The brain–behaviour interface as a discipline. From Phineas Gage's tamping iron to functional imaging — the patients, the lesions, the syndromes, and the clinicians who built the field.
The clinical discipline that infers the architecture of the human mind from the patterns of damage that disrupt it.
Neuropsychology sits between neurology and psychology. It uses the tools of cognitive psychology — controlled tasks, careful measurement of behaviour — applied to patients with neurological disease or injury, and uses the results to build a model of how the brain implements thought, language, perception, memory, and emotion.
The field's central inferential move is the dissociation. If a patient with damage in region A loses ability X but retains ability Y, and a patient with damage in region B loses Y but retains X, the two abilities depend on partly different brain systems. Almost everything we know about the modular structure of human cognition has been built from such pairings.
On 13 September 1848, near Cavendish, Vermont, the railroad foreman Phineas Gage, 25, was preparing a blasting charge when a premature explosion drove a 43-inch iron rod through his left cheek and out the top of his skull. He survived. The rod travelled through his left frontal lobe.
The treating physician John Martyn Harlow, in his 1848 and 1868 reports, described a man whose intelligence and language were preserved but whose personality had been transformed. The pre-injury Gage had been "a great favourite" with his employers — efficient, capable, balanced. The post-injury Gage was "fitful, irreverent, indulging at times in the grossest profanity," unable to plan, abandoning each scheme as soon as it was begun. "Gage was no longer Gage."
The case was the field's founding observation: the frontal lobes contribute something specific to personality, planning, and the regulation of social behaviour, distinct from intelligence in the narrow sense. The exact lesion — reconstructed from Gage's preserved skull by Hanna Damasio's lab in 1994 — implicated ventromedial prefrontal cortex.
In April 1861 the Parisian surgeon Paul Broca presented to the Société d'Anthropologie the case of a 51-year-old man named Louis Victor Leborgne. Leborgne had been admitted to the Bicêtre Hospital decades earlier and had progressively lost the capacity to speak, retaining only the syllable tan, which he repeated. He had also lost the use of his right side. He understood what was said to him.
Leborgne died on 17 April 1861. Broca performed the autopsy the next day. The lesion was in the posterior part of the left inferior frontal gyrus — the area now known as Broca's area.
The case settled an argument. The phrenologists Gall and Spurzheim had claimed cognitive functions were localised; their critics, led by Pierre Flourens's experimental work, had argued for equipotentiality. Broca's case provided the first clean clinical evidence that a specific cognitive function — articulate speech production — depended on a specific cortical region. Modern cortical localisationism begins here.
In 1874 the 26-year-old German neurologist Carl Wernicke published Der aphasische Symptomencomplex, describing patients with the opposite syndrome to Broca's: fluent but empty speech, with severe impairment of language comprehension. The lesion was in the posterior superior temporal gyrus — Wernicke's area.
The two-area scheme allowed Wernicke to predict a third syndrome: damage to the connecting fibres should disrupt repetition of speech while leaving comprehension and spontaneous speech intact. Conduction aphasia was duly described and is now associated with damage to the arcuate fasciculus.
The Wernicke–Geschwind model of language — comprehension in posterior temporal cortex, production in posterior frontal cortex, connecting fibres between — dominated clinical neurology for a century. Twenty-first-century imaging has complicated it (left frontotemporal language networks are broader and more distributed than the classical model suggests) but has not displaced it as a clinical heuristic.
Between Broca and World War I, European clinical neurology produced a steady catalogue of lesion-syndrome correlations. Hugo Liepmann described apraxia (loss of the capacity to perform learned motor sequences in the absence of weakness) in 1900, localising it to left parietal cortex. Joseph Babiński described anosognosia — the inability of right-hemisphere stroke patients to acknowledge their own paralysis — in 1914.
The First World War, which produced a generation of head-injured soldiers, expanded the empirical base. Kurt Goldstein's work with brain-injured veterans, summarised in Der Aufbau des Organismus (1934), introduced a more holistic perspective: lesions disrupted not just specific functions but the patient's overall capacity for what Goldstein called the "abstract attitude."
By the 1930s clinical neuropsychology had its modern shape. The discipline relied on careful single-case observation, detailed bedside testing, and post-mortem confirmation. The methodological vocabulary — dissociation, double dissociation, functional architecture — was largely in place.
Karl Lashley, working at Chicago and then Harvard from the 1920s through the 1950s, set himself the question: where in the brain is a memory stored? He trained rats to run mazes, then removed varying amounts of cortical tissue to identify the location of the memory trace — the engram.
The result was the most influential null finding in psychology. Maze-running performance declined in proportion to the amount of cortex removed, not the location. Lashley summarised his thirty-year search in 1950 — "In Search of the Engram" — with the rueful conclusion that "learning just is not possible."
The conclusion was wrong, but the lesson was right for its time. Memory is not stored at a single cortical location. The engram, as later work by Donald Hebb (Lashley's student) and Eric Kandel would show, is distributed across networks of synapses. Localisation in neuropsychology is real, but its grain is finer and its geometry more network-like than the early lesion tradition assumed.
The 1950s saw neuropsychology absorb the techniques of single-cell electrophysiology. David Hubel and Torsten Wiesel at Harvard, beginning in 1959, recorded from individual neurons in the cat visual cortex and discovered the orientation-selective simple cells and complex cells that gave rise to the modern understanding of hierarchical visual processing. Their Nobel followed in 1981.
The discoveries reshaped neuropsychology's theoretical vocabulary. Cognitive functions were no longer black boxes attached to cortical regions; they were the work of populations of neurons with definite tuning properties, connected in patterns that could in principle be charted.
The methodological pluralism that defines the modern field — lesion-deficit analysis, single-cell recording, electrical stimulation, eventually imaging — dates from this period. No single technique was now expected to deliver the whole picture.
The neurosurgeon Wilder Penfield at the Montreal Neurological Institute, treating epilepsy patients from the 1930s through the 1950s, developed a surgical technique that required mapping the cortex of awake patients with electrical stimulation before resecting epileptogenic tissue.
The maps that emerged — the cortical homunculus for somatosensory and motor cortex, the language-related sites of the dominant hemisphere — became the canonical illustrations of cortical functional organisation. Penfield also reported, more controversially, that stimulation of temporal lobe sites in some patients evoked vivid recollections of past experiences, complete with sensory detail.
The reminiscence findings have been disputed. Most contemporary work holds that what Penfield evoked was not stored memory but constructive imagery shaped by the patient's expectation. The cortical maps, however, were durable; their refinement by intraoperative mapping continues in modern neurosurgery.
The single most important patient in the history of memory research. Henry Gustav Molaison, born 1926, suffered intractable epilepsy. In 1953 the neurosurgeon William Beecher Scoville performed an experimental bilateral medial temporal lobectomy, removing both hippocampi and surrounding tissue. The seizures improved. The cost was catastrophic.
From the day of surgery until his death in 2008, H.M. could not form new explicit memories. He could hold information for as long as he attended to it; the moment his attention shifted, the trace was gone. His IQ was preserved. His personality was preserved. His memory of events before about 1942 was largely intact. The memories of his last decade before surgery were patchy.
What the case established: the medial temporal lobes are necessary for the consolidation of new declarative memories but not for their long-term storage and not for short-term holding. Memory is not a single faculty; it has separable systems. Almost everything modern cognitive neuroscience knows about memory begins with this dissociation.
The young Cambridge-trained psychologist Brenda Milner, working at the Montreal Neurological Institute under Donald Hebb and Wilder Penfield, was the clinician who studied H.M. most closely. Her 1957 paper with Scoville, "Loss of recent memory after bilateral hippocampal lesions," established the case in the literature.
Milner's 1962 demonstration that H.M. could learn the mirror-drawing task — improving across days even though he had no recollection of having done it before — was the first clear evidence that procedural memory was dissociable from declarative memory. The taxonomy of memory systems we now teach (declarative versus procedural; episodic versus semantic; short-term versus long-term) descended from this finding and the work of Endel Tulving (semantic/episodic, 1972) and Larry Squire (declarative/procedural, 1980s).
Milner remained a working scientist into her hundredth year. She was awarded the Kavli Prize in 2014, alongside John O'Keefe and Marcus Raichle. Her career — over seventy years at the Montreal Neurological Institute — is among the longest in the discipline's history.
Norman Geschwind, working at Boston Veterans Affairs Hospital and later Harvard, wrote a pair of papers in 1965 — "Disconnexion syndromes in animals and man" — that revived an idea Wernicke had originally proposed and the early-twentieth-century holists had buried.
The thesis: many neuropsychological syndromes are the result not of damage to a localised processor but of damage to the white-matter tracts connecting otherwise intact cortical regions. Pure alexia — the inability to read in the absence of any other language deficit — was Geschwind's signature example. The patient's right-hemisphere visual cortex still saw the words; the left-hemisphere language areas could still process them; but the corpus-callosal fibres that connected them were damaged. Words seen with the right hemisphere could not reach the left to be read.
Geschwind's revival of disconnectionism reshaped American neurology in the 1960s and 1970s. Modern white-matter imaging (diffusion tensor MRI) has restored most of his framework as quantitative anatomy.
Antonio Damasio's book Descartes' Error (1994) was the field's most-read trade publication of the 1990s. The work grew out of decades of clinical experience with patients who, after damage to ventromedial prefrontal cortex, retained their intelligence and factual knowledge but became disastrous decision-makers, particularly in social and personal life — Phineas Gage's modern descendants.
Damasio's somatic marker hypothesis proposed that emotion is not the enemy of reason but its substrate. Decisions in real-world contexts are guided by quick somatic signals — felt valences attached to imagined options — that constrain the search through possibilities. Patients with ventromedial damage lose the marker signal. Their reasoning continues unimpaired in the abstract. In life, they cannot tell good options from bad ones.
The Iowa Gambling Task, developed by Damasio and his colleagues, became a standard instrument for detecting the deficit. The wider claim — that emotion is constitutively involved in normal decision-making — has been broadly absorbed into cognitive neuroscience.
The British-born neurologist Oliver Sacks, working at Beth Abraham Hospital in the Bronx and later at Albert Einstein College of Medicine, did more than any clinician of his generation to bring neuropsychology into general literary culture.
His first major book, Awakenings (1973), described the temporary recovery of patients with post-encephalitic Parkinsonism after L-DOPA. A Leg to Stand On (1984) was an account of his own experience with a peripheral-nerve injury and the strange disowning of the affected limb. The Man Who Mistook His Wife for a Hat (1985) — twenty-four short case studies — became a permanent fixture of bookshop neurology.
Sacks's method was unusual. He wrote in the tradition of A. R. Luria's The Mind of a Mnemonist — clinical biography rather than case report — treating each patient as a person rather than a diagnostic instance, and finding in the alteration of cognition a window onto the structure of the unimpaired mind.
The title essay of Sacks's 1985 collection described a music teacher referred to as Dr. P, who suffered from visual agnosia — the inability to recognise objects despite intact basic vision. Asked to identify a glove, Dr. P examined it carefully and said: "A continuous surface, infolded on itself. It appears to have five outpouchings, if this is the word." Pressed, he conceded it might be "a container of some sort."
The eponymous moment came when Dr. P, leaving the consulting room, reached for his hat and grasped his wife's head, mistaking one for the other.
The clinical lesson is that visual recognition is not a single process. Dr. P's primary visual cortex was intact; he could see edges, surfaces, colour. What he had lost was the higher-level integration that turns visual features into recognised wholes — what cognitive neuroscience now associates with the lateral occipital complex and the fusiform gyrus. The case is a textbook illustration of apperceptive agnosia.
Lawrence Weiskrantz at Oxford, working with patient D.B. from the early 1970s, demonstrated that patients with damage to primary visual cortex (V1), who reported being completely blind in the affected visual field, could nevertheless reach toward, point at, or guess the orientation of stimuli they denied being able to see — at performance levels well above chance.
The phenomenon, named blindsight in Weiskrantz's 1974 paper with Warrington, is now well replicated. The standard interpretation invokes parallel pathways: visual information also reaches subcortical structures (the superior colliculus) and extrastriate cortex by routes bypassing V1, sufficient to support some kinds of behaviour but not the conscious visual experience that V1 input enables.
Blindsight has become a flagship case in the philosophy and neuroscience of consciousness. It shows that visual processing can occur without phenomenal vision, putting pressure on theories that identify perception with information uptake.
Prosopagnosia is the selective inability to recognise faces, including, in severe cases, one's own. The acquired form follows damage typically to bilateral or right fusiform cortex. The developmental form (estimated population prevalence 2%) appears without an identified lesion.
Patients can identify objects and read facial expressions but cannot match identity from face to face. They use compensatory strategies — voice, gait, hairstyle, distinctive features. Bill Choisser's online community in the 1990s and the work of Brad Duchaine and Ken Nakayama at Harvard documented the developmental form as a serious and prevalent condition.
The case for face-specific cortical machinery was sharpened by Nancy Kanwisher's 1997 fMRI identification of the fusiform face area, a cortical region that responds preferentially to faces. The dissociation — patients with prosopagnosia retain object recognition; some patients with object agnosia retain face recognition — is one of the cleanest in cognitive neuropsychology.
Synesthesia is the condition in which stimulation in one sensory modality reliably evokes experience in another — the most common forms being grapheme-colour (letters and numbers seen in particular colours) and chromesthesia (sound evoking colour).
Once dismissed as confabulation or metaphor, the phenomenon was rehabilitated by the careful work of Richard Cytowic, Simon Baron-Cohen, and (with subsequent neuroimaging) V. S. Ramachandran and Edward Hubbard in the late 1990s and 2000s. The synesthetic associations are stable across years (verified by retest), highly specific, and accompanied by altered patterns of cortical activation in fMRI.
Modern accounts emphasise increased structural and functional connectivity between regions that, in non-synesthetes, develop more independent specialisations — particularly between the visual word form area and colour-selective regions of fusiform cortex in grapheme-colour synesthesia. The condition runs in families and is several times more common in artists and writers than in the general population.
Beginning in the early 1960s, a small number of patients with severe intractable epilepsy underwent surgical section of the corpus callosum — the major fibre tract connecting the two cerebral hemispheres. Roger Sperry at Caltech, with the young Michael Gazzaniga, designed experiments that took advantage of the divided cortex.
The classic finding: when an object was presented to the right visual field (and so the left, language-dominant hemisphere), the patient could name it. Presented to the left visual field (right hemisphere), the patient denied seeing anything — but could pick out the object by feel with the left hand. Each hemisphere had its own access to the world; only the left hemisphere could speak.
Sperry's 1981 Nobel acknowledged the work as one of the foundational demonstrations of hemispheric specialisation. Gazzaniga went on to develop the influential interpreter hypothesis — that the left hemisphere generates running narrative explanations of behaviour, including behaviour driven by the right hemisphere it cannot directly access.
After damage to the right parietal lobe, particularly the inferior parietal lobule, patients may exhibit hemispatial neglect: a failure to attend to, respond to, or report stimuli on the contralesional (typically left) side of space. The neglect can extend to the patient's own body — patients may shave only the right side of their face, eat only the right half of a plate of food, or deny ownership of their left arm.
Edoardo Bisiach and Claudio Luzzatti's 1978 study had Milanese neglect patients imagine themselves standing at one end of the city's Piazza del Duomo and describe what they saw. Patients reported only the buildings on the right side of their imagined view. Asked to imagine standing at the opposite end, they now reported the previously neglected buildings — and ignored those they had previously named. The neglect was of a representational frame, not of the physical world as such.
The case shows that attention has a representational geometry. The right parietal cortex contributes to a spatial framework that subsequent cognition operates within.
The left hemisphere's specialisations — language, sequential analysis, fine motor control of the right hand — were charted first because their disruption produces dramatic, easily recognised deficits. The right hemisphere's contributions were charted later.
The contemporary picture: the right hemisphere is dominant for spatial attention (hence neglect); for the integration of facial expression and prosody in social cognition; for the appreciation of metaphor, narrative coherence, and inference beyond the literal; and for global as opposed to local visual processing.
Patients with right-hemisphere damage are often described as conversationally normal but socially off. They may interpret idioms literally, miss the point of stories, fail to read emotional tone, and produce flat, factually correct but pragmatically inappropriate speech. The constellation has been called cognitive-communicative disorder and represents an important and historically underappreciated branch of clinical neuropsychology.
The classical Broca/Wernicke/conduction taxonomy remains a useful clinical heuristic but has been substantially refined. Modern aphasiology distinguishes fluent and non-fluent aphasias, then subclassifies on phonology, syntax, and semantic content.
The primary progressive aphasias — distinct neurodegenerative syndromes affecting language without dementia at onset — were characterised by Marsel Mesulam at Northwestern from 1982 onward. The current classification recognises three variants: nonfluent/agrammatic, semantic, and logopenic. Each maps to a distinct pattern of cortical atrophy and points to a distinct neurodegenerative pathology (FTLD-tau, FTLD-TDP, and Alzheimer's, respectively).
Treatment has improved at the margins. Speech-language therapy in the post-stroke chronic phase produces measurable benefit (the COMPASS trial, 2018). Constraint-induced aphasia therapy and intensive home practice have produced gains in patients years out from the initial event. The wider question — why the language network reorganises in some patients but not others — remains an active research area.
Frontotemporal dementia (FTD) is the leading cause of dementia under 60. The behavioural variant — bvFTD — presents not with memory loss but with progressive personality change: disinhibition, loss of empathy, ritualistic behaviour, dietary changes, social inappropriateness. Insight is typically absent.
The clinical presentation often mimics late-life psychiatric disease. Patients are misdiagnosed with depression, mania, or personality disorder for years before the underlying neurodegeneration is recognised. The Lund-Manchester criteria (1994) and the revised 2011 international criteria standardised diagnosis.
The pathology is heterogeneous — TDP-43 inclusions in roughly half of cases, tau in most of the remainder. About a third of cases are familial, with mutations in MAPT, GRN, and C9ORF72 the principal causes. The condition's recognition has substantially altered the field's picture of frontal-lobe function: the personality changes Harlow described in Phineas Gage are seen, in slow motion, in the bvFTD clinic every week.
Alois Alzheimer's 1906 description of Auguste Deter, a 51-year-old woman with progressive dementia and the post-mortem plaques and tangles that bear his name, founded the disease. The 21st-century picture: roughly 50 million cases worldwide, projected to exceed 100 million by 2050.
The neuropsychological profile is stereotyped: early impairment of episodic memory (the entorhinal cortex and hippocampus are first affected), spreading to semantic memory, language, and visuospatial function as posterior cortical regions are implicated. Frontal and motor function are typically preserved until late.
The amyloid cascade hypothesis — that beta-amyloid accumulation drives the disease — has dominated treatment research for thirty years and produced, after a long history of failure, the first amyloid-clearing antibodies (aducanumab 2021, lecanemab 2023, donanemab 2024). Effects on cognition have been modest. Whether amyloid is causal or a co-traveller of the upstream pathology remains debated.
The textbook adult brain was, into the 1990s, regarded as essentially fixed. Stroke rehabilitation aimed at compensation rather than reorganisation. The picture has changed.
Edward Taub's constraint-induced movement therapy (developed at the University of Alabama from the late 1980s) demonstrated that immobilising the unaffected limb of a hemiparetic patient and requiring intensive use of the affected limb produced measurable cortical reorganisation and functional gains, often years post-stroke. The result challenged the view that the post-acute window was closed.
Modern stroke rehabilitation has absorbed the lesson. Intensive task-specific practice, mirror therapy, robot-assisted training, and non-invasive brain stimulation (transcranial direct-current stimulation, repetitive transcranial magnetic stimulation) all aim to drive plasticity in the spared cortex. The gains are real but variable. The patients who recover most are those with sparing of relevant white-matter tracts and intensive early therapy.
The 1990s saw functional neuroimaging move from a laboratory novelty to the dominant method of cognitive neuroscience. Positron emission tomography (PET), developed in the 1970s and refined by Marcus Raichle's group at Washington University, allowed regional cerebral blood flow to be measured non-invasively. Functional MRI, exploiting the BOLD signal first reported by Seiji Ogawa in 1990, replaced PET as the workhorse method through the 2000s.
The decade of activation studies — show subjects two conditions, subtract the brain images, name the resulting region a "centre" for the difference — produced both real progress and a methodological backlash. The early-2010s replication and statistical-power critiques (Eklund et al., 2016; Vul et al., 2009) chastened the field.
Modern imaging neuropsychology emphasises networks rather than single regions, with a substantial role for resting-state functional connectivity (Bharat Biswal, 1995; Marcus Raichle's default mode network, 2001). The patient lesion remains the gold standard for causal inference; imaging supplies the systems-level architecture.
The first sustained motor brain–computer interface in a human — BrainGate, an Utah-array implant in motor cortex enabling a tetraplegic patient to control a computer cursor — was reported by John Donoghue's group in 2006. The Cyberkinetics device was later developed by a Brown-Stanford-MGH consortium.
The 2020s have seen substantial acceleration. The 2021 BrainGate paper (Willett, Henderson, et al., Stanford) demonstrated handwriting decoded from motor cortex at 90 characters per minute. The 2023 Edward Chang group at UCSF restored speech to a paralysed patient via an electrocorticographic implant decoding intended phonemes. Neuralink's first human implant occurred in 2024.
The clinical neuropsychology question is no longer whether BCIs will work but which patients should get them, what they should control, and how the eventual learning curve compares to other assistive technologies. The discipline that began with patients losing function after lesions is finishing the century giving function back.
The contemporary clinical neuropsychologist is a doctorally trained psychologist (typically PhD or PsyD) who assesses patients with known or suspected brain disease using standardised cognitive testing. The standard battery includes premorbid IQ estimates (TOPF, NART), measures of attention (Trail Making, Stroop), memory (Wechsler Memory Scale, CVLT, RBMT), language (Boston Naming, semantic fluency), executive function (D-KEFS, Wisconsin Card Sorting), and visuospatial function (Rey-Osterrieth, Judgement of Line Orientation).
The reports inform diagnosis, capacity decisions, return-to-work judgements, surgical planning (Wada testing for language laterality before epilepsy surgery), forensic determinations, and treatment monitoring. The discipline overlaps with neurology and neurosurgery; the boundary is professional rather than substantive.
The scope continues to grow. As the population ages and the dementias become the dominant clinical challenge, neuropsychological assessment is increasingly the gateway to diagnostic biomarker testing and to the new disease-modifying therapies. The clinic Brenda Milner walked into in 1950 and the clinic a doctoral student walks into in 2026 are recognisably the same place — older, larger, and busier.
↑ Bringing new life to "Patient H.M.," the man who couldn't make memories
Watch · Phineas Gage · The man with a hole in his brain
Watch · Oliver Sacks · The Man Who Mistook His Wife for a Hat
Sacks's The Man Who Mistook His Wife for a Hat, Luria's The Mind of a Mnemonist, and Damasio's Descartes' Error — in that order. Sacks teaches you to look at the patient. Luria teaches you what a single deep case can do. Damasio teaches you what the lesions are telling us about the unimpaired mind.
Neuropsychology has charted a remarkable amount in 175 years. The functional anatomy of language, memory, perception, and attention is mapped at a coarse grain that is now mostly stable. The dissociations have produced the modular taxonomy of cognition the field uses to teach. The patients of the founding era — Gage, Tan, H.M. — remain the standard reference cases.
What we know less well: how the modules emerge developmentally, how they reorganise after injury, and how their integration produces the unified experience that ordinary consciousness reports. The bridge from neural mechanism to phenomenal experience — the so-called hard problem — sits where it has always sat, awaiting a theoretical move the field has not yet made.
The discipline's working method has not changed. A careful clinician sits with a patient, asks the right questions, watches the answers, and reads off, from the gaps, the structure of a mind whose intactness is taken for granted everywhere else. The gaps are still where the discoveries are.
Neuropsychology — Volume XII, Deck 12 of The Deck Catalog. Set in IBM Plex Sans with Tiempos Text body. Off-white #f5f3ee, navy ink #1d2d4a, sage #5a8a6a and clay #a85838 accents.
Thirty-one leaves on the patients, lesions, and clinicians who built our picture of the human brain by watching what breaks.
↑ Vol. XII · Psychology · Deck 12