The Return of Phineas Gage: Clues About the Brain from the Skull of a Famous Patient

Hanna Damaslo, Thomas Grabowski, Randall Frank, Albert M. Galaburda, Antonio R. Damasio*

When the landmark patient Phineas Gage died in 1861, no autopsy was performed, but language, motor function, and perception, his skull was later recovered. The brain lesion that caused the profound personality and now Gage's case indicated something changes for which his case became famous has been presumed to have involved the left even more surprising: Perhaps there were frontal region, but questions have been raised about the involvement of other regions and structures in the human brain dedicated to about the exact placement of the lesion within the vast frontal territory. Measurements from the planning and execution of personally Gage's skull and modern neuroimaging techniques were used to reconstitute the accident and socially suitable behavior, to the aspect and determine the probable location of the lesion. The damage involved both left and right of reasoning known as rationality. prefrontal cortices in a pattern that, as confirmed by Gage's modern counterparts, causes Given the power of this insight, Har- a defect in rational decision making and the processing of emotion. low's observation should have made the scientific impact that the comparable sug- gestions based on the patients of Broca and Wernicke made (2). The suggestions, al- On 13 September 1848, Phineas P. Gage, profound change in personality were al- though surrounded by controversy, became a 25-year-old construction foreman for the ready evident during the convalescence un- the foundation for the understanding of the Rutland and Burlington Railroad in New der the care of his physician, John Harlow. neural basis of language and were pursued England, became a victim of a bizarre acci- But as the months passed it became appar- actively, while Harlow's report on Gage did dent. In order to lay new rail tracks across ent that the transformation was not only not inspire a search for the neural basis of Vermont, it was necessary to level the radical but difficult to comprehend. In some reasoning, decision-making, or social be- on April 14, 2014 uneven terrain by controlled blasting. respects, Gage was fully recovered. He re- havior. One factor likely to have contrib- Among other tasks, Gage was in charge of mained as able-bodied and appeared to be uted to the indifferent reception accorded the detonations, which involved drilling as intelligent as before the accident; he had Harlow's work was that the intellectual holes in the stone, partially filling the holes no impairment of movement or speech; atmosphere of the time made it somewhat with explosive powder, covering the powder new learning was intact, and neither mem- with sand, and using a fuse and a tamping ory nor intelligence in the conventional iron to trigger an explosion into the rock. sense had been affected. On the other A On the fateful day, a momentary distraction hand, he had become irreverent and capri- let Gage begin tamping directly over the cious. His respect for the social conventions www.sciencemag.org powder before his assistant had had a chance by which he once abided had vanished. His to cover it with sand. The result was a abundant profanity offended those around powerful explosion away from the rock and him. Perhaps most troubling, he had taken toward Gage. The fine-pointed, 3-cm-thick, leave of his sense of responsibility. He could 109-cm-long tamping iron was hurled, rock- not be trusted to honor his commitments. et-like, through his face, skull, brain, and His employers had deemed him "the most then into the sky. Gage was momentarily efficient and capable" man in their "em- stunned but regained full consciousness im- ploy" but now had to dismiss him. In the Downloaded from mediately thereafter. He was able to talk and words of his physician, "the equilibrium or even walk with the help of his men. The balance, so to speak, between his intellec- iron landed many yards away (1). tual faculty and animal propensities" had Phineas Gage not only survived the been destroyed. In the words of his friends momentous injury, in itself enough to earn and acquaintances, "Gage was no longer him a place in the annals of medicine, but Gage" (1). Gage began a new life of wan- he survived as a different man, and therein dering that ended a dozen years later, in lies the greater significance of this case. San Francisco, under the custody of his Gage had been a responsible, intelligent, family. Gage never returned to a fully in- and socially well-adapted individual, a fa- dependent existence, never again held a job vorite with peers and elders. He had made comparable to the one he once had. His progress and showed promise. The signs of a accident had made headlines but his death went unnoticed. No autopsy was obtained. H. Damasio and A. R. Damasio are in the Department Twenty years after the accident, John of Neurology, University of Iowa Hospitals & Clinics, Iowa City, IA 52242, and the Salk Institute for Biolog- Harlow, unaided by the tools of experimen- ical Research, San Diego, CA 92186-5800, USA. T. tal neuropsychology available today, per- Grabowski and R. Frank are in the Department of ceptively correlated Gage's cognitive and Neurology, University of Iowa Hospitals & Clinics, Iowa City, IA 52242, USA. A. M. Galaburda is in the Depart- behavioral changes with a presumed area of ment of Neurology, Harvard Medical School, Beth focal damage in the frontal region (1). Israel Hospital, Boston, MA 02215, USA. Other cases of neurological damage were Fig. 1. Photographs of (A) several views of the *To whom correspondence should be addressed. then revealing the brain's foundation for skull of Phineas Gage and (B) the skull x-ray. 1102 SCIENCE * VOL. 264 * 20 MAY 1994 more acceptable that there was a neural eas Gage was once again the protagonist of mensional reconstruction of a standard hu- basis for processes such as movement or a grim event. As a result, the skull and the man skull (7) so that its dimensions matched even language rather than for moral reason- tamping iron, alongside which Gage had those of Phineas Gage's skull. We also con- ing and social behavior (3). But the prin- been buried, have been part of the Warren structed Talairach's stereotactic space for cipal explanation must rest with the sub- Anatomical Medical Museum at Harvard both this skull and Phineas Gage's real skull stance of Harlow's report. Broca and Wer- University. (8). On the basis of the skull photographs, nicke had autopsy results, Harlow did not. As new cases of frontal damage were the dimensions of the entry and exit holes Unsupported by anatomical evidence, Har- described in this century, some ofwhich did were scaled and mapped into the deformed low's observation was the more easily dis- resemble that of Gage, and as the enigmas standard skull. Based on measurements of missed. Because the exact position of the of frontal lobe function continued to resist the iron rod and on the recorded descrip- lesion was not known, some critics could elucidation, Gage gradually acquired land- tions of the accident, we determined the claim that the damage actually involved mark status. Our own interest in the case range of likely trajectories of the rod. Final- Broca's so-called language "center," and grew out of the idea that Gage exemplified ly, we simulated those trajectories in three- perhaps would also have involved the near- a particular type of cognitive and behavior- dimensional space using Brainvox (9). We by "motor centers." And because the pa- al defect caused by damage to ventral and modeled the rod's trajectory as a straight line tient showed neither paralysis nor aphasia, medial sectors of prefrontal cortex, rather connecting the center of the entry hole at some critics reached the conclusion that than to the left dorsolateral sector as im- orbital level to the center of the exit hole. there were no specialized regions at all (4). plicit in the traditional view. It then oc- This line was then carried downward to the The British physiologist David Ferrier was a curred to us that some of the image process- level of the mandibular ramus. The skull rare dissenting voice. He thoughtfully ven- ing techniques now used to investigate anatomy allowed us to consider entry points tured, in 1878, that the lesion spared both Gage's counterparts could be used to test within a 1.5-cm radius of this point (20 motor and language centers, that it had this idea by going back in time, reconsti- points in all) (Fig. 2). damaged the left prefrontal cortex, and that tuting the accident, and determining the Possible exit points were determined as such damage probably explained Gage's be- probable placement of his lesion. The fol- follows: We decided to constrain the exit havioral defects, which he aptly described lowing is the result of our neuroanthropo- point to be at least 1.5 cm (half the diam- as a "mental degradation" (5). logical effort. eter of the rod) from the lateral and poste- Harlow only learned of Gage's death We began by having one of us (A.M.G.) rior margins of the area of bone loss (Fig. 3) about 5 years after its occurrence. He pro- photograph Gage's skull inside and out and because there were no disruptions of the ceeded to ask Gage's family to have the obtain a skull x-ray (Fig. 1) as well as a set of outer table of the calvarium in these direc- body exhumed so that the skull could be precise measurements (6) relative to bone tions (Fig. 1, lower right panel). However, recovered and kept as a medical record. landmarks. Using these measurements, we we accepted that the rod might have passed The strange request was granted, and Phin- proceeded to deform linearly the three-di- up to 1.5 cm anterior to the area ofbone loss because inspection of the bone in this region revealed that it must have been separated Fig. 2. View of the entry-level area completely from the rest of the calvarium 4ntod)r with the a priori most likely first c 1 (Fig. 1). Furthermore, the wound was de- Skull with this first trajectory. (A) .. scribed as an inverted funnel (1). We tested vector and the level (red) at which M -15 is 16 points within the rectangular-shaped exit entry points were marked. (B) View Ah -10 A 0 1110 a- a area that we constructed (Fig. 3). of a segment of section 1. On the L m 5 1 The trajectory connecting each of the ramus, and - left is the mandibular W- ~Jul entry and exit points was tested at multiple on the right is the array of entry a 0 A of the ar- anatomical levels. The three-dimensional points. (C) Enlargement a a ray of entry points. One additional 1 skull was resampled in planes perpendicular point was added (L20) to ensure to the best a priori trajectory (C in Figs. 2 was j several that every viable entry point 0 and 3). We were helped by impor- surrounded by nonviable points. pi P1,10a tant anatomical constraints. We knew that Nonviable vectors are shown in red the left mandible was intact; that the zygo- aq a &SO . p.s PL 15 and viable vectors with labels iden- p.wtoribr1 matic arch was mostly intact but had a tifying their exit points are shown in M~ chipped area, at its medial and superior green. Abbreviations: A, anterior; edge, that suggested the rod had grazed it; L, lateral; P, posterior; AM, anteromesial; AL, anterolateral; FPL, posterolateral; C, central. and that the last superior molar socket was intact although the tooth was missing. Ac- ceptable trajectories were those which, at Fig. 3. (A) View from above the A 11 each level, did not violate the following deformed skull with the exit hole 12 * 10 in rules: The vectors representing the trajecto- and the anterior bone flap traced ries could not be closer than 1.5 cm from the black. The blue circle represents _ _ 1 - the first vector tested, and the gray mid-thickness of the zygomatic arch, 1 cm surface represents the area where from the last superior molar, and 0.5 cm exit points were tested. (B) Sche- from the coronoid process of the mandible matic enlargement of the exit hole 5 (10). Only seven trajectories satisfied these and of the area tested for exit 3 conditions (Fig. 4). Two of those seven points. The letter C marks the first invariably hit the anterior horn of the lateral tested vector (blue). The numbers ventricle and were therefore rejected as an- 1 through 15 mark the other exit atomically improbable because they would points tested. Red indicates nonvi- with survival able vectors, green indicates via- not have been compatible (the ble vectors, and the label identifies the entry point. Note that the a priori best fit C was not viable. resulting massive infection would not have SCIENCE * VOL. 264 * 20 MAY 1994 1103 Fig. 4. (A) Front and lateral skull views with the projection of the five final vectors (V). The two red lines show the position of the two sections seen in (B). (B) Skull sections 2 and 3: examples of two bottleneck levels at which the viability of vectors was checked. Next to each section is an enlargement of the critical area. Abbreviations: T, missing tooth; M, intact mandible; Z, intact zygoma with a chipped area (light blue). been controllable in the preantibiotic era). When checked in our collection of normal brains, one of the remaining five trajectories behaved better than any other relative to the lower constraints and was thus chosen as the most likely trajectory. The final step was to model the five acceptable trajectories of the iron rod in a three-dimensional reconstruc- tion of a human brain that closely fit Phineas Gage's assumed brain dimensions (1 1). Ta- lairach's stereotactic warpings were used for this final step. The modeling yielded the results shown in Fig. 5. In the left hemisphere, the lesion involved the anterior half of the orbital fron- tal cortex (Brodmann's cytoarchitectonic fields 11 and 12), the polar and anterior mesial frontal cortices (fields 8 to 10 and 32), The frontal operculum, which contains Bro- rior and mesial orbital region (field 12), the and the anterior-most sector of the anterior ca's area and includes fields 44, 45, and 47, mesial and polar frontal cortices (fields 8 to 10 cingulate gyrus (field 24). However, the le- was also spared, both cortically and in the and 32), and the anterior segment of the sion did not involve the mesial aspect of field underlying white matter. In the right hemi- anterior cingulate gyrus (field 24). The SMA 6 Ithe supplementary motor area (SMA)I. sphere, the lesion involved part of the ante- was spared. The white matter core of the frontal lobes was more extensively damaged in the left hemisphere than in the right. There Fig. 5. Normal brain A B was no damage outside of the frontal lobes. fitted with the five pos- Even allowing for error and taking into sible rods. The best rod consideration that additional white matter is highlighted in solid 8. 7 damage likely occurred in the surround of white [except for (B), 61 .5 the iron's trajectory, we can conclude that where it is shown in 4, 3 the lesion did not involve Broca's area or the red]. The areas spared 21 by the iron are high- motor cortices and that it favored the ven- lighted in color: Broca, tromedial region of both frontal lobes while yellow; motor, red; soma- sparing the dorsolateral. Thus, Ferrier was tosensory, green; Wer- correct, and Gage fits a neuroanatomical nicke, blue. (A) Lateral pattern that we have identified to date in 12 view of the brain. Num- patients within a group of 28 individuals bered black lines corre- with frontal damage (12). Their ability to spond to levels of the make rational decisions in personal and so- brain section shown in cial matters is invariably compromised and (C). (D and E) Medical view of left and right so is their processing of emotion. On the hemispheres, respec- contrary, their ability to tackle the logic of tively, with the rod an abstract problem, to perform calcula- shown in white. tions, and to call up appropriate knowledge and attend to it remains intact. The estab- lishment of such a pattern has led to the hypothesis that emotion and its underlying neural machinery participate in decision making within the social domain and has raised the possibility that the participation D E depends on the ventromedial frontal region (13). This region is reciprocally connected with subcortical nuclei that control basic biological regulation, emotional processing, 1U;-y100 and social cognition and behavior, for in- stance, in amygdala and hypothalamus (14). Moreover, this region shows a high concen- 1104 SCIENCE * VOL. 264 * 20 MAY 1994 .,.§k... *awe tration of serotonin S2 receptors in monkeys posteriorly to this hole; and the dimensions of the Phineas Gage [mean length, 169.9 mm (SD, 4.1); entry hole at the level of the orbit. mean height, 113.6 (SD, 2.3); mean width, 125 whose behavior is socially adapted as well as 7. Thin-cut standard computerized tomography im- (SD, 3.9). The seven brains were fitted with the a low concentration in aggressive, socially age of a cadaver head obtained at North Carolina possible trajectories to determine which brain uncooperative animals (15). In contrast, Memorial Hospital. areas were involved. There were no significant 8. We introduced the following changes to the meth- differences in the areas of damage. The modeling structures in the dorsolateral region are in- od described by P. Fox, J. Perlmutter, and M. we present here was performed on subject volved in other domains of cognition con- Raichle [J. Comput. Assist. Tomogr. 9, 141 1 600LL (length, 169 mm; height, 1 15.2 mm; width, cerning extrapersonal space, objects, lan- (1985)]. We calculated the mean distance from 125.6 mm). the anterior commissure (AC) to the posterior 12. Data from the Lesion Registry of the University of guage, and arithmetic (16). These structures commissure (PC) in a group of 27 normal brains Iowa's Division of Cognitive Neuroscience as of are largely intact in Gage-like patients, thus and used that distance for Gage (26.0 mm). We 1993. accounting for the patients' normal perfor- also did not consider the AC-frontal pole and the 13. P. Eslinger and A. R. Damasio, Neurology 35, in neuropsychologic tests PC-occipital pole distances as equal because 1731 (1985); J. L. Saver and A. R. Damasio, mance traditional our group of normals had a mean difference of 5 Neuropsychologia 29, 1241 (1991); A. R. Dama- that are aimed at such domains. mm between the two measures, and Talairach sio, D. Tranel, H. Damasio, in Frontal Lobe Func- The assignment of frontal regions to himself did not give these two measurements as tion and Dysfunction, H. S. Levin, H. M. Eisen- different cognitive domains is compatible equal [J. Talairach and G. Szikla, Atlas dAnat- berg, A. L. Benton, Eds. (Oxford Univ. Press, New omie Stereotaxique du Telencephale (Masson, York, 1991), pp. 217-229; S. Dehaene and J. P. with the idea that frontal neurons in any of Paris, 1967); J. Talairach and P. Tournoux, Co- Changeux, Cereb. Cortex 1, 62 (1991). those regions may be involved with atten- Planar Stereotaxic Atlas of the Human Brain 14. P. S. Goldman-Rakic, in Handbook of Physiology; tion, working memory, and the categoriza- (Thieme, New York, 1988)]. We introduced an The Nervous System, F. Plum, Ed. (American anterior shift of 3% to the center of the AC-PC Physiological Society, Bethesda, MD, 1987), vol. tion of contingent relationships regardless line and used that point as the center of the 5, pp. 373-401; D. N. Pandya and E. H. Yeterian, of the domain (17). This assignment also AC-PC segment. This shift meant that the ante- in The Prefrontal Cortex: Its Structure, Function agrees with the idea that in non-brain- rior sector of Talairach's space was 47% of the and Pathology, H. B. M. Uylings, Ed. (Elsevier, total length and that the posterior was 53%. We Amsterdam, 1990); H. Barbas and D. N. Pandya, damaged individuals the separate frontal had no means of calculating the difference be- J. Comp. Neurol. 286, 253 (1989). regions are interconnected and act cooper- tween the right and left width of Gage's brain; 15. M. J. Raleigh and G. L. Brammer, Soc. Neurosci. atively to support reasoning and decision therefore, we assumed them to be equal. Abstr. 19, 592 (1993). 9. H. Damasio and R. Frank, Arch. Neurol. 49, 137 16. M. Petrides and B. Milner, Neuropsychologia 20, making. The mysteries of frontal lobe func- (1992). 249 (1982); J. M. Fuster, The Prefrontal Cortex tion are slowly being solved, and it is only 10. There were two reasons to allow the vector this (Raven, New York, ed. 2, 1989); M. I. Posner and fair to establish, on a more substantial close to the mandible: (i) The zygomatic arch and S. E. Petersen, Annu. Rev. Neurosci. 13, 25 and Harlow the coronoid process were never more than 2 cm (1990). footing, the roles that Gage apart; (ii) we assumed that, in reality, this distance 17. P. S. Goldman-Rakic, Sci. Am. 267, 1 10 (Septem- played in the solution. might have been larger if the mouth were open or ber 1992); A. Bechara, A. R. Damasio, H. Dama- if the mandible, a movable structure, had been sio, S. Anderson, Cognition 50, 7 (1994); A. R. REFERENCES AND NOTES pushed by the impact of the iron rod. Damasio, Descartes' Error: Emotion, Reason and 11. The final dimensions of Phineas Gage's Talairach the Human Brain (Putnam, New York, in press). (1868). space were as follows: total length, 171.6 mm; 18. We thank A. Paul of the Warren Anatomical Mu- 1. J. M. Harlow, Pub. Mass. Med. Soc. 2,327 skull. 2. P. Broca, Bull. Soc. Anthropol. 6, 337 (1865); C. total height, 111.1 mm; and total width, 126.5 mm. seum for giving us access to Gage's Sup- Wernicke, Der aphasische Symptomencomplex Comparing these dimensions to a group of 27 ported by National Institute of Neurological Dis- (Cohn und Weigert, Breslau, Poland, 1874). A normal subjects, we found that in seven cases at eases and Stroke grant P01 NS19632 and by the remarkable number of basic insights on the func- least two of the dimensions were close to those of Mathers Foundation. tional specialization of the human brain, from motor function to sensory perception and to spo- ken and written language, came from the descrip- tion of such cases mostly during the second half of the 19th century. The cases usually acted as a 92RESEARCH ARTICLE::::.... springboard for further research, but on occasion their significance was overlooked, as in the case of Gage. Another such example is the description of color perception impairment (achromatopsia) caused by a ventral occipital lesion, by D. Verrey Highly Conjugated, Acetylenyl [Arch. Ophthalmol. (Paris) 8, 289 (1888)]. His astonishing finding was first denied and then ignored until the 1970s. Bridged Porphyrins: New Models for 3. Reasoning and social behavior were deemed inextricable from ethics and religion and not ame- nable to biological explanation. Light-Harvesting Antenna Systems 4. The reaction against claims for brain specializa- tion was in fact a reaction against phrenological doctrines, the curious and often unacknowledged Victor S.-Y. Lin, Stephen G. DiMagno, Michael J. Therien* inspiration for many of the early case reports. The views of E. Dupuy exemplify the attitude [Examen A new class of porphyrin-based chromophore systems has been prepared from ethyne- de Quelques Points de la Physiologie du Cerveau meth- (Delahaye, Paris, 1873); M. MacMillan, Brain Cog- elaborated porphyrin synthons through the use of metal-mediated cross-coupling nit. 5, 67 (1986)]. odologies. These systems feature porphyrin chromophores wired together through single 5. D. Ferrier, Br. Med. J. 1, 399 (1878). ethynyl linkages. This type of topological connectivity affords exceptional electronic inter- 6. The first measurements were those necessary to actions between the chromophores which are manifest in their room temperature photo- construct Gage's Talairach stereotactic space in- and deform a three-dimensional, computerized physics, optical spectroscopy, and electrochemistry; these spectroscopic signatures tomography skull: the maximum length of the dicate that these species model many of the essential characteristics of biological light- skull, the maximum height of the skull above the harvesting antenna systems. inion-glabella line, the distance from this line to the floor of the middle fossa, the maximum width of the skull, and the position of the section contour of Gage's skull relative to the inion-glabella line. The second measurements were those necessary biomimetic chemistry of to construct the entry and exit areas: on the top The optical, electronic, and photophysical the developing external view, the measure of edges of the trian- properties of porphyrins have made these multichromophoric assemblies in biology as gular exit hole; on the internal view the distances molecules desirable targets for incorpo- well as for potential application in sensing from its three corners to the mid-sagittal line and and (1), opto-electronic (2), magnetic (3), ar- to the nasion; the distance from the borders of the ration into supramolecular systems hole to the fracture lines seen anteriorly and polymers due to their central importance in tificial photosynthetic (4, 5), catalytic (6), SCIENCE * VOL. 264 * 20 MAY 1994 1105