The Organ of Tomorrow

An Overview of the Frontal Lobes, and the Problems This Unit Must Solve

A controller that has run out of present

We closed the last unit with a machine that could choose, and then learn which choices were worth making. The basal ganglia hold the motor programs switched off and release one by lifting its inhibition; the ventral striatum, supplied with a dopamine teaching signal, reshapes which release the contest will favor and energizes the pursuit once it is underway. By the end we could trace a single contest — eating against throwing — all the way through, and watch a shift in drive flip it. The architecture chose; no inner agent was needed.

But we ended on a confession. That machine reacts to value; it does not represent it. Our worked example flipped a contest by changing a drive — it did not weigh two genuine alternatives and judge between them. The striatum can tell you that a reward is coming and can teach you how to get it, but it cannot hold up two unlike options that are not currently present and ask which is worth more. It cannot compare an apple you might walk to against a mate you might court. It cannot represent a future that does not yet exist, place you inside it, and let that imagined future govern what you do now. Everything the last unit built was, in a sense, present-tense: a response to the affordances on offer, the drives now active, the reward now arriving or now withheld.

This is the gap the frontal lobes fill, and it is the largest gap in the book. To see its shape, return for a moment to the distinction that opened the very first unit. We said the brain is an embodied controller in the service of two things: homeostasis, the correction of error around a setpoint, and allostasis, the regulation of the body by anticipation — getting ahead of need rather than waiting to correct a deficit after it bites. We then spent five chapters building the homeostatic, reactive, present-tense machinery in detail: the hypothalamus correcting departures from setpoint, the sensory cortices handling the seen world, the cerebellum smoothing movement, the basal ganglia selecting among present candidates, the ventral striatum learning the value of what just happened. We built, thoroughly, the part of the controller that answers to now.

What we have not built is the part that answers to later. An animal that lives only in the present is at the mercy of it — captured by whatever affordance is brightest, whatever drive is loudest, whatever reward is nearest. To do better, a controller must be able to act on what is not in front of it: a goal it has chosen but not yet reached, a rule that holds even though the last stimulus argued against it, a consequence that has not happened, a version of the world that is merely possible. It must, in short, model the not-yet and let the model take charge. That capacity is what this unit is about, and the frontal lobes are where the vertebrate brain locates most of it.

Three things people say about the frontal lobes

Before we go further it is worth naming, and partly disarming, three claims that hang over this part of the brain, because they pull in different directions and the reader will meet all three elsewhere.

The first is that the frontal lobes are what make us human. There is something to this — the apparatus we are about to describe is unusually elaborated in our species, and the behaviors it supports (long-range planning, moral reasoning, the projection of a self across years) are among the ones we most associate with being human. But it is a dangerous half-truth. The frontal lobes are not a human invention; every vertebrate has frontal cortex, and the functions we will study run continuously down through the mammals and beyond. What is special about the human case is a matter of degree and elaboration, not of kind — and we will see that the “expansion” itself is less settled than the slogan suggests.

The second is that the frontal lobes are the brain’s executive — the boss, the chief, the place where decisions are made and behavior is commanded. This is the framing built into the very term “executive function,” and it is genuinely useful as a description of what is lost when this cortex is damaged: planning, control, the governance of behavior by goals. But taken literally it is exactly the idea this book has been renouncing since its first pages. An executive is a homunculus — a little agent inside the head who watches, decides, and issues orders, and whose own watching and deciding then go unexplained. We have refused that move at every level: there is no commander in the basal ganglia who chooses the movement, no connoisseur in the ventral striatum who feels the pleasure and issues the desire. We will refuse it here too. And we should be honest about why the temptation is sharpest in this unit: if there were a homunculus anywhere in the brain, this is where it would live. The frontal lobes sit atop the hierarchy; their damage looks, more than any other, like the loss of a person’s will and judgment rather than the loss of a specific competence. The whole burden of this unit is to show that what looks like a chief is in fact a set of mechanisms — and that “executive control” names what those mechanisms do, not an agent who does it.

The third claim is the one this book holds: the frontal lobes are part of the control system — the anticipatory part. They are not a boss riding above the homeostatic machinery of the earlier units; they are continuous with it, the same kind of control raised to a longer horizon. Where the hypothalamus regulates the body around its current setpoints, the frontal lobes regulate behavior around represented setpoints — goals, values, and possible futures that the present does not contain. On this view there is no seam between the “lower” reactive brain and a “higher” deciding one. There is one controller, working at the front of the brain on a problem the back of the brain cannot reach: the future.

These three framings are not equally wrong. “Make us human” is a true observation overstated into a slogan. “Executive” is a useful clinical shorthand that becomes a homunculus if you believe it. “Part of the control system” is the frame we will build on. The point of naming all three now is so that, as the syndromes accumulate, you can feel the pull of the first two and notice yourself resisting it.

What lesions can, and cannot, tell us

Much of what we know about the frontal lobes — and much of this unit — comes from people who have lost part of them, to stroke, tumor, injury, or surgery. It is worth pausing on what kind of knowledge that is, because the method has a logic, and the logic matters for the argument we are about to make.

When we study a patient with a focal lesion, we are running a subtraction. We cannot open the intact brain and read off what each region does; what we can do is remove a region and ask what the person can no longer do. The competence that disappears is our evidence for what the missing tissue contributed. This is an old and powerful idea — it is how the nineteenth century localized language — and it comes with old and well-known hazards. Lesions are rarely tidy: a stroke or a tumor respects blood supply and cell vulnerability, not the boundaries a neuroscientist would draw, so a “frontal” lesion routinely takes neighboring territory and severs passing fibers along with it. A lost function may reflect not the damaged region itself but a system the damage disconnected. And subtraction reveals what a region is necessary for, not the whole of what it normally does, nor how it does it. We will lean on lesions throughout this unit, and we will try to keep these cautions in view — pairing them, where we can, with recordings from intact brains and with imaging that shows the healthy system at work.

But there is a second reason the lesion method belongs at the front of this unit in particular, and it bears directly on the homunculus. Ask the diagnostic question the method forces: when we damage the frontal lobes, what can the person no longer do? Notice what the answer is never. It is never “the chief is gone, and the rest of the brain now sits leaderless, awaiting orders that never come.” It is always something specific. One patient can no longer hold a goal in mind against distraction. Another can no longer generate an alternative to the situation in front of her — cannot imagine the car crash turning out all right. Another cannot keep his hands off the objects in the room, reaching out to use whatever he sees. Another makes plans for tomorrow that somehow never arrive. Each is a particular competence subtracted, leaving the rest of the person — often including a normal IQ — intact around the hole. We never find the boss missing. We find a specific tool gone from the kit, and a person improvising, often badly, without it. That pattern is itself an argument, and the unit will let it accumulate.

One lobe, several organs: differentiating the frontal cortex

“The frontal lobe” is a single name for a large and heterogeneous stretch of cortex — by some estimates more than a third of the human cerebrum. Much of it, toward the back, we have already met: the precentral gyrus is primary motor cortex, and just in front of it lie the premotor and supplementary motor areas we discussed in the unit on movement. Our concern in this unit is everything anterior to that motor strip: the prefrontal cortex, the part of the frontal lobe defined less by where it sits than by its connections — classically, by its dense two-way traffic with the mediodorsal thalamus. It is here that the anticipatory work is done, and it is not done by one undifferentiated mass. It is worth drawing the major territories now, because the syndromes ahead respect them, and because their differences turn out to have deep roots.

A tale of two origins

The most useful organizing fact about prefrontal cortex is that it is not one thing with one history. Developmentally and architecturally it carries two trends, with two different ancestries, and almost everything else about the region — its connections, its chemistry, what its damage costs — follows from which trend a given patch belongs to.

One trend is ventral and medial: the orbital cortex on the underside of the frontal lobe, above the eye sockets, and the medial wall that faces its twin across the midline. Architecturally this cortex is agranular or dysgranular — it lacks, or barely develops, the granular fourth layer that elsewhere receives sensory input. It is the phylogenetically older trend, and it grew up in intimate company with the brain’s visceral and emotional core: the amygdala, the hypothalamus, the autonomic machinery of the brainstem, and — a telling detail — the olfactory system. Smell, the oldest of the distance senses, is in essence a value sense: is this good or bad, approach or avoid, eat or flee. That the orbital cortex sits where the olfactory inputs once dominated is a clue to what it became. This ventral-medial trend is the brain’s apparatus for value and the body — for registering what things are worth, and for doing so in the currency the earlier units established, the currency of bodily and emotional state.

The other trend is dorsal and lateral: the granular prefrontal cortex of the upper and outer surface, with a well-developed layer IV and a very different set of friends. It grew up with the sensory-association and parietal world — the spatial, visual, and motor-planning cortices — rather than with the visceral core. This is the most recently elaborated trend, the one most expanded in primates and the focus of the (contested) claims about human enlargement we turn to below. Where the ventral-medial cortex evaluates, the dorsolateral cortex holds and manipulates — keeping information online, sustaining a plan, maintaining a goal across the gap between intention and action.

This dual origin is not a piece of trivia; it is the backbone of the unit. The reason the ventral-medial system evaluates futures by running the body’s own signals forward — the reason an imagined consequence can be felt as a pang or a pull — is that this cortex was built out of the brain’s visceral and emotional machinery in the first place. The “feeling” attached to a possible outcome is not a metaphor laid over a cold computation; it is an evolutionary inheritance from the cortex’s limbic origins. And the reason the dorsolateral system holds locations and rules and action-plans online is that it was built out of the brain’s sensory and spatial association cortex. The two halves of the frontal lobe do different jobs because they came from different places.

A word of caution, in the spirit this book has tried to keep. The clean version of the dual-origin story is cleaner than the tissue. The borders between these territories are drawn in pencil, the lesions that teach us about them are large and rarely confined to one trend, and the homologies between rodent and primate prefrontal cortex — especially for the granular dorsolateral type — are genuinely disputed, with serious researchers (Wise among them) cautioning against easy cross-species mappings. Treat the two trends as a real and useful organization, not as two boxes with a firm wall between them.

The working territories

Within those two trends, four regional labels will recur, and it is worth fixing them now, along with an honest admission about how firm they are.

On the ventral and medial side:

The orbitofrontal cortex (OFC) — the underside of the frontal lobe — leans toward the value of objects and outcomes: what a particular thing is worth right now, and the rapid updating of that worth when circumstances change. It is closely tied to the amygdala and to the reward machinery of the last unit.

The ventromedial prefrontal cortex (vmPFC) — the lower medial wall — leans toward integrating value with the self and the body: binding an option to the emotional and somatic signals that mark it as good or bad for this animal, and projecting the self into possible situations so that they can be evaluated as if they were real. This is the territory of the “somatic marker” and of self-projection, and it connects to the amygdala through a fiber pathway, the uncinate fasciculus, that we will meet repeatedly.

Be warned that the literature does not use these two terms consistently. “vmPFC,” “OFC,” and “medial PFC” are drawn differently by different authors, and because lesions here are typically large, the distinctions blur in practice. We will use the working split above — OFC for object-and-outcome value, vmPFC for value bound to the self and body — while remembering that the boundary is a convenience, not a fact of nature.

On the dorsal and lateral side:

The dorsolateral prefrontal cortex (dlPFC) is the apparatus of holding and control: keeping a goal or a piece of information online when the eliciting stimulus is gone, manipulating that information, sustaining a plan, and protecting it against the pull of distraction and habit. This is the home of working memory and of much of what is meant by “executive function.”

The ventrolateral prefrontal cortex (vlPFC), just below it, is involved in selection and the interruption of ongoing behavior — in retrieving the right item, in stopping a prepotent response, and (in some accounts) in the “circuit-breaking” that lets a salient event commandeer attention. It sits at the boundary between the holding functions of the dorsolateral cortex and the value functions of the ventral cortex, and the two interact: a vivid emotional distractor that pulls the ventral system can knock the dorsolateral system off its goal.

These are not five independent organs with five clean jobs. They are regions of a continuous sheet, heavily interconnected, and most real behavior recruits several at once. The labels are a way of carving a large problem into approachable pieces — and, as with the direct and indirect pathways of the striatum, the carving is a teaching device that the tissue only partly respects.

On the human “expansion”

It is often said that the most anterior prefrontal cortex — the frontal pole, the cortex furthest from the motor strip — is disproportionately enlarged in humans, and that this expansion is the anatomical signature of our distinctive cognition. The claim is attractive and may well be partly true, but it is contested, and the honest version is more interesting than the slogan. Some analyses (Semendeferi’s among them) report that frontal cortex, and the frontal pole in particular, is larger in humans than a simple scaling from other primates would predict; others find that the frontal lobe as a whole is about the size expected for a primate brain of our overall dimensions, and that what is genuinely different is concentrated not in gross volume but in organization — more white matter connecting frontal cortex to the rest of the brain, larger and more richly branched pyramidal neurons, finer microstructure. The safe summary is that the anterior-frontal contribution is elaborated in humans, by some mix of volume and connectivity and cellular complexity, even if the bare claim “humans have a bigger frontal lobe” is shakier than it is usually stated.

Why does it earn a place here rather than in a footnote? Because there is an axis already visible in the frontal lobe that this expansion, if real, would illuminate. Tasks recruit progressively more anterior frontal cortex as they grow more abstract and more deeply nested — as they require holding a sub-goal while evaluating a branch, or reasoning about whether one should be doing this task at all. That posterior-to-anterior gradient of abstraction is, plausibly, the same axis as depth of self-projection: the further an animal can project itself into futures that do not yet exist, and the more deeply those projections can be nested inside one another, the more anterior-frontal apparatus the projection seems to demand. On this reading the contested expansion is not decoration but the phylogenetic face of the unit’s whole subject. An animal that lives further out along the axis of acting on the not-yet needs more of the cortex that does it — and humans live further out than anything else we know of. We will hold this as an interpretation worth entertaining, framed as such, not as a settled magnitude on which the chapter rests.

The shape of the answer, and the problems it leaves us

We can now state, in advance, the spine of this unit — not to settle it, but so that the syndromes ahead have something to hang on.

The frontal contribution to anticipatory control divides, roughly, along the two-origin line we have just drawn, into two complementary jobs. The ventral and medial system generates, represents, and values possible futures, and projects the self into them: it holds up alternatives that are not present, attaches worth to each on a common scale, and — drawing on its limbic inheritance — lets the body’s own signals stand in for the future so that a merely possible outcome can be felt and weighed. This is the system that answers which future is worth pursuing. The dorsolateral system holds a chosen future in command of action and protects its pursuit: once a goal is selected, it keeps that goal online against the steady pull of the brightest stimulus and the strongest habit, sustaining the plan that serves it. This is the system that answers how to get there, and how to keep going when the world offers a hundred reasons to do something else.

The two fail in different, almost opposite-feeling ways, and the difference is one of the clearest pieces of evidence that they are doing different jobs. Damage the ventral-medial system and choice itself goes wrong: decisions run against the person’s own interest, the imagined sting of a bad outcome never arrives to deter it, and — at the far end — the evaluation of others’ futures flattens into the cold arithmetic we will meet under the headings of psychopathy and moral judgment. Damage the dorsolateral system and choice is often fine but capture sets in: the goal is sound, the values are intact, yet behavior is pulled off course by whatever is most salient or most habitual — the distractor, the prepotent response, the object within reach. One lesion makes you choose badly; the other lets you choose well and never arrive.

Two further threads run alongside this spine. The first is a drive: exploration. A controller that values possible futures can only choose among the futures it has managed to imagine, and it can only imagine the ones it has the materials to build. Those materials have to be gathered — and they are gathered, in part, by a drive to wander and sample that surfaces precisely when the homeostatic drives fall quiet, when nothing is urgently needed. We wander, and our brains note where the food and the shelter and the mates are, against a future need we are not feeling now. We should be careful here exactly as we were careful with wanting and liking: the animal is not consciously gathering intelligence for later. It is following a curiosity that, over evolutionary time, was kept because the brains that wandered when sated arrived at their next hunger already knowing the territory. The purpose lives in the selection history; the mechanism is just a drive to sample the novel. This exploratory drive stands, interestingly, in tension with the dorsolateral system’s job of protecting an ongoing goal from capture — for exploration is, in a sense, productive capture, the willingness to be pulled off the current track by something new. The negotiation between the two — stay the course, or sample the unknown — is a genuine dilemma the frontal lobe must continually resolve, and we will return to it.

The second thread is the floor beneath the whole edifice, and it is what the most severe frontal damage exposes. Subtract enough of this apparatus and you do not get an animal that chooses badly, nor one that is merely distractible. You get an animal that no longer sets out — that retains its intelligence and its perceptions but has lost the drive to lean into any future at all. The patients with this profile make plans for tomorrow, the old clinical descriptions tell us, but somehow tomorrow never comes; they do enough for today, and then stop. That endpoint — the loss not of a particular future but of the impulse toward the future as such — is where this unit will close, and it is the deepest evidence that what the frontal lobes contribute is not a chief who commands but a capacity to be governed by what is not yet here.

So the unit ahead is organized not by anatomy marched through region by region, but by the structure of an anticipatory choice. We will ask, first, how the brain represents what an option is worth — the valuation work of the orbital and ventromedial cortex, and the common currency that lets unlike things be compared. We will ask how the brain holds a chosen goal online and defends it — the working memory and control work of the dorsolateral cortex, and the syndromes of distraction, perseveration, and capture that reveal it. We will ask how possible futures come to be felt — the somatic-marker machinery that re-uses the body to evaluate the imagined, and the bridge it builds back to the emotional and interoceptive material of the first unit. We will follow that capacity outward into its social and moral reach, where the future to be valued belongs to someone else. And we will end at the floor, with the patients in whom the drive toward tomorrow is gone, and let them hand the next unit its opening question — for you cannot simulate a future you cannot constructively remember, and so the road out of the frontal lobes runs into the machinery of memory.

If there were ever a place in the brain where the homunculus lived, it would be here. The work of this unit is to walk through its house, room by room, and find in each room not a chief but a mechanism — and to show that the sum of those mechanisms is exactly what we mean, and all we need to mean, by a controller that can act on tomorrow.