Questioning Capgras

New to the Phantom Self?  Read what this site is about, scan the Table of Contents,  or begin with the IntroductionContact the author – or comment on any post. 

CapgrasSomeone with the Capgras delusion has the stubbornly ingrained belief that a person or persons close to her—her spouse, her parents, her child—is someone else, a stranger, an imposter.

Although rare, Capgras is of interest because understanding it could shed light on the normal processes by which people recognize people. To decide that you are in the presence of your wife, not a stranger who bears an uncanny resemblance to your wife, is to make a personal identity judgement. We make them all the time, usually automatically and effortlessly with people we know well, sometimes with effort when we struggle to place a familiar face encountered in an unusual setting. An understanding of how beliefs about personal identity can go so spectacularly wrong in Capgras and related disorders may afford insight into the ordinary intuitions we rely on to recognize other people and ourselves.

Many Capgras cases are associated with mental illness. A paranoid interpretation of experience can develop into the Capgras delusion when the patient starts to see familiar caregivers—family members, doctors—as hostile agents with hidden agendas, masters of disguise intent on deception. If  those were the only examples of the Capgras delusion, it would be easy to dismiss as a psychological oddity. But Capgras has also followed traumatic brain injury in previously healthy patients, which strongly suggests an organic basis. There may be circuits in the brain specifically adapted for making personal identity judgements, damage to which can result in spectacular misidentifications. The delusion can also be drug-induced in healthy subjects.

Capgras-like delusions are not always about other people. Delusional misidentifications of pets and familiar places have been reported. Some subjects also misidentify themselves.

In this post, I will focus on three examples of the Capgras delusion. DS, aka “David”, described in a 1997 study by Ramachandran and Hirstein, was subsequently featured in a BBC documentary. Following a head injury resulting from a vehicle accident at the age of 22, David started to believe that his mother and father, who cared for him during his convalescence, were other people, strangers who resembled his parents. His delusion was not part of a paranoid pattern; David understood that their intentions towards him were benevolent, and he seemed to like them. Another young man, Clifford Beers, left a detailed autobiography of his bipolar disorder and eventual recovery. When institutionalized following a suicide attempt in 1900, he developed the delusion that his brother and other family members who visited were disguised agents of the police gathering evidence for a legal case against him. And a healthy 26-year old graduate student, whom I’ll call “Nicole,” experienced Capgras-like symptoms after being administered the drug ketamine in an experiment. Screened prior to the study, Nicole was found to have no neurological and psychiatric problems. When the effects wore off, she told the experimenter:

“every time you left the room, I thought another person dressed in your clothes was coming back into the room…it wasn’t scary, just another person dressed in your clothes, doing your job but the person was a little older in age and weighed more.” She also reported: “I’d look into the mirror and think that’s not me, I didn’t think the image staring back at me was myself…even the words I was saying were not words I would normally say…it just wasn’t me… it wasn’t my speech, my voice, my reactions…I felt like a different person because I would not have reacted in the way I usually would have.” [Corlett et al, 2010]

Capgras vs. Humpty Dumpty

About 2.5% of the human population have moderate to severe difficulty in recognizing faces, a condition known as prosopagnosia. To understand the Capgras delusion, it is important to know that it is not prosopagnosia.  Although both involve impaired recognition of people, prosopagnosia and Capgras are in some respects like mirror images of one another.

Although prosopagnosics see details as well as other people, they are unable to take in a face as a whole, perceive its uniqueness, and connect it with their memory of the person to whom it belongs. Their deficit can often be traced to the fusiform gyrus, a brain area specialized for holistic processing of visual imagery, believed to be especially well adapted for face recognition—a valuable skill in a social species. Prosopagnosics do not have delusions about the persons they fail to identify. They simply lack information. Since they cannot get it in the way most of us do—by a glance at the face—they rely on other identifying information, such as distinctive features (an extravagant moustache, a tattoo, a trademark ball cap), the person’s voice, or a hint from a companion. Once they have made the connection, by whatever means, their problem is over.

Famous prosopagnosics include Oliver Sacks, Jane Goodall, and Humpty Dumpty, as depicted in Through the Looking Glass. When Alice says “Good-bye, till we meet again”, he remarks grumpily that he wouldn’t know her if they did meet: “you’re so exactly like other people.”

“The face is what one goes by, generally,” Alice remarked in a thoughtful tone.

“That’s just what I complain of,” said Humpty Dumpty. “Your face is the same as everybody has—the two eyes, so—” (marking their places in the air with his thumb) “nose in the middle, mouth under. It’s always the same. Now if you had the two eyes on the same side of the nose, for instance—or the mouth at the top—that would be some help.”

Although both prosopagnosics and people with Capgras have trouble recognizing faces, it is quite a different kind of trouble. All a prosopagnosic needs is a specific identifying clue in order to connect the person in front of him with the mental model of one of his acquaintances. Someone with Capgras has already made the connection. David had no trouble recognizing his parents’ faces. He knew exactly who the strangers in his apartment looked like. Yet he could not be convinced they were his father and mother. Something was wrong with his experience, that made him think they were other people.

The Affective-Response Hypothesis

What is wrong with people with Capgras, on the prevailing neuroscientific view, is that they fail to experience the expected emotional response at the sight of a familiar person. This interpretation began with Joseph Capgras, the French psychiatrist who first diagnosed “l’illusion des sosies” (the illusion of doubles) in 1923. Capgras called it “the conclusion of an emotional judgement.” [Ramachandran and Hirstein, 1997, p 240]  Ramachandran (2011) explains Capgras as a disruption to the pathway connecting the fusiform gyrus to the amygdala, “which performs an emotional surveillance of the object or face and generates the appropriate emotional response.”  Because of damage to this pathway, David did not experience the emotional response he expected to have in the presence of his father and mother. This produced a profound sense of strangeness, from which he concluded they must be imposters.

At least two neural pathways, leading from visual centres to other parts of the brain, are involved in the ability to recognize faces. Following Bauer (1984), Ellis and Young (1990) identified one pathway for overt, or conscious, recognition, and another pathway they describe as resulting in “covert recognition (recognition at an unconscious level).” Prosopagnosics have damage to the overt pathway, whereas Capgras patients have an impairment in the covert pathway. The latter connects the visual cortex to the limbic centres responsible for emotional arousal, and…

…may carry some sort of affective tone. When patients find themselves in such a conflict (that is, receiving some information which indicates that the face in front of them belongs to X, but not receiving confirmation of this), they may adopt some sort of rationalization strategy in which the individual before them is deemed to be an impostor, a dummy, a robot, or whatever extant technology may suggest. [Ellis and Young 1990 p 235]

The two-pathway account is supported by skin conductance response (SCR) experiments on prosopagnosics and Capgras patients. When people are emotionally aroused, they perspire slightly on the skin of the hands, which can be measured as increased electrical conductivity. Normal people show higher SCR’s when shown photos of familiar faces, compared to the faces of strangers. Capgras patients like David do not; their SCR’s remain low for familiar and unfamiliar faces alike.[Ramachandran and Hirstein, 1998] Prosopagnosics, on the other hand, typically show a similar SCR pattern to normal people. Even when they cannot identify a face, their SCR’s will jump if it is someone familiar—a family member, friend, or famous celebrity.[Bauer, 1984] This ‘double dissociation’ between Capgras and prosopagnosia is evidence that face recognition involves two separate neural pathways, one of which results in overt recognition (the ability to retrieve explicit memories about a person) and the other in what might be called ‘emotional recognition.’

Although the covert channel does not provide prosopagnosics with enough information to correctly identify casual acquaintances, there is some evidence that it strengthens that ability in ordinary people. Capgras patients, who have no trouble overtly recognizing close family members (that is, knowing who they look like), are ‘significantly impaired’ in recognition tasks such as matching photographs of the same face taken from different viewpoints [Ellis and Young, 1990, p. 233].

The Representational Hypothesis

On the affective-response theory advanced by Hirstein and Ramachandran (1997), what triggers David’s delusion is lack of the emotional response he expects to have for his parents. His cognitive misidentification of them as strangers follows. But Hirstein later changed his views (2011), arguing that the cognitive misidentification comes first. On Hirstein’s new model, the brain’s representation of a person has two components, one containing information about the person’s appearance, the other representing his mind: beliefs, preferences, emotions, etc.  Although the Capgras patient recognizes the outward appearance of his father, his representation of his father’s mind has become damaged or inaccessible. Consequently, he sees his father as having the mind of a stranger, and decides he is an imposter. So convinced, he naturally suppresses the affection he would normally feel for his father.

Hirstein asks why so many Capgras patients, with different histories, have arrived at the same improbable conclusion that a loved one has been replaced by a lookalike. Other, more plausible hypotheses are consistent with lack of an expected emotional response. It would certainly be disturbing to suddenly not have the usual feelings for one’s wife. Most people would seek an explanation in psychology (‘We no longer share the same interests’) or events (“Maybe she’s cheating on me!’).

Hirstein argues that because the Capgras patient cannot summon up his model of the other person’s mind, his brain substitutes a default model, one used to represent the minds of strangers. This give him a vivid experience of an unfamiliar mind in a familiar-looking body, from which the imposter explanation naturally follows.

So we are left with the question which comes first, the cognitive misidentification or the incongruous emotional response? Does David feel, emotionally, towards his father as he would towards a stranger, and then conclude that this person, who looks like his father, has a stranger’s mind? Or does he ‘experience the mind of a stranger,’ and, as a result, adopt the emotional reserve he would have for someone unknown?

A problem for the representational hypothesis is that the Capgras delusion is not always limited to persons. A 70-year-old woman who developed Capgras towards her husband following a cerebral hemorrhage frequently expressed the desire to “go home” in the evening, although she was at home the whole time. [Thiel et al, 2013] Several other examples of  ‘Capgras for place’ have been reported in the clinical literature. It is easier to understand how lack of an expected emotional response could make your own home seem like a strange place than to explain it as a breakdown of your ability to represent minds. However fond people are of their homes, they do not normally regard them as sentient.

The Two-Factor Theory of Delusion

Hirstein challenges the affective-response theory to explain why an incongruent emotional response is elevated into a delusional belief about identity. If you felt unexpected emotions towards your mother one day, would you conclude that the woman in front of you, who looks exactly like your mother, and who evidently knows you as well as your mother does, was an imposter? Or would you decide (accurately) that your feelings for your mother had changed?

This counterfactual question is hard to answer, as those of us who have not experienced Capgras cannot claim to know what that is like. Nevertheless, the high improbability that someone looks just like your mother and is impersonating her creates a presumption that if you did reach that conclusion, you must have a rational impairment of some kind, a defect in the normal ability to evaluate your own beliefs and reject those that are extremely improbable.

A delusion is a false belief firmly held in the face of strong countervailing evidence. People with delusions typically support them by confabulating—offering explanations that strike normal listeners as incredible stories, made up on the spot to paper over the glaring holes in their account of reality. Patients with left-arm paralysis following a stroke of the right hemisphere sometimes deny their disability, a condition called anosognosia. They claim they can move both arms equally well. When asked by the doctor to lift their left arm, they make lame excuses like, “My arm is tired today,” or, “I don’t take orders from you, doctor!” They may even claim that they are lifting the arm.[Ramachandran and Blakesee, 1998, pp. 127-131] Capgras patients sometimes confabulate in support of their delusion—Clifford Beers developed an elaborate theory of the motives of the police agent he thought was impersonating his brother, and David thought his father’s look-alike was someone hired by his father to help him—but sometimes they admit that the impersonation is an improbable fact that they just can’t explain.

Hirstein argues that a second factor is needed to explain why the deluded person fails to correct his highly improbable beliefs. Prefrontal brain damage has frequently been found in delusory patients. Hirstein cites a 2005 study on anosognosia by Berti, which showed that:

…patients who denied paralysis differed from those with paralysis but no denial in that the denial patients had additional damage in the frontal portions of a large brain network involved in the planning of motor actions. These frontal areas are directly connected to the damaged inferior parietal areas. [Hirstein, 2009, p 238]

It may well be true that two independent factors are at work in Capgras, as well as other delusions: a primary factor causing an unusual experience that suggests a bizarre interpretation of reality, and a secondary, rational impairment that allows the subject to incorporate the unlikely interpretation into his belief system. The two-factor theory is consistent with both the affective-response and the representational theories of Capgras. Dependence on two separate factors could explain the rarity of the delusion. It also suggests that other people may be out there who have the unusual experience but not the rational deficit. If there are, studying them might shed light on whether the affective-response theory or the mind-representation theory of Capgras is closer to the truth.

Nicole, whose Capgras-like symptoms were induced by the drug ketamine, could be one of those. She reported thinking that the experimenter was someone else dressed in the same clothes, and that the person reflected in the mirror was not herself, but the evidence does not show that she believed those things. What’s clear is that she experienced a sense of unfamiliarity. The study authors suggest that the full-blown Capgras delusion would likely have taken hold had the dose been stronger or repeated.

Syndrome or Symptom?

The diversity of cases of Capras delusion raises the possibility that the search for a common explanation of them all may be in vain. In seeking to understand it, we should not rule out the possibility that there may not be a Capgras syndrome as such. If, instead, we regard the delusion as a symptom, we remain open to the discovery that it has more than one aetiology. If so, then both the affective-response and the representational accounts of Capgras could be true, albeit of different cases. One subject may develop the imposter delusion because his representation of the other person’s mind is damaged; another, because of an aberrant emotional reaction to the person.

Emotion as Perception

Emotions are not standardly classified with the senses—vision, hearing, touch, taste, smell, proprioception—that provide an animal with information about its environment and its body.  But in fact, our affective responses do give us information. If I’m hungry, it’s a safe bet that my blood sugar levels are low. Your feeling of mistrust towards someone may be giving you accurate information about that person. Although your feeling is undoubtedly rooted in visual and auditory clues, they may be too subtle for you to notice explicitly as signs of malevolence or deception. The feeling itself may be the only evidence available to you.

With the Gambling Game experiment (described more fully in an earlier post), Antonio Damasio demonstrated that an emotional response to a subtle threat can influence behaviour well in advance of any cognitive awareness of danger. Players were instructed to pick cards from four decks. Some cards gave the players rewards, while others imposed penalties. Two decks contained high reward cards that initially attracted players, but were ultimately stacked against them, whereas the other two decks were slightly biased in the players’ favour. Early in the game, normal players’ SCR’s began to spike when they reached for the ‘toxic’ decks. A little later, they began to avoid those decks, but when asked, could not explain why. Only when the cards were nearly exhausted did any players consciously ‘figure out’ the bias in the decks.

Detecting the statistical bias in a deck of cards is an inductive process. Induction is hard to do by conscious reasoning, as anyone who has played Robert Abbott’s card game Eleusis can testify. Conscious ‘figuring out’ relies on attention and short-term memory, which are too limited in capacity to do statistical analysis of dozens of independent facts. The Gambling Game experiment shows that our brains have that capacity nonetheless—statistical patterns are detected by processes that take place without conscious awareness, and that are first manifested in emotional responses. They change our behaviour before we have a clue what’s going on.

Our emotional responses to other people are a fount of information about them, inductive distillations of all we have experienced in their company. The experience of talking to someone is full of emotional reactions which influence our behaviour—how friendly or unfriendly we are, how far we are willing to trust. Emotions are not infallible sources of information (neither are any of the standard senses), but are nonetheless very useful. Members of a social species who have skewed or poorly developed emotional ‘feelers’ for others are at a disadvantage.

Think about how you respond emotionally to someone you know well—a lover, a sibling, your boss, or a co-worker. Your response shapes your behaviour towards that person. It sets up a complex set of dispositions that determine your body language, your tone of voice, and whether or not you kiss that person on the nape of the neck by way of greeting. It governs trust—what you will reveal to that person, what you will withhold, how you present yourself, what you pretend to be. Your emotional response embodies and reflects the history you share with that person. It expresses your relationship to that person, as individual and expressive as a portrait.

Seen in this light, the affective-response and representational theories of Capgras seem less far apart. It is difficult to distinguish one’s intricate emotional response to the presence of a family member from one’s beliefs about that person’s mind. Emotional responses are inseparable from behavioural dispositions, and those in turn are inextricably bound to a mental model of the person’s beliefs, attitudes and inclinations. My love for my wife is expressed in spontaneous affectionate behaviour towards her, in trust, in lack of inhibition. Those emotional and behavioural patterns would be shattered if my  beliefs about her mind changed—for example, if I stopped believing that she loved me in return. Although I might still love her, I could not express it in the same way. A profound alteration in my relationship towards her would inevitably follow. Certainly my emotional response, when she entered the room, would be different  from what it is now.

I might wake up one morning feeling emotionally flat towards my wife—and be perplexed about that—while still recognizing that her love for me had not changed.  If my feelings for my wife were to suddenly change, I would not be forced to change my beliefs about her mind. But if I stopped believing that she loved me, I could not maintain the same feelings.

This asymmetry seems to favour the representational hypothesis, and is, I suspect, at the heart of Hirstein’s intuition that the emotional-response hypothesis cannot explain Capgras. But it does not settle the question.

Capgras for Vision, Capgras for Voice

Another difficulty for the representational model is that the Capgras delusion is usually associated with a single sensory modality. David believed his mother was an imposter when he saw her, but he had no difficulty recognizing her voice on the telephone. On the phone, he treated her the way he used to treat his mother. His representation of his mother’s mind was apparently undamaged at the time. If his delusion was caused by damage to that representation, why would it be repaired during a telephone conversation? If it wasn’t actually damaged, could it have become inaccessible, for some reason, when David actually saw his mother? Hirstein does not suggest a mechanism which could explain how that happened.

The affective-response hypothesis offers an explanation of how the Capgras delusion could be specific to vision. Ramanchandran argues that David’s problem was caused by disruption to a neural pathway connecting the fusiform gyrus to the amygdala. On Ramachandran’s model, face recognition comes first, but its results are not communicated to the emotional centres of the brain, resulting in a bizarrely discordant experience.

But the Capgras delusion is not always associated with vision. Several cases of blind Capgras patients have been reported, and at least one sighted person had an apparently voice-specific Capgras delusion. Lewis and Sherwood  (2001) found that this subject showed a normal differentiation in SCR responses to familiar and unfamiliar faces, but not to voices, suggesting that two neural pathways, one connected to emotional centres, are also at work in our ability to recognize people by their voices.

The SCR evidence, the association of the delusion with a specific sensory modality, in combination with independent neurological evidence of separate pathways conveying sensory information to the emotional centres and other parts of the brain, all are consistent with the affective-response explanation of Capgras.

Still, there is another nagging question. David heard his mother’s voice when she was physically present, not just over the phone. If the pathway between David’s voice recognition areas and his emotional centres was intact, as the telephone evidence suggests, why wasn’t the familiar sound of her voice enough to summon up his usual emotional response when she was there in the room? If the problem was just that the connection between David’s face recognition areas and his emotional centres was damaged, why wouldn’t the intact pathway from his voice recognition areas to the emotional centres compensate for that? Why would he not recognize his mother by her voice?

A plausible, although unconfirmed, answer is that something may have blocked that recognition. We respond emotionally to strangers, as well as to our intimates. For many animals, and all social animals, the physical closeness of a member of the same species is of comparable importance to that of a predator or prey. A conspecific may be a potential rival for food or social position, or a potential mate, all matters that significantly affect an organism’s chances of survival and reproduction. When strangers are nearby, we are circumspect, more inhibited than if we were alone. In crowded elevators, we avoid one another’s eyes and watch the numbers or, in recent years, our phones. It is plausible that when David sees his mother but does not recognize her emotionally, the emotional response he would have towards an unknown person kicks in. If that response became associated with the sight of the woman in his apartment, it might block the normal response to his mother which the sound of her voice, alone, would arouse. As Ramachandran says, “The brain abhors discrepancies;” and it tends to resolve them in favour of the visual channel. The rich detail and high salience of visual information, and the disproportionate amount of neural processing devoted to it, are indicators that we humans rely more heavily on vision than on other senses for information about the world—a fact encapsulated in the aphorism, “Seeing is believing.”

Capgras and Human Replication

I had hoped that the Capgras delusion and related recognition failures would shed light on the processes by which normal people make personal identity judgements—recognize themselves, and others, as the same persons over time. But the gaps remaining in the scientific understanding make it too early to draw strong conclusions. There are still more questions than answers.

The evidence contrasting Capgras and prosopagnosia strongly suggests that recognition of persons has more than one component. One deals with the outward signs of identity; it consists in knowing who the person looks like, and sounds like. The other, which may be described as “emotional recognition,” determines how one feels and behaves towards the person. Normally, when people who know each other meet again, these two channels of recognition concur. Reinforcing one another, they re-establish the relationship between the two people.

These psychological facts augur well for the eventual public acceptance of human replication technology. Suppose that, a few decades hence, a family member you know and love develops a life-threatening illness. Technology has advanced to the point that the surest and cheapest ‘cure’ is digital reconstruction: a process which involves recording physical information about the person’s body and brain as a large electronic file, editing the file to remove information stemming from the disease, manufacturing a replica of the person from the modified file, and destroying the original, diseased body. The replication process is good enough to preserve all the person’s memories and other psychological and bodily attributes (as well as they are normally preserved from one day to the next), excepting only the undesired attributes of the disease. Suppose that your family member undergoes this ‘cure,’ and you find yourself sitting across the breakfast table from the survivor. You might adopt one of two attitudes towards him or her. Knowing that your loved one’s body was destroyed, and that the person across from you is the product of a copying process, you might decide that your loved one is dead, and this person is someone else—a ‘mere replica.’ You would experience grief, knowing that your loved one’s life was cut short. You would probably feel ambivalent, at best, towards this other person, who now addresses you familiarly, asking what’s wrong.

If you would, your attitude strikingly resembles David’s attitude towards his parents. It is Capgras-like. The main difference is that there is no delusional component. You have the facts straight.

The other attitude available to you is to decide that the person you are now looking at is your family member, who has, to your immense relief, been restored to health by this wonderful new treatment. Everything about the person is familiar, and dear to you. Your normal emotional response is aroused, and with it, the behavioural dispositions which define your entire relationship. This, of course, would be the natural response of a normal human being who was unencumbered by metaphysical notions that personal identity inheres in the continuity of some underlying substance, either physical (such as the body) or spiritual. All the channels of recognition, overt and covert, would lead to the same conclusion.


Bauer (1984) “Autonomic recognition of names and faces in prosopagnosia – a neuropsychological application of the Guilty Knowledge Test, prosopagnosia”

Beers, Clifford (1908) A Mind that Found Itself (5th ed. revised 1921)

Corlett et al (2010) “Capgras syndrome induced by Ketamine in a Healthy Subject,” Biol Psychiatry Jul 1, 2010, 68(1).

Ellis, HD and Young, AW (1990) “Accounting for delusional misidentifications,” in Young, AW (ed.) Face and Mind, 1998, Oxford University Press, Oxford.

Hirstein, W (2010) “The misidentification syndromes as mindreading disorders,” Cognitive Neuropsychiatry, 2010, 15.

Hirstein and Ramachandran (1997) “Capgras syndrome: a novel probe for understanding the neural representation of the identity and familiarity of persons,” Proceedings of the Royal Society of London, 264.

Lewis and Sherwood (2001) “Autonomic responses to familiar faces without autonomic responses to familiar voices: Evidence for voice-specific Capgras delusion,” Cognitive Neuropsychiatry, 2001, 6.

Ramachandran, VS (2010) The Tell-Tale Brain, W.W. Norton & Company, Inc., New York.

Ramchandran and Blakesee (1998) Phantoms in the Brain, HarperCollins, New York.

Thiel et al (2013) “When a loved one feels unfamiliar: A case study on the neural basis of Capgras delusion,” Cortex, vol 52, March 2014.


Return to the Phantom Self home page.




Tags: , , , ,

2 Responses to “Questioning Capgras”

  1. susan kahn says:

    Gordon, this article is a most interesting and comprehensive account of the Capgras predicament, and its vast implications for how being human works. Bravo and thank you!!

Leave a Reply