45Descartes' Error

Antonio Damasio, Descartes' Error: Emotion, Reason, and the Human Brain, Putnam, 1994; Penguin edition with a new preface, 2005

Linked to 24 chapters: supports 14 · contradicts 6 · mechanism 10 · pairing 6

Damasio was brought up on the advice that sound decisions come from a cool head, and he had in front of him the coolest and least emotional intelligent man he could imagine, whose practical reason had collapsed. The patient had lost his savings to a partner everyone had warned him against, had been through two divorces and could not hold a job, and he scored in the superior range on every instrument the laboratory could give him, from intelligence and memory to estimation under uncertainty and moral reasoning. One lesion had produced both the ruined deciding and a flattening of feeling so complete that he recounted his own downfall without a sign of distress. The book argues that the pairing is not a coincidence. It moves from a foreman's skull in 1848 to a modern patient with the same lesion, to lesions elsewhere that produce the same two losses together, and from there to a mechanism. An emotion, in Damasio's terms, is a set of changes in the body; a feeling is the perception of those changes beside the thought that caused them; and the feelings experience has attached to the imagined outcomes of our options act as markers, biasing the field before any weighing of costs against benefits begins. That is the somatic-marker hypothesis, and the title is its consequence. Reason was built out of the machinery that keeps a body alive rather than walled off from it, and the separation Descartes drew between thinking thing and extended thing is the error the book is named after.

Unpleasantness in Vermont

In the summer of 1848 Phineas Gage, twenty-five, was foreman of a gang blasting rock for the Rutland and Burlington Railroad near Cavendish, Vermont, and was called by his employers the most efficient and capable man in their employ. Blasting meant tamping sand down over a charge of powder with an iron bar, and Gage's bar had been made to his own specification: three feet seven inches long, thirteen and a quarter pounds, tapered to a quarter-inch point. At half past four on a hot afternoon someone called out, he looked away, and he began tamping before the sand had been poured. The charge blew upward into his face. The iron entered his left cheek, passed through the base of the skull and the front of his brain, left through the top of his head and landed more than a hundred feet away. Gage spoke within minutes, rode upright in an ox cart to the town hotel and climbed out of it with a little assistance; John Harlow, the town physician, pronounced him cured in under two months.

Gage could see with his right eye, walk firmly, use his hands and speak without difficulty. What had gone was the man. Harlow's account, written twenty years afterwards, describes someone fitful, irreverent, grossly profane where profanity had not been his custom, impatient of restraint or advice, obstinate and vacillating at once, devising plans that were abandoned as soon as they were made. Friends said that "Gage was no longer Gage", and his employers would not take him back, not for want of skill but because of the change in his mind. He worked on horse farms, was exhibited with his iron at Barnum's Museum, drove a stagecoach in Chile, and died of seizures in San Francisco in 1861, at thirty-eight.

What the case dissociates is Damasio's interest. Character was destroyed and the instruments of mind were not: observing social convention, behaving ethically and choosing in one's own interest turned out to need a particular brain system, and the need could go unmet in a man with no defect of attention, perception, memory, language or intelligence. The argument Gage's survival provoked was about whether anything in the brain could be localised at all, and it passed over what he was pointing at.

No autopsy was performed. Harlow heard of the death five years late, asked Gage's sister to have the body exhumed, and received the skull and the tamping iron, which have been at the Warren Medical Museum at Harvard ever since. In 1994 Hanna Damasio finished the work: Albert Galaburda photographed and measured the skull, she and Thomas Grabowski rebuilt it in three dimensions and derived the brain that best fitted it, and with Randall Frank she drove a simulated iron of Gage's dimensions through that brain along the trajectories the wounds allowed.

The damage was bilateral, heavier on the left, and concentrated in the ventral and inner surfaces of the prefrontal cortices, the sector Damasio calls ventromedial. Motor and premotor cortex and the region carrying Broca's area were untouched, as David Ferrier had argued against everyone who cited Gage as proof that nothing in the brain is localised; so were the lateral prefrontal cortices, whose damage disrupts attention, calculation and shifting from one task to another. Gage had lost the tissue his contemporaries assumed was the best fitted of any part of the brain to sustain such an injury.

A modern Phineas Gage

Elliot was in his thirties, a husband and father with a job in a business firm and a standing as the model for younger colleagues, when the headaches began and his work started to have to be finished by other people. The tumour was a meningioma the size of a small orange, growing from the midline above the nasal cavities and pressing both frontal lobes upward. It was removed, and with it the frontal tissue it had destroyed: the orbital and medial sectors on both sides and, on the right, the core of white matter beneath the cortex. Everything else was intact, from the motor cortices and Broca's area to the other three lobes and the deep nuclei. As in Gage, the ventromedial sector had taken the disproportionate share.

What followed was a life coming apart at a steady rate. He needed prompting to get ready for work and could not be trusted with a schedule. Given a client's documents to sort, he might read one of them carefully and intelligently for an entire day, or spend an afternoon deliberating whether to classify by date, by size or by pertinence; the step at which he stopped was being carried out too well, at the expense of the purpose. He was dismissed, took other jobs and was dismissed again, put his savings into a venture with a disreputable partner against everyone's advice and lost all of it, and was twice divorced. Then his disability payments were refused, on the reasonable ground that several professionals had found his mental faculties intact.

Damasio predicted that Elliot would pass the general tests and fail the small number that detect frontal dysfunction. He failed none of them.

Jeffrey Saver then built a battery for the social knowledge itself: generating responses to predicaments, considering the consequences of an action, finding effective means to a social end, predicting how a situation would turn out, and reasoning through a moral dilemma. Elliot matched controls on four, exceeded them on consequences, and reached a late-conventional level on the Standard Issue Moral Judgment Interview. At the end of a session in which he had produced a long list of valid and implementable options, he smiled and remarked that after all of it he still would not know what to do.

Damasio took time away from the problem and came back to find he had been attending to the instruments of Elliot's rationality and not to his emotions. Elliot described the ruin of his own life as an uninvolved spectator, and Damasio found himself more distressed by the stories than the man telling them. This was not a stiff upper lip; there was no turmoil being hushed. When Daniel Tranel showed him images of earthquakes, burning houses and injured people, Elliot volunteered that subjects which had once produced a strong response in him no longer produced any. Damasio summarises the condition as to know but not to feel.

Eleven more patients with the same kind of prefrontal damage followed, and not one separated the two defects. Nor does the historical record: a New York stockbroker described in 1932 who lost the same regions to a meningioma, became boastful and casually cruel and never worked again, and patients damaged at sixteen and around birth whose social development stopped or never began. The prefrontal leucotomies Egas Moniz introduced in 1936 abolished anxiety while sparing language and memory, and left patients flat in emotion and less decisive than before.

Then the argument widens past the frontal lobe. Anosognosia, the inability to acknowledge one's own illness, follows damage to the somatosensory cortices of the right hemisphere, in the insula and the parietal region. Damasio's patient DJ, completely paralysed on the left side, would report that her movements were normal, hunt around for the arm when asked to move it and, having looked at it, ask whether he really wanted it to move by itself. Such patients take the news of a stroke or a spreading cancer with equanimity and never with anguish, and their planning for their own future is wrecked. A patient with bilateral amygdala damage showed the same combination at normal intelligence.

Emotions and feelings

The usual division of labour puts basic regulation downstairs in the old brain and deliberation upstairs in the neocortex, and Damasio's evidence does not fit it. Nature appears to have built the apparatus of rationality not merely on top of the apparatus of biological regulation but out of it and with it, since the innate dispositions of the brain stem, hypothalamus and limbic system shape the modifiable circuits above them through modulator neurons distributing dopamine, norepinephrine, serotonin and acetylcholine across the cortex. Emotion and feeling are where that contact becomes visible.

William James, Damasio holds, had the mechanism. Strip from a strong emotion all the feelings of its bodily symptoms, James wrote, and nothing is left but a cold and neutral state of intellectual perception; try to picture rage with limp muscles and a placid face. What James left out was the mental evaluation that stands between a situation and the emotion it provokes, and any alternative to running the response through the body every time. Damasio supplies both, and splits emotions in two. Primary emotions are preorganised and depend on the amygdala and the anterior cingulate. Nobody is wired for bear fear as such; what is wired is a response to features — size, a wide span, a kind of motion, a growl — which is why a chick that has never seen an eagle hides its head when a wide-winged shape passes overhead at a certain speed. Secondary emotions arrive later in a life, once we have begun forming systematic connections between categories of situation and the primary emotions. Images of a situation held in thought activate acquired dispositions in the prefrontal cortices, which signal the amygdala and cingulate, which do what they already knew how to do: place the viscera in the appropriate state, arrange the face and posture, release hormones, and fire the neurotransmitter nuclei that alter the speed and style of thought.

The distinction between emotion and feeling carries the rest of the book. An emotion is the collection of changes in body state induced in the organs by a dedicated brain system responding to a thought about some particular thing. A feeling is the continuous monitoring of those changes as they unfold over seconds and minutes, set beside the image that started the cycle — a face, a melody, a piece of news — so that the body's condition ends up qualifying whatever else is being perceived. Completing it is a parallel shift in the mode of thinking: with a negative body state images come slowly and inference is poor, which at the extreme is depression; with a positive one they come fast and various, which at the extreme is mania. Beneath both Damasio posits a third kind, background feeling, the state of the body between emotions and the sense of being alive, which we notice so little that we can nonetheless answer at once when asked how we feel. Anosognosia is what its loss looks like: unable to update the map of a body that has changed, the patient reports the body he last knew.

That the body is doing this work rather than merely accompanying it is what the supporting evidence is for. Paul Ekman, instructing people to move particular facial muscles without telling them why, produced in them the feeling that matched the expression they had unknowingly composed. Against this Damasio sets a second mechanism he thinks development installs: an "as if" loop, in which the prefrontal cortices and amygdala tell the somatosensory cortices to organise themselves as though a body state had been enacted and reported, bypassing the body and its cost in time and energy. He treats the body loop as basic and the "as if" loop as a later economy rather than the whole story, on the ground that the brain cannot predict what a barrage of neural and chemical commands will do in a body whose local chemistry it does not fully represent. It waits for the body to report; if all feelings were of the second kind they would be rebroadcasts rather than live performances.

The somatic-marker hypothesis

Set out the problem the way a decision course would. You own a business and must decide whether to meet a client who would bring valuable work and happens to be your closest friend's enemy. The commonsense method is to take the scenarios apart, estimate gains and losses at several points in the projected future, discount them, convert unlike quantities into a common currency and compare. Damasio's objection is not that this is wrong but that it does not run: attention and working memory are limited, and the intermediate ledgers vanish from the slate while you consult the next ones, so that at best the decision takes far longer than the day allows and at worst you never reach one. The cool strategy, he suggests, describes how a ventromedial patient goes about deciding rather than how anyone else does.

What happens instead, on his account, is that before any weighing begins, the bad outcome attached to an option comes into mind however briefly and you experience an unpleasant gut feeling. He calls the state somatic because it is about the body, and the signal a marker because it marks an image. A negative marker beside an imagined outcome works as an alarm bell, a positive one as a beacon of incentive. It forces attention onto the outcome and can eliminate the option immediately, so that deduction and cost-benefit analysis are left to operate over a much smaller field.

Damasio is careful about what the hypothesis does not claim, and the 2005 preface is careful again where readers had gone past him. Markers do not deliberate. They are not sufficient for deciding, since in most cases a process of reasoning and final selection still follows, and the hypothesis says nothing about those later steps; what markers do is raise the accuracy and efficiency of the process, and their absence lowers it. He did not propose that the help emotion gives reason must be unconscious — his first version equated somatic markers with conscious gut feelings, while leaving room for a covert variety. He did not treat skin conductance responses as markers, but as indices of them. And he did not propose that emotion substitutes for reasoning.

The markers themselves are learned, mostly during education and socialisation, by connecting classes of situation to classes of body state under an internal preference system that avoids pain and seeks pleasure and an external regime of reward and punishment. Developmental psychopathy, in which the threshold for emotion is so high that the person appears unflappable, is Damasio's example of a build that failed from the start. The prefrontal cortices are the right place for the mechanism because they are the one region privy to everything it needs: signals from every sensory region in which images are formed, the somatosensory cortices included; signals from the bioregulatory sectors, so that innate preferences reach the reasoning apparatus rather than sit outside it; categorisations of the contingencies of one particular life, which is what makes the marking personal; and projections to the effectors that enact a body state.

A marker can also act covertly, without being attended and so without becoming a feeling, inhibiting the circuits that mediate approach or enhancing those that mediate withdrawal. That covert route is what Damasio names as the source of intuition, and the passage he quotes for it is Poincaré on mathematical creation: sterile combinations never present themselves to the inventor's mind at all. The second half of the hypothesis, easy to miss, is that the same signal boosts working memory and attention, holding the field of options stable long enough to be reasoned over. The mechanism cuts both ways. One of his ventromedial patients drove calmly and correctly across an ice patch moments after watching the car ahead of him panic, brake and slide into a ditch. The next day, asked to choose between two appointment dates a few days apart, he produced half an hour of reasons for and against each, until the laboratory quietly told him which to take.

Testing the somatic-marker hypothesis

The first test, with Daniel Tranel, used skin conductance, the small rise in sweat-gland secretion that follows a change of body state and lowers the skin's resistance to a current; startle and a deep breath established first that the machinery was intact in the frontal patients. Subjects then sat before a long sequence of projected slides, most of them bland scenery and abstract patterns, with disturbing images appearing at random intervals. Controls generated abundant responses to the disturbing pictures and none to the bland. The patients with frontal damage produced nothing; their recordings were flat. It was not inattention, since afterwards they could describe the content in detail, name the emotion a picture ought to arouse, and place a given slide in the order of the set. One said plainly that in spite of realising the content ought to be disturbing, he himself had not been disturbed, and that his flesh no longer responded to such things as it once had.

The second test was a card game designed by Antoine Bechara and refined with Hanna Damasio and Steven Anderson. The player sits before four decks labelled A, B, C and D with a loan of two thousand dollars in play money, and turns cards one at a time from any deck until told to stop. Every card pays; some also demand a payment back. Nothing is disclosed — not the amounts, not the order, not the running total, not that the game ends after a hundred turns — and notes are not allowed.

Ordinary players sample all four, are drawn for a while to the hundred-dollar decks, and within the first thirty moves switch to the fifty-dollar decks and stay there, without ever being able to compute the balance. They develop a hunch that A and B are dangerous. The patients did the reverse: they turned more and more cards from A and B, and by halfway through they were bankrupt and taking further loans. Elliot played, and still describes himself as a conservative, low-risk person; normal subjects who called themselves gamblers played prudently. At the end of the game he knew which decks were bad, and months later, with different cards and labels, he played the same way again.

Bechara then inverted the schedule so that punishment came first with every card and reward was interspersed. Elliot and the other frontal patients now performed like controls, which disposes of the simplest explanation, that they had become insensitive to punishment. The fine grain of the first task points the same way: immediately after paying a penalty the patients avoided that deck exactly as controls did, and then, unlike controls, came back to it. What was missing was not the response to punishment but its conversion into a prediction. Damasio's name for the result is myopia for the future.

The third test put the two instruments together: Hanna Damasio suggested recording skin conductance during the game, and Bechara and Tranel did it. Both groups responded to each reward and each penalty in the seconds after it arrived, which confirms again that the patients respond to what is in front of them. Then, after some tens of turns, something began in the controls only. In the seconds while they deliberated over a card from a bad deck, before turning it, they generated a response, and it grew as the game went on. The patients generated none.

Damasio does not close the account further than the data allow. He does not know how the prediction develops: whether the player first estimates a deck cognitively and the somatic signal is then attached, or whether a covert weighing of good and bad body states runs first and helps the player into a theory about the game. He inclines to the second, and notes that the first cannot be the whole story, since the patients cannot play normally even once they know which decks are bad.

Descartes' error

The last part of the book takes the body from being the theatre of emotion to being the ground reference for mind in general. Damasio will not grant a normal mind to the philosophers' brain in a vat, and his reason is instructive: to give it one you would have to build a body surrogate capable of delivering realistic configurations of body signals, which concedes the point. The self, on this account, is not a small person inspecting the contents of the mind but a biological state reconstructed from the ground up at each moment, out of the memories that define a person and the primordial representations of the body.

The error is then named. Taken literally, cogito ergo sum makes thinking the substrate of being, and Descartes said so without ambiguity a few lines later: that the soul is a substance whose whole essence is to think, that it needs no place and depends on no material thing, and that even if body were not, the soul would not cease to be what it is. Being came first, Damasio replies, in evolution and in each life; we are, and then we think, and we think only inasmuch as we are. What he objects to is the abyssal separation between body and mind, and the modern descendants that keep it alive without calling themselves dualist: the mind as a software program run on the hardware of a brain, and the brain explained in its own terms with the rest of the organism left outside the account. The postscript locates the practical damage in medicine, where schools teach psychopathology to students who have never been taught normal psychology. His illustration is a man with intractable trigeminal neuralgia, crouched and immobile before a leucotomy and two days afterwards playing cards, who reported cheerfully that the pains were exactly the same and that he now felt fine — the image of pain intact, the suffering abolished.

Damasio denies that any of this cheapens what it explains: the dependence of high reason on low brain does not turn high reason into low reason, and the edifice of ethics does not collapse because acting on a principle requires circuitry in the brain core. He refuses, too, the reading that would make the book a licence for feeling over thinking. Knowing how much feelings do in the work of reason is an argument for cultivating reason further, and for protecting it from the deliberate manipulation of normal ones.

Since 1994

The clinical core has held. Damage to ventromedial prefrontal cortex reliably produces the combination Damasio described, and the region has become the standard site for subjective value in the decision neuroscience that grew up afterwards. The 2005 preface reports a discovery the book provoked: parents who recognised their grown children in these patients wrote in, and the cases published in 1999 showed early frontal damage producing a worse pattern, since the adult-onset patients knew the rules and failed to act on them while the early-onset patients had never learned them. Later work found ventromedial patients choosing, in moral dilemmas, the outcome that saves more lives in exactly the cases where the harm is personal and most people cannot endorse it.

The card game has had a harder time, and the critiques accumulated as it spread through neuropsychology. The claim that players choose advantageously before they know why did not survive better questioning: asked more finely what they knew, most participants turned out to have enough conscious knowledge to account for their choices well before the point at which the original studies reported a mere hunch. The anticipatory response was shown to track the size of the reward or penalty on offer rather than the long-run value of a deck. The decks confound two things, since one of the bad ones punishes rarely and heavily and healthy players over-select it. A review in 2006 concluded that the evidence for somatic markers as the book states them is thinner than the presentation suggests, and patients whose autonomic responses are abolished by pure autonomic failure have been reported to play normally, which the body loop does not predict and the "as if" loop absorbs at the cost of being hard to test. Ekman's facial feedback has had a difficult decade, a large multi-laboratory replication of its best-known version failing before later meta-analyses recovered a small effect.

What has travelled furthest is the general position rather than the specific mechanism. The neglect of emotion the preface describes ended within a few years of the book, with LeDoux's and Panksepp's volumes and then a whole field, and the proposition that emotion is in the loop of reason rather than an intruder in it is now the default. Damasio's separation of emotion from feeling has aged better than the categories around it: the search for discrete neural signatures of basic emotions has largely failed, and LeDoux now insists that defensive circuits in the amygdala do not produce the feeling of fear. The interoceptive claim — that the brain builds a reading of the body's condition, and that this reading is the substance of feeling — has been the most productive part of the book: the pathway from the body through the posterior to the anterior insula was mapped in the following decade, and predictive accounts of interoception have made bodily regulation a candidate explanation of selfhood. The Gage story has been revised in the other direction: Malcolm Macmillan's history argued that the standard version overstates the ruin, since holding a demanding stagecoach job in Chile for years implies substantial re-adaptation, and in 2012 the trajectory was recomputed with diffusion imaging, which put the loss of white-matter connections well beyond the loss of cortex. Damasio went on to write the books this one keeps promising: an account of consciousness and the self in 1999, a return to Spinoza and the biology of feeling in 2003, and, in 2018, the argument the postscript sketches in a paragraph, that homeostasis is the blueprint from which cultures build their values.