrain) stays undivided. The hindbrain (the rhombencephalon) divides into the metencephalon and the myelencephalon. Five vesicles: telencephalon, diencephalon, mesencephalon, metencephalon, myelencephalon, running front to back. And out of each, by the differential expression of regulatory genes we discussed in Chapter 2, grow the adult structures.
The map in this chapter is built as a set of nested, expandable lists — a “concept map.” Each box is a structure; the boxes nest, so that a box can be opened to reveal what it contains. A telencephalon box opens to reveal cortex, basal ganglia, hippocampus, and amygdala; the cortex box in turn opens to reveal the lobes; and so on.
Here is the important convention. When a box is opened, it means we are unpacking this structure now. When a box is left closed, it means this structure has contents, and we are deliberately leaving them for a later unit. A closed box is a promise, not a gap. So as you read, notice which boxes I open and which I leave shut — the shut ones are a map of what is still to come, and you can return to them as each later unit pries one open.
Caption / what this figure should show: A horizontally-oriented concept map (mind-map style), reading right-to-left or left-to-right, in which the five vesicles form one column of nodes — Telencephalon, Diencephalon, Mesencephalon, Metencephalon, Myelencephalon — color-coded, one color per vesicle. Each vesicle node connects to the adult structures that derive from it, in the vesicle’s color:
- Telencephalon → Cortex; Basal ganglia (Striatum); Hippocampus; Amygdala
- Diencephalon → Thalamus; Hypothalamus; Epithalamus
- Mesencephalon → Tectum (Colliculi); Tegmentum
- Metencephalon → Cerebellum; Pons
- Myelencephalon → Medulla
Each adult-structure node should carry a small “expand” affordance (e.g. a ⊕ icon) indicating it can be opened further — except where it is opened in a later figure. This is the master map of the chapter; later figures open individual nodes. (GMc: this is your existing CNS01-style mind map; please insert and relink.)
Notice the shape of the thing. The five bumps run rostral to caudal, and — this is the through-line of the whole unit, so I will keep returning to it — they run, very roughly, from the structures that handle the body’s most basic, ancient regulation at the back to the structures that perform the most elaborate, recent integration at the front. The medulla, at the caudal end, keeps you breathing. The telencephalon, at the rostral end, is where the vast expansions of human cognition live. The brain is, in a loose but useful sense, built in layers of ascending elaboration along this axis — an ancient vertebrate core at the back, with newer apparatus added in front. (I will immediately complicate this picture later, because the tidy “ancient back, modern front” story is exactly the kind of thing that gets oversold. But it is a fair first sketch.)
Two ways to build a brain: nuclei and layered cortex
Before we open any boxes, I need to give you two words, because they name the two fundamental ways the brain organizes its neurons — and almost every structure on our map is one or the other (or a bundle of cables connecting them, which is the next section). Get these two architectural motifs straight and the rest of the anatomy has somewhere to land.
First, some groundwork you already half-know from earlier chapters. The brain’s information-processing cells are neurons. A neuron receives signals (chiefly on its branching dendrites) and sends signals (down its single axon, which may itself branch to reach many targets). When neurons’ cell bodies are packed together, the tissue looks gray to the eye — “gray matter.” When we look instead at regions full of axons bundled and traveling, those look pale, because axons are often sheathed in fatty, light-colored myelin — “white matter.” Gray matter is where the computing happens; white matter is the wiring that connects one patch of computing to another. Hold that distinction; the rest of the chapter rides on it.
Now the two motifs.
A nucleus is a cluster of neuronal cell bodies, grouped together because they share connections and a job. (This is the neuroanatomical sense of “nucleus” — nothing to do with the nucleus of a cell.) A nucleus is a clump: a discrete blob of gray matter, with inputs arriving and outputs leaving, doing some identifiable piece of work. The nucleus of the solitary tract from Chapter 3 was one such clump — the place visceral afferents land. When I tell you, later, that the hypothalamus is “a collection of sixteen nuclei,” you now know exactly what that means: sixteen little functional clumps of cell bodies, packed into one region, each tuned to its own job. The thalamus is organized this way; the amygdala is; the basal ganglia are; the brainstem is studded with them. Whenever you meet a roster of nuclei in a later unit, you are meeting the contents of a clump-organized structure — and the reason I am defining the word now, rather than there, is so that the roster reads as a list of functional parts rather than a list of names to memorize.
Layered cortex is the other motif: not a clump but a sheet. In the telencephalon, the neurons are mostly not balled into nuclei. Instead they are spread into a continuous thin sheet — the cortex (Latin for “bark,” because it wraps the surface like bark on a tree). The human cortical sheet is only about 2 to 4 millimeters thick, but if you unfolded it, it would cover something like 1.3 square feet per hemisphere. And within that sheet, the neurons are stacked in layers — strata that differ in the kinds of neurons they contain and, crucially, in their connections: some layers are mostly input layers, receiving signals from elsewhere; others are mostly output layers, sending signals away. A sheet, layered by connectivity, is a fundamentally different architecture from a clump — and the difference will matter when we ask what cortex is for.
There is a wrinkle worth one paragraph, because it pays off later and ties back to our cross-species theme. Not all cortex has the same number of layers. The evolutionarily ancient cortex — allocortex, found in the rim of older structures like the hippocampus and the olfactory cortex — has only three or four layers. The newer neocortex, which makes up the great majority of the human cortical sheet, has six layers, and is found only in mammals. This is the same lesson as the rest of the book in miniature: the elaborate six-layered sheet is a mammalian addition built around an older three-layered core, not a thing that appeared from nowhere. (Recall the warning from Chapter 2 against assuming human brains have unique magic structures: the neocortex is genuinely a mammalian elaboration, but it is an elaboration of conserved cortical tissue, expanded and re-layered, not a novelty without ancestry.)
The cortical sheet is too large to fit smoothly inside the skull, so it is crumpled. The ridges that show on the surface are gyri (singular gyrus); the grooves folded down between them are sulci (singular sulcus). You will hear these words constantly, so: gyrus = bump, sulcus = groove.
Here is the part that is more than vocabulary. The folding is not random. The major sulci and gyri sit in recognizably the same places across individuals — the central sulcus, the fusiform gyrus, and the rest have consistent locations and names. Why would folds be reproducible? A leading idea is that the folding pattern is shaped by the connections the sheet has to make: regions that need to communicate are drawn together, and the sheet buckles to accommodate the wiring. If that is right — and it remains an active question — then even the shape of the brain’s surface is a story about connectivity. Which is the real subject of this chapter, and where we now turn.
The real subject: places are defined by their connections
Here is the move on which this whole chapter turns, and it is worth stating baldly. A place in the brain is not fully described by where it is or even by what cells it contains. It is described, more than anything, by what it is connected to. A structure’s function is largely a function of its inputs and its outputs — of the routes that run in and out of it.
Let me make this concrete with a metaphor I will ride for a while. Think of the brain as a map of places, the way a country is a map of regions. Iowa is, first of all, just a location — you can point to it before you know anything about it, which is exactly what our vesicle map lets us do for brain structures. But the moment you ask what Iowa is for, you find that its character is bound up with what flows in and out of it. Iowa grows corn; that corn ships out along routes to food producers in other states. So already Iowa is defined not just by its borders but by its trade routes — by where its output goes and where its inputs come from. To understand Iowa as an economic place is to understand its connections.
This metaphor is going to do a lot of work, so let me load it deliberately, because each feature of it corresponds to something real about brains.
Places have characteristic outputs, but rarely a single one. Iowa’s corn does not only become food. Some of it becomes ethanol, which becomes fuel. So you cannot say “Iowa = corn = food” and stop; the same output feeds into more than one downstream system. This is the single most important thing to internalize before we open any boxes, because the brain is full of structures that do more than one job, and students reliably trip over it. The basal ganglia help select movements and carry reward signals. The cerebellum coordinates movement and, it turns out, does more than that. The superior colliculus handles vision and the orienting of the eyes and head. You will be tempted, every time, to want each structure to have one tidy function. Resist it. A place can supply more than one system, and the brain’s places usually do. (You already accept this for Iowa without effort. Extend the courtesy to the basal ganglia.)
Places are connected in characteristic patterns. When many routes converge on one place, we call that convergence; when one place sends routes out to many, we call that divergence. These are not idle terms — they describe the computational shape of a connection, and you will meet them throughout the course. The nucleus of the solitary tract, from Chapter 3, is a convergence zone: visceral reports from all over the body funnel into it. The neuromodulatory systems we will meet (the ones that spray dopamine or noradrenaline widely across the brain) are divergence in pure form: a small source broadcasting to a vast territory. When you later study a sensory system and learn that many receptors feed into one neuron, that is convergence; when you learn that one alarm signal reaches the whole brain at once, that is divergence. The words let you describe what a wiring diagram is doing.
Routes run in both directions, and the return route carries regulation. Here is the feature of the metaphor I most want you to hold, because it is the one that makes this a chapter about a control system rather than a chapter about a pipeline. Iowa ships corn forward to the consumer — but the consumer signals back. If demand falls, word returns to Iowa to cut production. The route is not a one-way chute; it is a loop with a forward leg and a return leg. We have names for these. A connection that carries signals “forward” — from input toward output, from a lower station to a higher one — is feedforward. A connection that carries signals back the other way, or that loops within a system, is feedback or, when it forms loops, recurrent. And here is the empirical punchline, which genuinely surprises students: in the brain, the return routes are not an afterthought. They are everywhere, and they are often more numerous than the forward ones. The cortex and the thalamus are massively interconnected in both directions. Sensory pathways that carry information up to the cortex are shadowed by descending pathways carrying signals back down. Almost nothing in the brain is a one-way street.
This should sound familiar, because it is the same fact we met in Chapter 3 from the other side. There, the surprise was that the vagus is mostly afferent — the body reports up far more than the brain commands down. Here, the surprise is that brain structures are mostly talking back and forth, not just forward. These are two views of one deep principle: the brain is built out of loops, not chains. A control system must be — you cannot regulate anything without feedback, without the return signal that tells you whether your last action worked. The recurrence is not a complication on top of the architecture. The recurrence is the architecture, because regulation is what the architecture is for. When you find yourself imagining information flowing one way through the brain, from senses to thoughts to actions, stop and add the return arrows. They are always there, and they are the point.
I promised to tell you when the metaphor stops earning its keep. Here are its two main failures, and they are worth knowing precisely because each points at something true.
Map regions have crisp, fixed borders; brain territories often do not. Iowa has a hard line around it. Many functional territories in the brain shade into one another, overlap, and are defined by gradients rather than fences. When we “carve” the brain into regions, we are sometimes carving at real joints (as Chapter 2 discussed) and sometimes imposing tidier borders than nature drew. Hold the parcels loosely.
A map is static; the brain’s functional geography is not. States and their economies sit still on the map. The brain’s functional assignments are shaped by development and can be renegotiated — by experience, by learning, and dramatically by injury (recall from Chapter 3 how a stroke’s deficit reveals what a region was doing precisely by removing it). A map that can rewire itself is straining the metaphor. But the strain is informative: it tells you that “what this place does” is partly a fact about the present wiring, not an eternal property of the location.
So: a useful map, with soft borders and the capacity to change. Keep both caveats in your pocket as we walk it.
Walking the map, back to front
Now we open boxes. We will walk the five vesicles from caudal (the ancient regulatory core) to rostral (the elaborated forebrain), and for each structure I will do two things: locate it on the map, and sketch what it connects to — because the connections are what make the location mean something. Remember the convention: where I open a box, we unpack it now; where I leave one shut, its contents wait for a later unit. I will be explicit, once per major structure, about which is which, and then I will stop narrating the deferrals and trust you to read the shut boxes for yourself.
The myelencephalon: the medulla
At the caudal end, sitting directly atop the spinal cord as it enters the skull, is the medulla oblongata, the sole derivative of the myelencephalon. The medulla is small and it is essential in the most literal sense: it contains the nuclei that run breathing and heart rate, along with reflexes like vomiting. It also houses the nuclei of origin for four cranial nerves, and — this is the connectivity point — a great many white-matter tracts simply pass through it, because the medulla is the gateway between the spinal cord and everything above. Damage here is so often fatal precisely because the routes for life-support converge in this small place.
Notice that we have already met the medulla’s most important resident under a different heading. The nucleus of the solitary tract — the great convergence zone for visceral afferents from Chapter 3 — lives here, in the medulla. This is what the map buys us: a structure we discussed functionally, three chapters ago, now gets an address. The body’s status reports land in the medulla because the medulla is the brain’s ground floor, where the body’s wiring first arrives.
The metencephalon: pons and cerebellum
Moving rostral, the metencephalon gives us two structures, the pons and the cerebellum.
The pons (“bridge”) is, true to its name, a great crossing-point and relay. It contains the nuclei of four more cranial nerves and numerous nuclei involved in basic functions including aspects of sleep, and — the connectivity headline — it is a major waystation between the cerebellum and the rest of the brain. Vast numbers of fibers from the cortex synapse in the pons and are relayed onward to the cerebellum. So the pons is, in our terms, a convergence-and-relay station on the route to the cerebellum.
The cerebellum (“little brain”) deserves a careful word, because it is a showcase for the multi-function principle. The traditional and well-established story is that the cerebellum is a motor structure: it coordinates movement, refines its timing and accuracy, and maintains posture and balance. That story is true. Here is the Iowa lesson made flesh, though: the cerebellum has turned out to do more than motor coordination — there is now substantial evidence that it contributes to aspects of cognition and the timing of non-motor processes as well. I will not develop that here; it is a frontier, and it belongs to a later treatment. But I flag it now as a clean example: you will be tempted to file the cerebellum under “movement” and move on, and that file is not wrong, only incomplete. The corn also makes fuel. (The deeper story of cerebellar function is a shut box; we will open it later.)
One quantitative fact about the cerebellum that I want to plant now because it is genuinely startling, and we have carried it since Chapter 1: the cerebellum is only about 10% of the brain’s mass, but it contains roughly 80% of the brain’s neurons. The cortex, by contrast — that vast folded sheet — is about 82% of the mass but holds only around 19% of the neurons. Sit with that the next time you are tempted to equate “cortex” with “the brain.” Most of your neurons are in the little structure at the back. (Why the cerebellum is built from so many neurons, and what that architecture computes, is another shut box.)
The mesencephalon: the tectum
The midbrain (mesencephalon) is the one vesicle that did not subdivide. For our purposes its key feature is the tectum (“roof”), which consists of two pairs of bumps called the colliculi. The superior colliculus (which in non-mammals is called the optic tectum, and is a major visual structure) handles aspects of vision and — note the dual role again — the registration of space across the senses for the purpose of directing the eyes: it helps point your gaze at things. The inferior colliculus is a key station in the auditory pathway. So the tectum is, in connectivity terms, a sensory-integration roof on the midbrain, with the superior colliculus a convergence point where visual and other spatial signals meet to steer orienting movements.
The midbrain also contains the tegmentum beneath the tectum, home to nuclei we will meet later — including the red nucleus (a motor-related relay) and, importantly, several of the neuromodulatory sources whose divergent projections we flagged earlier. (The neuromodulatory nuclei of the midbrain are a shut box, and an important one; they open when we study arousal, reward, and motivation.)
The diencephalon: thalamus, hypothalamus, epithalamus
Now we reach the diencephalon, and the first of two structures here is so central to the whole connectivity story that it almost deserves to be the chapter’s mascot.
The thalamus is a collection of roughly thirteen pairs of nuclei sitting in the middle of the forebrain — and it is, more than any other structure, the brain’s great relay and convergence hub. Here is the pattern that will recur through every sensory unit you study: information from the body and the senses does not march straight into the cortex. With one telling exception (smell, which we will note when we get there), the senses reach the cortex by way of the thalamus. Vision routes through one thalamic nucleus, hearing through another, body sensation through another, and each is handed onward to its proper patch of cortex. To take the canonical example, which I will state here precisely so that it slots into place when you meet it again: the lateral geniculate nucleus of the thalamus receives input from the eyes and relays it to the primary visual cortex in the occipital lobe. You do not need to memorize that now. I am placing it here so that when the vision unit opens the box and shows you the lateral geniculate nucleus in detail, you will think: ah — this is the visual thalamus the framework promised. That is the whole method of this chapter in one example: build the slot now, fill it later.
And do not forget the return routes. The thalamus is not a one-way relay station passing sensation up to the cortex; the cortex projects massively back to the thalamus, in numbers that often exceed the forward projection. The thalamo-cortical loop is one of the most heavily recurrent circuits in the brain. Whatever the thalamus is doing, it is doing it in continuous two-way conversation with the cortex — feedforward up, feedback down, exactly the loop architecture this chapter keeps insisting on. (What that recurrent conversation accomplishes — for attention, for the regulation of cortical state, for consciousness itself — is a deep and partly unsettled set of shut boxes.)
The hypothalamus (“under the thalamus”) is small — about sixteen nuclei — and it is the master regulator of bodily homeostasis, which makes it the structure where the entire spine of this unit comes to rest. Thermoregulation, hunger and satiety, thirst, sexual arousal, sleep–wake cycles: the hypothalamus tunes them, and it does so through exactly the two output channels we built in Chapter 3 — the autonomic nervous system and the endocrine system. Some hypothalamic nuclei contain secretory cells that release hormones, making the hypothalamus the bridge between the nervous system and the bloodstream. This is the convergence point of everything we have built: the body’s afferent reports (via the medulla and thalamus) inform the hypothalamus, and the hypothalamus acts back on the body (via autonomic and hormonal efferents) to hold it in balance, or — in allostatic mode — to adjust the balance ahead of need. (The full roster of hypothalamic nuclei, and the detailed stress-axis machinery — the paraventricular nucleus, CRH, ACTH, cortisol — is a shut box. We previewed it in Chapter 3 and will open it fully when we study stress and motivated behavior. I am deliberately not dumping the nucleus table on you here; this is the framework, and that is the fill.)
The epithalamus is the smallest of the three and contains the pineal gland, which — unusually for a brain structure — is not a left-right pair but a single midline organ. The pineal secretes melatonin and participates in the regulation of circadian rhythms. (We met it in Chapter 3 as one of the secretory circumventricular organs — another structure now getting its map address.)
The telencephalon: cortex and the deep structures
Finally we reach the rostral end, the telencephalon, the most expanded part of the human brain and the one with the most to unpack. Its most visible feature is that it comes in two halves — two hemispheres, joined at the midline, looking nearly identical to the eye but, as we will see when we study split-brain patients, differing in important ways (the left hemisphere, for instance, is critical for language in most people). (Hemispheric specialization and the split-brain findings are a shut box flagged for later — and a fascinating one.)
The telencephalon contains, first, the great sheet of cortex we anatomized earlier — and the cortex itself opens into territories. It is customary to divide the cortical sheet into lobes: the frontal, temporal, parietal, and occipital lobes, with two further regions often added — the insula (a patch of cortex hidden in the fold between the frontal and temporal lobes) and the limbic lobe (the cingulate gyrus and adjacent cortex on the brain’s medial rim). I will locate them here and say a word about each lobe’s broad associations, but I want to be careful about how I say it, because this is the single easiest place in all of neuroanatomy to fall into the error this chapter has been warning against.
Before I tell you what the lobes “do,” I want to vaccinate you against a famous and seductive way of being wrong about brain organization, because the lobes invite exactly this mistake.
In the 1960s, Paul MacLean proposed the Triune Brain: the idea that the human brain is three brains stacked by evolution — a “reptilian” core (basal ganglia and brainstem) handling primal drives, a “paleomammalian” limbic system handling emotion, and a “neomammalian” neocortex handling reason — added in that order, like geological strata, with rational cortex sitting on top governing the beasts below. It is a wonderfully tidy story. It is also wrong, and it is instructive because it is wrong, which is why I keep it.
It fails on the evolution (the “reptilian brain” is not a leftover reptile brain; reptiles have their own elaborated forebrains, and the structures MacLean called primitive are present and developed across vertebrates — recall Chapter 2’s warning against recapitulation and against assuming a neat ladder of ascent). It fails on the anatomy (the “three brains” are not separable layers; they are densely interconnected and develop together). And it fails on the function (emotion is not confined to a “limbic” layer with reason quarantined above it — the two are thoroughly intertwined, as anyone who has made a decision can attest). The reason the Triune Brain has been so hard to kill is that it feels right and it is easy to draw. But “easy to draw and feels right” is not evidence, and a tidy diagram that misrepresents the biology is worse than no diagram. I show you this one as a cautionary example: when a scheme for brain organization seems suspiciously clean — three neat layers, one structure one function, reason atop emotion atop instinct — that very cleanness is a reason for suspicion. The brain did not read the diagram.
With that warning in hand: the lobes have broad associations, which are real as tendencies and false as strict assignments. The occipital lobe at the back is dominated by vision. The parietal lobe is heavily involved in body sensation and spatial processing. The temporal lobe handles much of hearing and, on its inner surface, is bound up with memory and the recognition of objects. The frontal lobe contains the motor machinery at its rear and, forward of that, the cortex most associated with planning, decision, and the flexible control of behavior. The insula is deeply involved in interoception — the sense of the body’s internal state we made so much of in Chapter 3 — and the limbic cortex with emotion and motivation. State these as centers of gravity, not as borders: vision is “mostly occipital” the way corn is “mostly Iowa,” with the real story spilling across lines and involving connections to everywhere else. (The detailed functional anatomy of each lobe is the substance of the units to come; here I am only locating the territories and warning you not to over-trust the parcellation.)
Beneath the cortical sheet, the telencephalon also contains three large deep structures — and these are clumps, nuclei-organized, not sheet-organized. I introduce them here by location and headline function, and each is a perfect instance of the multi-function rule:
- The amygdala (a cluster of about thirteen nuclei) is associated with emotion, fear, and the steering of motivated behavior. You will see it reduced, in popular accounts, to “the fear center.” It is more than that, and the reduction is the Iowa error again. (Shut box; opens with emotion and motivation.)
- The hippocampus is associated with memory and with spatial navigation — and, as we noted in Chapter 3, it is a target of stress. (Recall too that the hippocampus is built of the ancient three-layered allocortex, not six-layered neocortex — a structural clue to its antiquity.) (Shut box; opens with learning and memory.)
- The basal ganglia (also called the striatum, themselves several nuclei) are involved in the selection and inhibition of movements and behaviors — both the “go” and the “stop” — and are tied into the reward system. This is the textbook multi-function structure: action selection and reward, two systems served by one place. (Shut box; opens with movement and with reward.)
The routes themselves: white matter as the trade network
We have walked the places. Now we have to take seriously the thing the whole chapter has insisted matters most — the connections — and that means looking directly at the white matter, the brain’s physical wiring. If the gray-matter structures are the places on the map, the white-matter tracts are the trade routes, and a map without its routes is just a scatter of dots.
First, what a tract is, building on the neuron groundwork from earlier. Anatomists distinguish local-circuit neurons, whose axons stay nearby and do local processing (for instance, within the layers of a cortical patch), from projection neurons, whose axons travel long distances to reach a different region entirely. When the axons of many projection neurons run together from one region toward a common destination, they form a bundle — a tract (also called a fiber tract or pathway). These bundles are consistent enough across individuals, and even across species, that they have names, just as the structures do. A tract is, quite literally, a named route between two places, and like Iowa’s trade routes it has a direction (or two) and a shape (fan-in or fan-out). Let me organize the major routes the way their geography organizes them — and as we go, notice how the connectivity vocabulary from earlier (feedforward and recurrent, convergence and divergence) lets us describe each one’s job rather than merely its path.
Routes between the brain and the body: the spinal tracts
The spinal cord is the great trunk line between the brain and the body, and it carries traffic in both directions — which, by now, you will expect.
The descending tracts carry motor commands down from the brain to the spinal cord — feedforward routes from the controller toward the muscles. The principal one is the corticospinal tract, which runs from the motor and sensory cortex all the way to the motor neurons in the ventral horn of the spinal cord: the main highway for voluntary movement. (When the vision unit’s logic recurs in the motor unit, this is the tract that will carry “the cortex’s decision to move” to the body.) Alongside it run several brainstem-origin descending tracts that handle posture, tone, and balance: the rubrospinal tract from the red nucleus of the midbrain, the reticulospinal tract from the pons and medulla (which also participates in breathing and cardiac regulation), and the vestibulospinal tracts from the vestibular nuclei (controlling muscle tone, posture, and the stabilizing of head position). Notice that these descending routes originate at different levels of the map we just walked — cortex, midbrain, hindbrain — which is itself a lesson: motor control is not issued from one place but is a layered collaboration down the neuraxis. (Exactly how these layers divide the labor of movement is a shut box; it opens in the motor unit, where each of these tracts will be threaded into the functional story rather than listed.)
The ascending tracts carry sensory information up from the body to the brain — feedforward in the other direction, body toward cortex, and almost all of them routing (as promised) toward the thalamus. The dorsal column–medial lemniscus system carries fine touch and proprioception (the sense of where your body is) up to the thalamus. The anterolateral system (including the spinothalamic tracts) carries pain, temperature, and crude touch up to the thalamus — the lateral part for pain and temperature, the ventral part for pressure and touch. And the spinocerebellar tracts carry proprioceptive information from muscle and tendon up to the cerebellum, feeding the motor-coordination machinery the body-position data it needs. (Which receptors feed these tracts, and how the information is transformed along the way, is shut; it opens in the somatosensory and pain units. Here, the point is only the architecture: body sensation ascends, mostly via the thalamus, to reach the cortex.)
Routes within a hemisphere: association and projection fibers
Within a single hemisphere, tracts connect the thalamus and the cortical regions to one another. The corona radiata is a spectacular example of divergence made visible: fibers fan outward from the thalamus to the entire cortex like rays from the sun — early anatomists named it for exactly that appearance. (Your instructor has shown, in class, a preparation in which the cortex has been digested away to leave these white-matter “ropes” exposed; it is a striking thing to see.)
Other intra-hemispheric tracts — the association fibers — connect cortical regions to each other, and they are where the connectivity vocabulary earns its keep most clearly. The superior longitudinal fasciculus connects the parietal, occipital, and temporal lobes with the frontal cortex; it is explicitly bidirectional, made of sub-pathways running both ways, and it is heavily involved in language and attention — a recurrent route knitting the back of the brain to the front. The inferior longitudinal fasciculus links the occipital lobe to the front of the temporal lobe (a route in the service of recognizing what we see). The uncinate fasciculus is a bidirectional route connecting the front of the temporal lobe to the orbitofrontal cortex, usually counted as part of the limbic system. And the cingulum runs within the limbic rim, connecting the cingulate cortex to the entorhinal cortex — which is the major gateway of input into the hippocampus.
A subset of these limbic routes are, specifically, the output cables of the deep structures, and they all converge on a familiar destination — the hypothalamus, the regulator. The fornix carries output from the hippocampus to the hypothalamus. The stria terminalis carries output from the amygdala to the hypothalamus (and to the bed nucleus of the stria terminalis). The ventral amygdalofugal pathway carries another stream of amygdala output to a spread of targets including the hypothalamus, the thalamus, the nucleus accumbens, and frontal cortex. I point these out not to add three names to your list, but because they show the loop closing yet again: the emotion-and-memory structures send their outputs down to the same homeostatic regulator that Chapter 3 built, which is precisely why your emotional and remembered states can reach out and change your heart rate, your hormones, your gut. The wiring is the explanation. (How these limbic outputs shape motivated behavior is shut; it opens with emotion, memory, and stress.)
Routes between the hemispheres: the commissures
The two hemispheres are connected by commissures — tracts that cross the midline. By far the largest is the corpus callosum, a massive band connecting corresponding (and some non-corresponding) regions of the two hemispheres: the main channel by which the two halves of the cortex stay coordinated. It carries a famous clinical and scientific significance — it is sometimes surgically cut to stop severe epileptic seizures from spreading between hemispheres, and those “split-brain” patients have taught us a great deal about hemispheric specialization. (The split-brain story is one of the most illuminating shut boxes in the book; we open it later.) The smaller anterior commissure connects the temporal lobes and the amygdalae across the midline; the tiny posterior commissure links pretectal nuclei and serves the pupillary light reflex; and the hippocampal commissure (part of the fornix) joins the two hippocampi.
Routes to the cerebellum: the peduncles
Finally, the cerebellum hangs off the brainstem by three thick fiber bundles per side called peduncles (“little feet”), and they are a tidy illustration of how naming a route’s direction tells you its job. The middle cerebellar peduncle is the great afferent route into the cerebellum — carrying the pontine relay of cortical information we mentioned earlier. The inferior cerebellar peduncle is also largely afferent, gathering the spinocerebellar tracts (the body-position information) and delivering them to the cerebellum. And the superior cerebellar peduncle is the major efferent route out of the cerebellum, carrying its computed output onward to the thalamus and the red nucleus. Three cables: two bringing data in, one sending results out. The cerebellum’s whole conversation with the rest of the brain runs through these three routes, and simply knowing which way each one points tells you the shape of that conversation.
For most of the history of neuroanatomy, the tracts could be studied only in dissected, preserved tissue — which is why the classic preparations involve digesting the cortex away to expose the white-matter “ropes.” Something has changed in the last two decades, and it is worth knowing because it is why white-matter connectivity has become such an active field. A form of MRI called diffusion-weighted imaging detects the direction in which water diffuses along axons, and from that it can reconstruct the major tracts — non-invasively, in living people, in large numbers. The resulting images, with tracts color-coded by their direction of travel, let us study the brain’s wiring quantitatively in health and disease. The trade network, once visible only in the dissecting room, can now be photographed in you. (How these methods work, and what they have and have not established, is a shut box for the imaging unit.)
Caption / what this figure should show: A schematic “trade-route map” overlaid on a simplified mid-sagittal or lateral brain outline, showing the major tract categories in distinct colors with arrowheads indicating direction (and double arrowheads where bidirectional/recurrent):
- Brain–body (spinal): descending corticospinal (cortex → ventral horn) in one color; ascending dorsal-column and spinothalamic (body → thalamus) in another; spinocerebellar (body → cerebellum) in a third.
- Within-hemisphere (association): superior longitudinal fasciculus (double-headed, parieto-occipito-temporal ↔︎ frontal); the corona radiata fanning from thalamus to cortex (to depict divergence).
- Between-hemisphere (commissural): corpus callosum crossing the midline.
- Cerebellar peduncles: middle (afferent, in), inferior (afferent, in), superior (efferent, out), with arrowheads making the in/in/out pattern explicit.
The intent is pedagogical, not exhaustive: the figure should show feedforward vs. recurrent (arrowheads), convergence (many→one, e.g. spinocerebellar into the inferior peduncle), and divergence (one→many, the corona radiata). A clean schematic is far better here than an anatomically dense tractography image. (GMc: candidate for a redrawn schematic; the existing intracortical-tract and peduncle figures could be adapted, but a single unified route-map would serve the chapter’s thesis best.)
The other network: blood supply over the same map
There is a second network laid over the brain’s map, and although we treated it at length in Chapter 3, it earns a brief reprise here because it completes the connectivity picture — and because it fails the trade-route metaphor in an instructive way.
Recall the essentials: the brain is fed by two arterial systems — the carotid (anterior) circulation and the vertebral–basilar (posterior) circulation — which meet in the ring of the Circle of Willis, from which the major cerebral arteries branch to their territories. The brain stores almost no energy and so depends on this delivery continuously; cut a vessel and the territory it feeds is starved within seconds, which is a stroke.
Here is the instructive part, the place where blood supply and the trade-route metaphor come apart — and the coming-apart is exactly what makes strokes so revealing. Vascular territories do not respect functional borders. A single artery feeds whatever tissue happens to lie downstream of it, regardless of which functional systems that tissue belongs to. So when an artery is blocked, the resulting deficit is carved along vascular lines, not functional ones — which is why a stroke can knock out a peculiar-looking combination of abilities that share no functional logic, only a shared blood supply. The road network and the trade network are different maps laid over the same country, and the brain has both: the white-matter tracts wiring places by function, and the blood vessels supplying places by geography. A complete map of the brain needs both networks drawn on it. And the mismatch between them is not a nuisance — it is a gift to neuroscience, because the vascular accidents that cut across functional systems have, for over a century, been one of our primary windows into what each piece of the functional map actually does.
Coda: from a map of places to a system that regulates
Let me close by collecting what we have built and pointing it forward, because this chapter was always preparation for what comes next.
We started with a problem I admitted was genuinely hard: you cannot use the brain’s anatomical language until you have a map, but a map of names without functions is inert. The solution was to let the map grow out of the five developmental vesicles of Chapter 2 — so that its shape is explained rather than memorized — and then to insist that the real content is not the places but the connections between them. We gave ourselves the two architectural motifs every structure is built from (the nucleus, a clump; and layered cortex, a sheet), and the vocabulary of connection (feedforward and recurrent; convergence and divergence). We walked the map from the ancient regulatory medulla at the back to the elaborated forebrain at the front, locating every structure we had previously used on credit — the nucleus of the solitary tract, the hypothalamus, the thalamus, the hippocampus, the amygdala — and sketching what each connects to. And we laid down the trade network of white-matter tracts and the second network of blood supply over the same terrain.
Now notice what the map is, when you step back and look at the connections rather than the boxes. The senses and the body feed information inward and upward — through the spinal tracts, through the thalamic relays, converging on the integrating structures. The integrating structures — the cortex, the deep telencephalic clumps, and above all the hypothalamus — process and combine these signals. And commands flow outward and downward — through the descending tracts to the muscles, through the autonomic and endocrine outputs to the viscera. Sensing, integrating, acting. But — and this is the whole point, the thing the recurrence kept insisting on — it is not a one-way assembly line from sensation to action. Every forward route is shadowed by a return route. The body reports up far more than the brain commands down (Chapter 3); the cortex talks back to the thalamus as much as the thalamus talks to the cortex (this chapter); the emotional and mnemonic structures send their outputs back down to the homeostatic regulator. The brain is built of loops, and it is built of loops because it is a control system, and a control system is nothing without the feedback that tells it whether its regulation is working.
That is the bridge out of Unit I. We began the unit with a claim that might have sounded like a slogan: the expensive brain earns its keep by buying prediction — by regulating the body allostatically, forecasting needs before the errors arrive, rather than merely reacting. Across four chapters we have assembled what that claim requires: a reason for a brain at all (Chapter 1), the developmental plan that builds one (Chapter 2), the channels that wire it into the body it regulates (Chapter 3), and now the map of the regulating machinery itself, drawn as a network of connected places (Chapter 4). The control system is now before us, in outline, with all its major parts located and its major routes traced.
What remains is to open the boxes we have left shut — to take each functional system in turn and study it in the depth it deserves. We have built the framework deliberately, and deliberately left the details for where they belong. The next unit begins that work, starting with the cells that all of this is built from — the neurons and glia whose properties we have so far simply assumed. We have been describing the wiring diagram of the brain. It is time to meet the wires.
In the spirit of the unit, an honest ledger for a chapter that is more framework than findings — which shapes what “sure” even means here.
What we’re sure of.
- The gross developmental map is solid: the five vesicles and the adult structures deriving from each are textbook anatomy, not in dispute.
- The two architectural motifs — nucleus (clump) and layered cortex (sheet), with the allocortex/neocortex layer distinction — are well established.
- The major white-matter tracts and their basic connectivity (what connects to what, and in which direction) are well established, and now directly visualizable by diffusion imaging.
- The thalamus as the obligatory relay for the senses (smell excepted), and the massive recurrence of thalamo-cortical and other connections, are solid and central.
- The dual arterial supply, the Circle of Willis, and the mismatch between vascular and functional territories (hence the diagnostic value of strokes) are firmly established.
What we’re not sure of, or oversimplified on purpose.
- The “ancient back, modern front” gradient and the lobe-by-function assignments are centers of gravity, not borders. I stated them simply here and flagged the simplification; the real functional anatomy is distributed and overlapping, and the units to come will complicate every one of these tidy assignments.
- The multi-function structures (basal ganglia, cerebellum, amygdala, superior colliculus) genuinely do more than their headline job, and in several cases (cerebellar contributions to cognition, especially) how much more is an active frontier.
- The functional claims attached to each structure here are deliberately shallow — slots, not fillings. Where this chapter says “the amygdala is involved in emotion and fear,” that is a placeholder for a much more contested and detailed story the later unit will tell. Do not mistake the headline for the science; the science is in the units that open these boxes.
- The Triune Brain is in here as a kept mistake — a model that is wrong but instructive. If you remember one cautionary lesson from this chapter, let it be the one it teaches: a scheme for brain organization that is suspiciously tidy — neat layers, one structure one function — is suspect precisely because of its tidiness. The brain is built of overlapping, multi-functional, densely interconnected loops, and any map that makes it look simpler than that is lying to you a little. Including, where it must simplify to be teachable, this one — which is why the boxes are shut, not sealed.