Abstract
In some patients with chronic physical complaints, detailed examination fails to reveal a well-recognized underlying disease process. In this situation, the physician may suspect a psychological cause. In this review, we critically evaluated the evidence for this causal claim, focusing on complaints presenting as neurological disorders. There were four main conclusions. First, patients with these complaints frequently exhibit psychopathology but not consistently more often than patients with a comparable “organic” diagnosis, so a causal role cannot be inferred. Second, these patients report a high incidence of adverse life experiences, but again, there is insufficient evidence to indicate a causal role for any particular type of experience. Third, although psychogenic illnesses are believed to be more responsive to psychological interventions than comparable “organic” illnesses, there is currently no evidence to support this claim. Finally, recent evidence suggests that biological and physical factors play a much greater causal role in these illnesses than previously believed. We conclude that there is currently little evidential support for psychogenic theories of illness in the neurological domain. In future research, researchers need to take a wider view concerning the etiology of these illnesses.
Keywords
In a significant proportion of patients with chronic physical complaints, detailed examination fails to reveal a well-recognized disease process. In this situation, the physician may suspect a psychological cause. Indeed, in neurological clinics alone, an estimated 9%−11% of patients receive a diagnosis suggesting a probable psychological cause (Carson et al., 2000; Lempert, Dieterich, Huppert, & Brandt, 1990), and an estimated 4%−22% of patients in seizure clinics are diagnosed with “psychogenic” seizures (Benbadis & Hauser, 2000). In this article, we examine whether such psychological causal explanations have evidential support, focusing on illnesses presenting as neurological disorders. We ask: What sort of evidence would be required to provide positive support for a psychogenic explanation of these illnesses? We see no reason why a lower standard of evidence should apply to a psychological than to a medical explanation. Clearly, psychological causation cannot be demonstrated by the mere absence of a confirmed disease diagnosis (argument from ignorance). Medical practitioners simply cannot assume that the current knowledge of disease and its markers is 100% perfect or that all complaints not otherwise accounted for must have a psychological origin. Furthermore, from a scientific perspective, psychogenic theories of illness are conceptually complex, requiring significant psychological dysfunction and often also specific learning experiences. These theories therefore demand particularly strong empirical justification.
For this review, we define a psychogenic explanation as one in which (a) the primary causal factors are formulated at a psychological level (e.g., distress, fears, beliefs, attributions, or expectations) and (b) addressing these factors could lead to major improvement or even complete recovery. We first consider several conceptual issues that are relevant to interpreting evidence on this topic. We then examine the features of two major types of neurological disorders commonly explained in psychogenic terms, and outline some prominent psychogenic accounts of these disorders. Finally, we evaluate the evidence supporting these theories. Although the article is structured as a narrative review, in key sections we apply a more systematic search method to identify studies meeting specific criteria.
Theoretical and Conceptual Issues
Psychogenic theories make the strong claim that psychological factors play a primary causal role in the development or maintenance of illness. Establishing causation—in the absence of experimental control—is challenging. It is not sufficient merely to demonstrate an association between an outcome and a hypothesized causal factor (e.g., illness and depression). It is also necessary to rule out a primary causal relationship in the opposite direction (i.e., that the illness caused the depression). There is also the possibility that some third, possibly unknown factor might be responsible for the association. In addition, one must distinguish between causation and modulation. For example, high temperatures can exacerbate the symptoms of multiple sclerosis (Guthrie & Nelson, 1995), but that does not mean it causes the illness. Similarly, anxiety, stress, and tiredness may precipitate epileptic seizures (Spector, Cull, & Goldstein, 2000), but we would never conclude that these are causes of epilepsy.
One argument often claimed in support of psychogenic illness models is that psychological distress commonly manifests itself as physical complaints. For example, the psychological phenomena that characterize anxiety disorders are often accompanied by muscle tension and gastrointestinal symptoms (Aldao, Mennin, Linardatos, & Fresco, 2010). Similarly, the sad feelings and negative thoughts that define depression are often accompanied by insomnia and fatigue (Beck, 1961; Ward, 2006). Consequently, severe psychological distress might very possibly lead to more chronic, debilitating physical symptoms.
However, there are problems with this reasoning. First, the causal relationship between these psychological and somatic phenomena remains unclear. For example, insomnia may actually increase a person’s risk of developing depression (Riemann & Voderholzer, 2003). A further problem is that only a small proportion of severely psychologically distressed people develop chronic, debilitating physical illnesses of the kind examined in this article. Conversely, many individuals with such illnesses do not exhibit any significant psychopathology. To explain these inconsistencies, additional psychological variables must be proposed.
Two such variables are conversion and somatization. Conversion refers to a process whereby psychological tension or conflict is “converted” into physical signs and symptoms (see Brown, 2005, for a review). Somatization refers to a marked tendency to experience or express psychological distress as bodily symptoms or to misinterpret normal bodily sensations as signs of illness (Barsky, 1992; Lipowski, 1988). This tendency may be heightened in those with anxiety, depression, or a strong introspective focus. It has been suggested that somatization may contribute to the development of chronic physical complaints (Brown, 2004). Another possible contributor is alexithymia: a person who cannot recognize the physiological sensations that normally accompany certain emotions may misinterpret them as signs of illness (Taylor, Bagby, & Parker, 1991).
However, these concepts are problematic. The status of somatization (and indeed conversion) as a psychological construct rests entirely on indirect inference from self-reported physical complaints. There are no measurable affective or cognitive components. The concept also is highly underspecified: no constraints have been placed on which specific physical complaints should—and should not—be considered indicative. It is commonly assumed that any physical complaint without a medical explanation must reflect somatization. Given our current incomplete understanding of disease and its markers, this assumption is difficult to justify.
“Psychogenic” Movement Disorders and Seizures: Presenting Features
Multiple physical complaints have been attributed to psychogenic causes (e.g., chronic pain, gastrointestinal disturbances, fatigue, and even sensory disturbances). In the neurological domain, names for such disorders include functional neurological (symptom) disorder, conversion disorder, hysteria, somatoform disorder, and somatic symptom disorder and more specific terms for particular presentations. Here we consider in detail two presentations that have been extensively studied: psychogenic movement disorders (PMDs) and psychogenic nonepileptic seizures (PNESs). Both are associated with a set of well-defined presenting features, and both can be readily compared with “organic” diseases with similar features.
PMDs may themselves take many forms, including limb weakness, hemiparesis, tremor, dystonia, jerks, Parkinsonism, or even complete paralysis. The primary criterion is that the presentation (including test results) should be inconsistent with any well-established neurological movement disorder. In addition, other “positive” diagnostic signs are usually sought; the most common are listed in Table 1. These signs provide reassurance that the diagnosis is not based solely on negative criteria. However, it is unclear whether they significantly impact on diagnostic outcome (see, e.g., Morgante et al., 2012). PMDs can be chronic and intractable. Follow-up studies published since 1985 have reported that anywhere from 20% to 62% of PMD patients were the same or worse at least 6 months later (for a review, see Gelauff, Stone, Edwards, & Carson, 2013).
Presenting Features Often Used to Distinguish Psychogenic Movement Disorders (PMDs) and Psychogenic Nonepileptic Seizures (PNESs) From Their Organic Counterparts
Note: The concluding statements about the usefulness of these features are drawn from an informal review of relevant studies where an appropriate organic control group was used and where the examiners were blinded to diagnosis (for subjective measures only). ES = epileptic seizure; EEG = electroencephalography.
Examiner blinding was not required for studies addressing these features because they do not rely on subjective judgement.
PNES are primarily defined by the absence of accompanying epileptiform discharges on electroencephalography (EEG) during the seizure (after alternative medical diagnoses have been ruled out). PNESs vary widely in their presentation: they may present as periods of unresponsiveness alone or may be accompanied by tremor or, in some cases, gross motor movements or even vocalizations (Hubsch et al., 2011; see also Gröppel, Kapitany, & Baumgartner, 2000). However, the mere absence of an EEG signature is not a definitive diagnosis in itself, as it is also absent in some types of epileptic seizures (e.g., orbitofrontal and mesial frontal lobe seizures; Saygi, Katz, Marks, & Spencer, 1992). Consequently, certain positive distinguishing signs are also sought, if possible, with continuous video EEG monitoring. The most common of these are listed in Table 1. However, in practice, positive features are generally used in a confirmatory way; their absence is rarely sufficient to disconfirm a psychogenic diagnosis.
PNESs can also be chronic and intractable. Studies conducted after 1985 have reported that from 48% to 65% of PNES patients continue to experience seizures 1 year or more after diagnosis (considering studies of 30 patients or more; see Bodde, Brooks, Baker, Boon, Hendriksen, Mulder, & Aldenkamp, 2009).
Psychogenic Theories of Illness
Various theories have been put forward to explain the symptoms observed in these disorders. Virtually all of these meet our working definition of psychogenic—that is, they propose that the primary causal (and/or maintaining) factors are psychological. Classic psychodynamic theories of these illnesses center on the concepts of dissociation and conversion (see, e.g., Nijenhuis, 2001; Prigatano, Stonnington, & Fisher, 2002; Sierra & Berrios, 1999; C. Spitzer, Spelsberg, Grabe, Mundt, & Freyberger, 1999). Here, the term dissociation refers to a process whereby distressing memories or thoughts are excluded from conscious awareness, enabling the person to avoid distress. However, these cognitions continue to create psychological tension at an unconscious level, which may be relieved through physical signs and symptoms (i.e., conversion). These can include movement abnormalities or, in cases of acute distress, even a seizure. Illness behavior may be further reinforced if the person gains benefits such as sympathy, attention, or special assistance (secondary gain).
More recent theories place less emphasis on trauma and more on personality and learning. Brown (2004) proposed that the central factor in psychogenic illnesses is a tendency to experience certain normal bodily sensations as unusually intense and to misinterpret them as signs of illness. This is most likely to occur if the sensations have personal salience—for example, the individual previously experienced them during a period of illness or distress or has thought about them or discussed them extensively. However, these complaints are unlikely to become severe and chronic unless the person also possesses certain psychological characteristics. These include a tendency to focus attention on internal states (similar to somatization) and a proneness to anxiety or depression (which may heighten any introspective tendencies and lead to rumination upon symptoms). According to Brown, PMDs may occur if a person develops the false expectation that his or her movements will be limited or abnormal in some way. This expectation may be incorporated into an action plan, resulting in abnormal movements. PNESs may occur when a single, aberrant action plan is activated as an acute response to certain stimuli or experiences. Having once been activated, the action plan then is more readily reactivated under similar conditions in the future.
Another, more recent account of PMDs also attributes them to distorted expectations about one’s movement capabilities (M. J. Edwards, Adams, Brown, Pareés, & Friston, 2012). These expectations may be created in any of the ways Brown (2004) suggested. However, the expectations only lead to persistent movement difficulties if there is great certainty attached to them—for example, if a person is focused strongly upon internal bodily functions or has developed intense health anxiety. In these circumstances, the individual works unconsciously to avoid disconfirmation of these expectations by adjusting his or her movements to fulfill them.
Concerning PNESs, contemporary theories identify a range of biological, psychological, and social factors that may play a causal role in seizure development or maintenance. These include past traumatic experiences, medical history, psychological characteristics, and prior learning (Bodde, Brooks, Baker, Boon, Hendriksen, & Aldenkamp, 2009; Bodde, Brooks, Baker, Boon, Hendriksen, Mulder, & Aldenkamp, 2009; Reuber, 2009). However, again, the individual’s psychological characteristics are crucial. Key predisposing characteristics include a proneness to somatization, a high level of anxiety or depression, difficulty regulating emotion, or a maladaptive coping style. Persons who have experienced epileptic seizures themselves or observed them in others are particularly vulnerable to PNESs, as they may “model” these previous experiences when acutely distressed (symptom modeling; see Aldenkamp & Mulder, 1997). Factors that further perpetuate the occurrence of PNESs include negative affective states, misattributions, avoidance of underlying psychological issues, and secondary gain (Reuber, 2009). These various accounts recognize that different factors may operate in different individuals. Indeed, Baslet (2011) suggested two alternative sets of predisposing factors. One is a proneness to somatization, combined with extreme emotionality and suggestibility. The other is a tendency toward dissociation, which when combined with poor emotional awareness and limited cognitive resources, may create a system more readily “tipped into crisis.”
Testing Predictions of Psychogenic Theories
In this review, we formulate a series of predictions based on the theories previously described and examined the existing evidence for each. We consider predictions relating to (a) psychological characteristics, (b) stressful or traumatic life experiences, (c) illness features consistent with a psychological origin (e.g., suggestibility), (d) secondary gain, and (e) the effectiveness of psychological interventions. In the final sections, we consider some other relevant sources of evidence, including whether psychogenic diagnoses are stable over time and whether there is evidence of accompanying biological abnormalities in these individuals.
The role of psychological characteristics
Virtually all theories we review predict a disproportionately high incidence of anxiety, depression, and somatization disorders in patients with PMDs and PNESs. Some also predict high levels of introversion or a proneness to dissociative experiences. Finally, several recent theories predict marked disturbances in emotional regulation or awareness, poor coping skills, or a high incidence of associated personality disorders (e.g., borderline personality disorder).
However, before such predictions can be tested, several important methodological issues must be considered. One is that psychopathology can often be a consequence, rather than a cause, of physical illness. Indeed, even organic illnesses are associated with very high rates of psychopathology (Katon, 2003). In one large U.S. cohort study, researchers estimated that a major (organic) disease diagnosis was associated with a 41% higher likelihood of a recent diagnosis of depression or anxiety (Cassem, 1995). Estimates are even higher for conditions such as epilepsy (Gaitatzis, Trimble, & Sander, 2004) and Parkinson’s disease (Nuti et al., 2004). Further, the risk of psychopathology increases in these diseases with the number and severity of the physical complaints (Kisely, Goldberg, & Simon, 1997; Kroenke et al., 1994). Some portion of these associations may be a measurement artifact. For example, the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013) criteria for depressive disorders include disturbances of sleep, energy, concentration, and appetite, all of which can be affected by physical illness. Self-report scales also frequently contain similar somatic items (e.g., the Beck Depression Inventory–II: Beck, Steer, & Brown, 1996). Such schemes may have a tendency to lead to overdiagnosis of psychiatric problems in those with physical illnesses. Consequently, to provide evidence concerning the causal role of psychopathology in illness, investigators need to compare the disorder of interest with one of known etiology with similar physical complaints—and ideally, these complaints should be of similar frequency and severity.
It also is important that the psychological or psychiatric assessment is made independently of the medical diagnosis. If the interviewer is aware the patient has “medically unexplained” symptoms, his or her judgments may be influenced in numerous subtle ways (e.g., a patient worried about unexplained symptoms may be rated as more anxious than one whose symptoms are medically explained).
We systematically searched the literature for studies that (a) included a control group with qualitatively similar symptoms due to organic disease and (b) had psychological measures that had been obtained independently of the medical diagnosis, either by a blinded interviewer or with an objectively scored self-report scale. We did not require the control group to be fully matched on symptom frequency and severity because to do so would have yielded only one qualifying study in each illness domain, but we did record this information. To ensure sufficient power to detect differences, we further required at least 30 participants in the group of interest. 1 Where there were at least three studies with comparable measures and samples, we also calculated average effect sizes (see Supplementary Materials for full details of method).
Results
In the domain of movement disorders, there were only the three qualifying studies (see Table 2). Rates of self-reported depression and anxiety were higher, on average, in PMD patients than in organic controls. Nevertheless, it is difficult to draw any firm conclusions from such a small number of studies. Also, in two studies, controls were patients with cervical or hand dystonia, which may be painful and disabling but perhaps not fully comparable in psychological and functional impact to tremor, myoclonus, or limb paresis.
Studies Meeting Selection Criteria That Examined the Psychological Characteristics of Patients Diagnosed With Psychogenic Movement Disorders (PMDs)
Note: The Brief Symptom Inventory is from Derogatis (2000); Beck Depression Inventory–II (BDI–II) is from Beck, Steer, and Brown (1996); Beck Anxiety Inventory (BAI) is from Beck, Epstein, Brown, and Steer (1988); Dissociative Experiences Scale (DES) is from Wiener (1992); Montgomery-Asberg Depression Rating Scale (MDRS) is from Montgomery and Asberg (1979); Toronto Alexithymia Scale (TAS-20) is from Bagby, Parker, and Taylor (1994). PMD = movement disorder diagnosed as psychogenic; OMD = movement disorder with confirmed organic basis.
In the domain of seizures, the outcomes of qualifying studies are summarized in Table 3 (descriptions of each study are presented in Tables S2–S4 in the supplementary online materials). In the two studies in which an interview format was used, one reported a reliably higher incidence of anxiety and personality disorders (specifically, borderline personality disorder) in PNES patients than in controls. No other reliable differences were observed. Of course, despite initial blinding, the participants’ medical diagnosis was likely to have been revealed during these interviews, so these assessments may not have been entirely free of interviewer bias.
Summary of Findings From Studies Meeting Selection Criteria Examining the Psychological Characteristics of Patients Diagnosed With Psychogenic Nonepileptic Seizures (PNESs)
Note: See Table S1 in the supplemental online materials for a detailed description of each study. The table also notes the number of studies where the PNES and control groups were matched for seizure frequency. Full references to the scales and instruments used are presented in Table S2–S4 in the supplementary online materials.
In the studies in which self-report measures were used, results for anxiety and depression were inconsistent. Some studies reported higher mean scores on self-report measures of these constructs, but these studies were outnumbered by those who reported null effects. In studies in which average effect sizes were calculable, the effects for both anxiety and depression were statistically reliable but small (see Table S5 for details). It is interesting to note that in studies where such data was available and reported, the largest group differences on scales of anxiety and depression were invariably observed on physiological–somatic subscale items. In the three studies in which introversion/extraversion was examined, no reliable group differences were observed.
The findings with respect to dissociative disorders or experiences were inconsistent, with half the studies reporting group differences and the other half reporting null effects. An average effect size, calculable from four studies examining dissociative experiences, was not statistically significant (see Table S5).
With respect to emotional awareness and regulation, a wide variety of constructs has been examined, but again, findings were inconsistent across studies, with null findings equaling or outnumbering reports of group differences. An average effect size was calculable from three studies examining alexithymia; it was not statistically significant (see Table S5).
Finally—in stark contrast to the findings discussed so far—mean scores on scales measuring somatization and related constructs were consistently higher for PNES patients than for controls. One caveat is that it is not always clear whether the evidence regarding somatization was gathered independently from the medical diagnosis. In some studies, complaints attributable to organic disease (e.g., epilepsy) were explicitly or implicitly excluded, so the measure of interest is no longer independent of the medical diagnosis.
It is worth noting that in those studies documenting this information, PNES patients reported a considerably higher frequency of seizures than epileptic controls (see Tables S2–S4). This confound is of some concern: in epilepsy, seizure frequency is a significant predictor of depression and anxiety (see, e.g., Cramer, Blum, Reed, & Fanning, 2003; Grabowska-Grzyb, Je˛drzejczak, Nagan´ska, & Fiszer, 2006; Kimiskidis et al., 2007).
Discussion
If psychopathology played a primary causal role in the development or maintenance of the (severe) disorders considered here, we would expect patients to score consistently and substantially higher on the predicted psychological measures than organic controls. Our review of the existing literature did not support this prediction. With the exception of somatization (which we discuss in more detail later), studies comparing the psychological characteristics of PMD or PNES patients with appropriate controls have not shown consistent group differences. Where average effect sizes were calculable, self-report measures of anxiety and depression were slightly higher for PNES patients than for controls, but the size of these effects was no greater than we would expect on the basis of seizure frequency alone.
The only psychological construct consistently associated with the illnesses of interest was somatization. Nevertheless, this finding is difficult to interpret. The measurement of somatization relies entirely on self-reported physical complaints, which is problematic when the groups being compared are not equated as to overall disease severity or disability. Also, in some cases, physical complaints attributable to organic disease were explicitly excluded, which rendered the group comparisons nonindependent. In addition, somatization itself is a problematic construct because it assumes that all medically unexplained complaints must have a psychological cause, which is a strong assumption given the incomplete understanding of disease and its markers. Until there is a way to define and operationalize somatization that is free from these problems, this evidence must be interpreted with extreme caution.
What is needed in the future to support claims about the causal role of psychopathology in these illnesses is (a) a theoretical framework that specifies exactly which characteristics should and should not play a causal role; (b) independent evidence concerning these characteristics in the relevant population, preferably gathered prior to illness onset; and (c) comparison against a control group with symptoms of a similar nature, frequency, and severity.
Stressful and traumatic life experiences
Many theories propose that traumatic or stressful experiences may predispose a person to psychogenic illness. However, testing such predictions is far from straightforward. One problem is that even those with confirmed organic disease report more adverse life events than healthy individuals, and this tendency may simply reflect a reporting bias (see Sharpe & Faye, 2006). Consequently, any study that does not include a control group matched for levels of physical disability is difficult to interpret. One must also control for gender, since reporting rates for physical and sexual abuse are generally higher in women (see Fiszman, Alves-Leon, Nunes, Isabella, & Figueira, 2004). Age and socioeconomic status also need to be considered, since both variables are likely to influence rates of abuse and other negative life experiences.
To identify studies in which at least some of these concerns were addressed, we performed a systematic literature search applying the same quality criteria set out previously but excluding studies in which the gender composition of the two groups differed widely (more than a 20-point difference in the percentage of females). We did not set any criteria with respect to age and socioeconomic status, but we have commented on this where relevant. The qualifying studies are summarized in Table 4.
Studies Meeting Selection Criteria in Which Researchers Examined Aspects of Negative Life Experiences in Patients Diagnosed With a Psychogenic Movement Disorder (PMDs) or Psychogenic Nonepileptic Seizures (PNESs)
Note: The Childhood Trauma Questionnaire (CTQ) is from Bernstein et al. (1994); Trauma Life Events Questionnaire is from Kubany et al. (2000); Parental Bonding Instrument (PBI) is from Parker, Tupling, and Brown (1979); Social Readjustment Scale is from Holmes and Rahe (1967); Structured Clinical Interview I and II based on DSM-IV is from First, Gibbon, Spitzer, Williams, and Benjamin (1997; First, Spitzer, Gibbon, & Williams, 1996); Family Environment Scale (FES) is from Moos (1974); Psychiatric Epidemiology Research Interview (PERI) Life Events Scale is from Dohrenwend, Krasnoff, Askenasy, and Dohrenwend (1978). ES = epileptic seizure.
Only one study of PMDs met our quality criteria. Kranick et al. (2011) examined 11 different childhood and adult life experience variables. There were three statistically significant group differences likely to survive correction for multiple comparisons: childhood emotional (but not sexual) abuse, childhood physical neglect, and overall incidence of lifetime traumatic events. However, it is unclear whether the control group used in this study (those with focal hand dystonia) was sufficiently well matched to the PMD group in terms of disability severity.
Five studies of PNESs met our quality criteria. Two interview-based studies showed a higher incidence of sexual abuse in PNES patients than in epileptic controls. One study focused on childhood events (Betts & Boden, 1992), and the other examined lifetime history (Scévola et al., 2013). It is unclear whether the former study used blinded interviewers; in the latter, interviewers were initially blinded to the medical diagnosis but may not have remained so throughout the interview.
Three studies used written self-report measures. One was difficult to interpret because the PNES patients were considerably older than the controls, and age was found to be an independent predictor of the key measures (van Merode et al., 2004). In the remaining two studies, one examined 13 life childhood and adult experience variables (Salmon, Al-Marzooqi, Baker, & Reilly, 2003). There were three group differences likely to survive correction for multiple comparisons: adult sexual abuse, childhood psychological (not sexual) abuse, and control within the family. However, once intercorrelations between these variables were taken into account, the authors identified only childhood psychological abuse as being uniquely predictive of PNESs. In the final study, researchers examined recent stressful events in various domains (e.g., school, work, love, family, legal matters, and health) and the distress they caused (Testa, Krauss, Lesser, & Brandt, 2012). No reliable group differences were found for any of the key measures.
In conclusion, the existing evidence that adverse life events play a causal role in suspected psychogenic neurological illness is weak (see also Sharpe & Faye, 2006, for a similar conclusion). First, the number of well-controlled studies in this domain is small. Second, investigators in these studies measured a wide range of different variables and did not always agree on which aspects of life history are most crucial (e.g., sexual vs. psychological abuse). The problem here is theoretical as well as empirical, since few psychogenic theories specify which types of adverse experiences should and should not be considered critical. Third, many of the variables measured are strongly intercorrelated. For example, a childhood history of sexual abuse is strongly associated with general family dysfunction, and both are strongly associated with low socioeconomic status (Rind, Tromovitch, & Bauserman, 1998). Consequently, even consistent reliable differences in personal history may be driven by factors other than those being directly measured.
An even bigger problem is the reliance on retrospective self-report, which may be subject to recall or reporting bias. There is evidence that the recollection of past experiences, especially negative ones, is strongly influenced by one’s current state of mind (Merckelbach & Muris, 2001). In addition, many patients with psychogenic diagnoses have been told their symptoms are due to unresolved psychological problems, perhaps relating to past experiences. Consequently, they are especially likely to search for attributions from within their own life history (see Briere, 1992; Pope & Hudson, 1995). What is needed to demonstrate a causal role for adverse events is (a) a model specifying the precise nature of the adverse events that do—and do not—play a causal role in the development of the illness and (b) prospective cohort data, where measures relating to predefined adverse life events can be obtained for the period before the onset of illness (as well as demographic, medical, and psychiatric information).
Presenting features suggestive of psychological causation
Some of the positive presenting features associated with PMDs and PNESs might themselves speak to illness etiology. Table 1 summarizes the results of an informal review of key studies of these features, where an appropriate organic control group was included, and where examiners were blinded to the medical diagnosis.
Two features were found to be highly specific to PMDs, both of which pertain to cases of limb weakness: Hoover’s sign and collapsing weakness. In Hoover’s sign, the bad leg appears weak when the patient attempts to move it but often is much stronger when it supports other types of movements. In collapsing weakness, the limb seems to give away suddenly upon light pressure. Both features suggest psychogenic limb paresis may be more amenable to intentional or attentional factors than is the case for organic movement disorders. Nevertheless, one must bear in mind that some organic disorders can also be modulated by intentional or attentional factors, so this is not definitive evidence of psychogenic causation (e.g., motor neglect: Buxbaum et al., 2004; Parkinson’s disease: Landers, Wulf, Wallmann, & Guadagnoli, 2005). Other positive signs in PMDs offer little insight, either because they are so rare that they do not speak to psychogenic illness as a whole (e.g., dragging gait) or because they also occur with reasonable frequency in organic illness and therefore cannot be considered reliable indicators of psychogenic etiology (e.g., distractibility, suggestibility, and entrainment).
Turning to seizures, our review revealed two features that are highly specific to PNESs. First, patients are often able to recall elements of the episode afterward, which suggests that their awareness may be less severely compromised than is usual during an epileptic event. This feature certainly suggests the operation of a distinctly different neural mechanism in PNESs, but its relevance to the issue of psychological causation is unclear. Second, PNESs rarely occur during sleep, which suggests they may be under a higher degree of voluntary control than epileptic seizures. Again, this feature could be attributable to a number of factors (e.g., physiological differences between sleep and wakefulness and their associated brain states, the presence or absence of external stimuli, and the triggering role of affect and cognition), so its interpretation is less than straightforward.
A third feature highly specific to PNESs is that they may be more readily induced through suggestion. In at least four studies with an appropriate control group, researchers found that PNESs are more likely to be induced through the use of a nocebo, or through the patient’s own mental efforts, than epileptic seizures. However, in these studies, investigators did not control for baseline seizure rate, which is often considerably higher in PNESs (see Tables S2–S4). Of course, the more frequently a person experiences seizures, the higher the probability of exhibiting one under any conditions. Again, although these findings suggest PNESs may be qualitatively different, they do not offer strong support for a psychogenic causal explanation, since psychological factors such as anxiety and stress can also precipitate epileptic seizures (Spector et al., 2000). There is a risk here of confusing causation with moderation.
Other proposed differences between PNESs and epileptic seizures are relative rather than absolute, so they do not provide reliable evidence as to causation (e.g., eye closure during seizure, awareness during event, and frequency of self-injury or incontinence) or are not statistically reliable (e.g., length of seizure and incidence of bizarre movements or emotive vocalizations during seizure).
In conclusion, some presenting features of PMDs and PNESs appear to be useful for distinguishing them from organic disorders. PNESs, in particular, appear to be qualitatively different events from typical epileptic seizures. Therefore, from a diagnostic point of view, it makes sense to document such features. However, the features themselves do not make a strong case for a psychogenic explanation. In many instances, the link between the relevant feature and the suspected cause is not made clear or is open to alternative interpretations. What is needed to make stronger claims is a theory that specifies the precise mechanism that gives rise to the feature in question.
Presence of secondary gain
A core concept in many psychogenic theories is that of secondary gain—the ancillary benefits obtained by assuming the sick role. The person’s symptoms may allow him or her to gain sympathy, attention, or assistance or to avoid unpleasant situations or stress. If secondary gain can be demonstrated, it may provide positive support for a psychogenic explanation.
Secondary gain is not easy to measure. It is usually measured by identifying changes associated with the sick role that might be potentially reinforcing, such as sympathy, attention, or avoidance of work (but, of course, whether these changes are actually reinforcing for any particular individual cannot be known). However, the incidence of secondary gains is likely to be related to the person’s overall level of physical disability, so a carefully chosen comparison group—one composed of individuals well matched for the level of incapacity—is essential. Studies compared secondary gain in suspected psychogenic illness with those in well-matched groups with neurologically confirmed disease have failed to show differences in their incidence (Chabrol, Peresson, & Clanet, 1995; Gould, Miller, Goldberg, & Benson, 1986). A further problem is that secondary gains often are evaluated without taking into account the associated losses, which may be wide and devastating (e.g., loss of financial and personal independence, loss of social contacts and leisure pursuits, and so on). Clearly, gains must considerably outweigh losses if they are to be motivating for the patient.
Finally, unless a set of objective criteria is applied independently of the medical diagnosis, reporting can be subject to bias. To illustrate, a recent case report described a young woman diagnosed with conversion disorder whose physical symptoms were so debilitating that she required residential care (Solvason, Harris, Zeifert, Flores, & Hayward, 2002). In the original report, several secondary gains were identified: the woman’s sickness role gave her the chance to rebuild her relationship with her estranged family and ex-lover and also provided relief from the stress of working. The woman died shortly thereafter and was found on autopsy to have Creutzfeldt–Jakob disease, a rare degenerative neurological disorder. This case illustrates how readily secondary gains can be found—even in severe organic illness—and the dangers of seeking them after a psychogenic diagnosis has been made.
Effect of psychological interventions
According to interventionist theories of causation (Woodward, 2007), an illness in which psychological factors play a primary causal role should respond substantially to an intervention targeting those factors, such as psychotherapy. Several studies of PMD patients have reported reductions in physical symptoms following various forms of psychotherapy, including combined psychodynamic psychotherapy and psychopharmacological treatment (Hinson, Weinstein, Bernard, Leurgans, & Goetz, 2006) and hypnosis (Moene, Spinhoven, Hoogduin, & van Dyck, 2002). In those with PNESs, reductions have been reported following psychodynamic psychotherapy (Hinson et al., 2006; Mayor, Howlett, Grünewald, & Reuber, 2010), general psychotherapy (Brooks, Goodfellow, Bodde, Aldenkamp, & Baker, 2007), and cognitive behavioral therapy (CBT; Goldstein, Deale, O’Malley, Toone, & Mellers, 2004; LaFrance et al., 2009).
However, few of these studies were randomized clinical trials, so they might very possibly represent overestimates of the effectiveness of these interventions (see Staines & Cleland, 2007, for discussion). Also, very few considered the possible impact of dropouts, which can be very high in this domain (up to 30% or 40%; see, e.g., Chen et al., 2014; Kompoliti, Wilson, Stebbins, Bernard, & Hinson, 2014) and are very likely to be nonrandom (for discussion, see Jüni, Altman, & Egger, 2001; Montori & Guyatt, 2001).
Therefore, to gain a more accurate picture of the effectiveness of psychotherapy on the physical symptoms of PNES and PMDs, we systematically searched the literature for studies that met the following criteria: (a) random allocation of participants to a treatment or control condition (other than waitlist); (b) use of intention-to-treat analysis, in which dropouts were counted as treatment failures; and (c) at least 15 patients in the treatment arm (although we also noted any qualifying studies with lower sample sizes). One study met these criteria; its participants were PNES patients. Goldstein et al. (2010) allocated 66 PNES patients to either standard medical care alone or accompanied by CBT. The CBT group showed a reliably larger reduction in seizure frequency than the control group at end of treatment (4 months) and at 6-month follow-up. One other study met all our criteria except for sample size. In a pilot study, LaFrance et al. (2014) used a two-way design to examine the effect of CBT or treatment with sertraline on seizure frequency in PNES patients. However, due to the small sample size (no more than 10 patients per treatment arm), between-group comparisons were not performed.
We conclude that evidence for the effectiveness of psychotherapy in these disorders is weak. This conclusion is in line with recent systematic reviews (PNESs: Martlew, Pulman, & Marson, 2014; conversion disorder: Ruddy & House, 2005).
The reader should also bear in mind that psychotherapy, particularly CBT, has demonstrated broad effectiveness in reducing physical symptoms even in organic diseases, including epilepsy (for a review, see Tang, Michaelis, & Kwan, 2014), and at least one such study meets the minimum quality requirements we set out (McLaughlin & McFarland, 2011). Few would assume that such interventions modify the underlying causes of epilepsy. More likely, they operate via a moderator variable. Therefore, in order for researchers to distinguish causation from mere modulation of symptoms, a treatment intervention would need to induce greater symptomatic improvement in suspected psychogenic cases than in organic controls. To our knowledge, no studies to date have examined this question.
Stability of psychogenic diagnoses over time
It has been suggested that some presentations diagnosed as psychogenic may actually represent the early subclinical manifestations of organic disease (E. T. Slater & Glithero, 1965). Indeed, in studies conducted prior to 1970, researchers reported that between 10% and 30% of neurological patients receiving a psychogenic diagnosis were diagnosed later with an organic disease that could have explained their earlier symptoms (see Stone et al., 2005, and references therein). However, since then, these rates have been steadily decreasing. For studies of psychogenic neurological disorders in general, where the diagnosis was made after 1985 and where a formal re-examination was conducted at least 2 years later, the reported rates of diagnostic change ranged from 5% to 12% (e.g., Crimlisk et al., 1998; Moene et al., 2000; see Stone et al., 2005). Considering seizures specifically, one study reported a 14% change rate (those where a sole PNES diagnosis was changed to definite or probable epilepsy; Lempert & Schmidt, 1990).
These rates are an improvement over those in earlier reports. However, there is a limit to what can be inferred from an unchanged diagnosis alone. An organic diagnosis generally requires a higher standard of supporting evidence than a psychogenic one, and for a patient to fulfill the requirements at follow-up, significant new evidence is needed. A failure to find such evidence could occur for many reasons, some of which have little to do with the accuracy of the original diagnosis—for example, the number and specificity of new presenting symptoms or the ability of new diagnostic methods to reveal abnormalities that might have been missed previously. Consequently, current rates of diagnostic change do not pose a serious challenge to psychogenic theories, but neither do they provide strong positive support for these theories.
Biological abnormalities and physical precipitating events
Although psychogenic theories do not make specific predictions regarding the medical history of patients with a psychogenic diagnosis, it is clearly worth considering this issue. Table 5 summarizes the results of the key studies reporting on these variables in PMDs and PNESs. Many PMD patients have an accompanying organic movement disorder (25% according to one study; Factor, Podskalny, & Molho, 1995), and more than a third have a significant neurological history. An even larger number report a physical injury or illness prior to PMD onset (up to 80% in some studies).
Summary of Results of Several Key Studies of the Precipitating Role of Illness or Injury in Psychogenic Movement Disorders (PMDs) or Psychogenic Nonepileptic Seizures (PNESs) and the Incidence of Concomitant Organic Abnormalities
Note: ES = epileptic seizure; EEG = electroencephalography; MRIs = magnetic resonance images.
Similarly, many PNES patients also suffer from epileptic seizures. Exact estimates range from 5% to 30%, depending upon whether the diagnosis required actual observation of an epileptic seizure or merely a validated history of epilepsy. Curiously, in one study of patients with both epilepsy and PNES, frontal focal epileptic seizures were unusually common, affecting around 25% of such patients compared with only 4% of those with a diagnosis of epilepsy alone (Pillai & Haut, 2012). PNES patients also commonly exhibit abnormalities on structural magnetic resonance images. Finally, as with PMD, the onset of PNESs is often preceded by an event such as a head injury—or even surgery. It is interesting that such events are also risk factors for epilepsy (Foy, Copeland, & Shaw, 1981). In fact, in one study, investigators who retrospectively examined individuals experiencing seizure-like events after a head injury found that approximately two thirds developed epileptic seizures, and the remaining third developed PNESs (Hudak et al., 2004).
Many of these studies relied on retrospective self-reports, which are subject to the recall biases outlined earlier. Further, most did not include control groups, so it is difficult to assess baseline rates of occurrence. There is a need for further, more systematic investigation of these variables. However, the evidence so far does suggest that biological factors are likely to play a significant role in the development of psychogenic neurological disorders.
Such findings have been accommodated within psychogenic frameworks in various ways. It has been suggested that illness or injury may itself be psychologically traumatic, leading to feelings of helplessness, perhaps triggering memories of earlier trauma (see Stone et al., 2009, for discussion). Alternatively, previously experienced illnesses may shape the manner in which patients express current psychological distress (symptom modeling: Bautista, Gonzales-Salazar, & Ochoa, 2008). A third suggestion is that an earlier illness or injury might increase a person’s general concern about bodily sensations (see, e.g., Brown, 2004), or the illness might shape a person’s expectations about his or her present capabilities (Pareés et al., 2013).
Other findings are difficult to explain within a psychogenic framework. One example is the development of PNESs after surgery. Glosser, Roberts, and Glosser (1999) suggested that the psychological distress patients experience at this time (a postsurgical period of “psychiatric instability,” p. 1750) may be expressed by modeling previous epileptic episodes. In a similar vein, Pillai, Haut, and Masur (2015) suggested that abnormalities involving the orbitofrontal cortex may compromise emotional or self-monitoring functions, which when combined with trauma or other stressors, may increase a person’s susceptibility to psychogenic disorders. Both accounts invoke different levels of explanation in service of a single phenomenon: a neurological causal factor is used to infer a psychological-level vulnerability, which in turn is used to explain symptoms of an apparently neurological nature. However, such evidence could quite reasonably lead one to question whether simpler accounts might be possible, in which psychological mediators play no role.
General Discussion
In this review, we sought to evaluate the evidence supporting psychogenic theories of illness in the neurological domain, applying the same evidence standards as one would demand for any other causal explanation. Our main conclusions were as follows. First, the incidence of psychopathology is not consistently higher in these illnesses than in comparable organic illnesses. The only psychological characteristic found to be associated consistently with a psychogenic diagnosis was somatization, but there are major problems interpreting this association (i.e., the nonindependence of the measures being examined and the untestable assumptions underlying the measurement of somatization). Second, researchers in a small number of studies have found that patients with a psychogenic diagnosis report more adverse life experiences than organic controls, but they disagree as to which types of experiences are implicated. Also, reporting biases may operate to exaggerate any group effects, and there are problems teasing apart intercorrelated variables. Third, although some of the presenting features of these illnesses differ from those of comparable organic cases, these differences are usually relative rather than absolute, which limits their value as support for a psychogenic explanation. Fourth, researchers examining whether PNESs are more inducible than epileptic seizures may fail to control for differences in baseline seizure rate. They also run the risk of confusing modulation and causation. Finally, it is not yet known whether suspected psychogenic illnesses are more responsive to psychological interventions than comparable organic illnesses.
Therefore, the overall conclusion from this review is that there is insufficient evidence to support the view that psychological factors are the primary causal or maintaining factors in these disorders. Further, if we take into account the effects of publication bias, which would likely operate to exaggerate any apparent support for psychogenic accounts, the “reality” may be even less favorable to these accounts than this review indicate. This is not to say that psychogenic explanations of these illnesses cannot possibly be true but simply that the evidence is current lacking. Perhaps the psychological line of inquiry might eventually bear fruit if one searches long and hard enough. However, the time may have come to focus on lines of inquiry that could be more beneficial to patients.
Our conclusion is not entirely at odds with historical precedent. In medicine, there is an ever-lengthening list of physical illnesses formerly attributed to psychological causes that are no longer believed to be primarily psychogenic. These include gastric ulcers (see e.g., Alp & Grant, 1970), multiple sclerosis (Caplan & Nadelson, 1980; Langworthy, 1948), ulcerative colitis (McKegney, Gordon, & Levine, 1970), rheumatoid arthritis (Moos, 1964); systemic lupus erythematosus (McClary, Meyer, & Weitzman, 1955), and, most recently, postconcussive syndrome (King, 2003 vs. Gosselin et al., 2011; Shenton et al., 2012). As new methodologies and diagnostic techniques are developed, other illnesses are likely to be added to the list. Nevertheless, our conclusion is clearly at odds with the prevailing view in literature within this domain, where one commonly sees strong statements such as “Psychogenic nonepileptic seizures (PNES) . . . have a psychological origin” (Brown et al., 2011, p. 85) and “Psychogenic movement disorders (PMDs) are movement disorders that result from a psychological or psychiatric rather than neurological disturbance” (Gupta & Laing, 2009: p. 430).
There are a number of possible reasons for our different conclusions. First, previous research has been founded on the premise that these illnesses are indeed psychogenic in origin. Researchers have not sought to scrutinize that premise. In contrast, we made no prior assumptions but instead carefully examined the supporting evidence from a fresh perspective. Second, in key topic areas, we identified studies using a more systematic method than is adopted in narrative reviews. Our method may have provided some protection against citation bias, the tendency for studies confirming a prevailing viewpoint to be cited more often than those disconfirming it (Greenberg, 2009; Jannot, Agoritsas, Gayet-Ageron, & Perneger, 2013). Citation bias can become amplified over time and through the mediating effect of influential narrative reviews, creating a distorted picture of the evidence (see, especially, Greenberg, 2009).
Third, we adopted a conservative approach to the evaluation of the evidence. In each topic area, we first identified potential problems in study design that might compromise the conclusions. We then located those studies that best addressed these problems. Finally, we critically examined any remaining difficulties that might impact on interpretation.
Some might object that our conservative approach has “thrown the baby out with the bathwater.” Perhaps many studies reporting genuine evidence were rejected as flawed. Perhaps behavioral researchers need to adopt more relaxed standards than other fields, since behavior is so complex, noisy, and difficult to measure. Perhaps a psychological explanation is harmless and offers some comfort to the patient, irrespective of its accuracy. Our response to these arguments is twofold. First, the sheer noisiness of behavioral data should not be a justification for applying lower standards of evidence but rather a reason to adopt even greater care. Recently, there has been growing concern within psychology that many of its key findings may not be secure (see, Ferguson & Heene, 2012; Ioannidis, 2012; Simmons, Nelson, & Simonsohn, 2011). This has led to a call for generally tighter standards of evidence.
Second, psychogenic explanations are far from harmless. At a population level, if practitioners falsely conclude that an illness is due to psychological factors, research avenues may be pursued that are ultimately unfruitful, most likely at the expense of more valuable ones—at a real cost in terms of future patients’ treatment opportunities (see Davey-Smith, 2005, for discussion). The patients’ own self-reports about their symptoms and their precipitators, which may be a valuable source of information about their illness, may be disregarded, thereby further delaying progress.
There may also be serious risks for the individual patient. From the moment a psychogenic explanation is offered, the patient’s own perceptions about his or her symptoms are recast as dysfunctional and become targets for “correction” by those claiming greater insight. A psychogenic interpretation also carries implications about the patient’s own responsibility for the continuing illness. Behavioral therapies may reinforce that message, and a failure to respond to therapy may sometimes even be blamed on the patient (“The patient’s own psychopathology interferes with treatment”: Harden, Burgut, & Kanner, 2003, p. 454). If these illnesses turn out not to be of psychological origin, such approaches are clearly inappropriate and potentially highly damaging to the individual. Indeed, frequent themes in the narratives of those diagnosed with a psychogenic illness are (a) a feeling that their own reports of their symptoms have not been sufficiently heard and considered (e.g., Barnum, 2014; Karterud, Knizek, & Nakken, 2010), (b) distress or anger at the suggestion they are psychologically unstable (Fairclough, Fox, Mercer, Reuber, & Brown, 2014; Karterud et al., 2010), and (c) feelings of self-blame for being unable to control their illness (Nettleton et al., 2005). These themes do not appear to reflect any specific psychopathology related to psychogenic illness, since they also appear in the narratives of patients who later received an organic diagnosis (e.g., multiple sclerosis: R. G. Edwards, Barlow, & Turner, 2008; Ehler–Danlos syndrome: Barnum, 2014; and Sneddon’s syndrome: Mildon, 2015).
A psychogenic diagnosis also influences the way patients are viewed by others. Findings in studies in which artificial patient vignettes were used suggest that a patient’s complaints may be rated as less serious if they are believed to be psychogenic, and the patient is more likely to be characterized negatively (De Ruddere et al., 2014; De Ruddere, Goubert, Stevens, Williams, & Crombez, 2013).
Most important, psychogenic explanations also have serious practical consequences. Many experts advocate the minimization of disability aids in such cases, since they may reinforce false illness beliefs or provide secondary gains (Davison, Sharpe, Wade, & Bass, 1999). Such denial of support would be a catastrophic error if the person’s disability were in fact intractable. A psychogenic explanation also may affect the patient’s future medical care. It is often recommended that further medical investigations be minimized (Deary, Chalder, & Sharpe, 2007; Hotopf, 2004). Physicians also may be more inclined to attribute any future physical complaints to psychological causes. Both these factors could combine to delay diagnosis and treatment of underlying or subsequently developing disease. Although there are no systematic studies of this issue, one recent survey of 12,000 patients with rare diseases showed that a psychological or psychiatric misdiagnosis substantially increased time to correct diagnosis compared with a nonpsychological misdiagnosis—in some illnesses, by more than 5 years (Kole & Faurisson, 2009). 2
Given the insufficient supporting evidence and the dangers of making an incorrect inference, we recommend researchers and practitioners refrain from claiming that the illnesses we have considered here have a psychological cause. Until better evidence exists, their causes are simply unknown. Also, the terms used to describe these illnesses (e.g., psychogenic) convey a sense of certainty about the underlying cause, which is extremely misleading in the light of current evidence. We recommend the alternative terms nonepileptic attack disorder, functional movement disorder, or functional neurological disorder, which although not entirely free of connotations, at least do not make explicit reference to psychological factors.
In this review, we identified a number of problems with current research approaches in this domain. These include the problems in teasing apart the direction of causal relationships within observational studies; the difficulties associated with inferring psychological traits from physical complaints alone (e.g., somatization); the biases that can occur when the psychological and medical assessments are not performed independently; the intercorrelated nature of many of the variables proposed to play a causal role in these illnesses; and the problems associated with retrospective self-report measures. To obtain more reliable evidence about the role of psychological variables in illness, researchers may need to use an entirely different approach. For example, by selecting participants who possess the psychological characteristics of interest before they exhibit any physical symptoms, investigators may be able to establish which of these variables reliably predicts illness outcomes (both explained and unexplained) at a later point.
Another serious problem concerns the “default” nature of psychogenic explanations. There is a massive gap in evidence standards required for classification of an illness as medically explained (organic) on the one hand and as psychogenic on the other. The former requires positive substantiating evidence and the latter little more than the absence of evidence (we have seen that even if positive signs are sought, these are unlikely to impact substantially on diagnostic outcome). In practice, disorders that fail to meet the criteria for a known organic illness readily slip through this evidence gap and become presumed to be psychogenic by default. This category becomes a wastebasket for any illness awaiting a full explanation and is likely to be highly heterogeneous. It is quite possible that some patients may suffer from a genuine psychological problem, whereas others might have some disease that is currently poorly understood; some might even be feigning illness. The only way to address these problems in future is to confront the unwarranted assumption that all “unexplained” illnesses have a single common cause. The next step is to seek positive criteria that may help identify specific subgroups.
Finally, we also identified some significant problems with existing theory in this area. Current psychogenic theories are often phrased in such general terms that they are difficult to falsify. For example, evidence suggestive of a role for biological factors in this disease would seem to call for a re-evaluation of certain theories. Instead, it has been accommodated within existing theories without alteration of their central tenets— for example, by proposing that patients may “model” the symptoms of previous illnesses or that neurological abnormalities in turn create “psychological vulnerabilities,” which in turn cause the illness. The possibility of a simpler, more direct causal relation is not considered. It is time to widen the net and consider other possible explanations for these illnesses.
In the domain of PMDs, some new ideas are already beginning to emerge. For example, it has recently been suggested that pain may play a critical mediating role in the development of some PMDs (Stone et al., 2009; van Hilten, Geraedts, & Marinus, 2007). The central idea is that the difficulty in movement develops gradually as a direct response to complex and chronic pain. The mechanisms responsible operate at a subcortical level, engendering a gradual change to movement behavior that is outside the person’s awareness.
Another recent proposal is that some PMDs may arise from inadequate top-down, goal-driven control of movement (Voon, Brezing, Gallea, & Hallett, 2011). Consequently, movement behavior is more easily shaped by factors outside the person’s current goals and intentions (including states of high emotional arousal). States of high emotional arousal in particular may activate problematic, abnormal movement plans, which then may compete for selection with the target movement. The poor top-down control itself need not have a psychological origin. It could be due to a loss of functionality within brain networks responsible for movement—perhaps as a consequence of an illness or injury—that may be difficult to detect with standard diagnostic techniques (see Felicio et al., 2010; Lang, Koller, & Fahn, 1995; Vuilleumier et al., 2001, for studies that support this hypothesis).
In the domain of PNESs, there has been relatively little consideration of the proximal neural mechanisms responsible for the PNESs themselves. However, a better understanding of these mechanisms is likely to reduce the explanatory burden placed on psychological factors. Research is beginning to make inroads here. In a recent study by Arzy et al. (2014), EEG was used to record electrophysiological activity during the actual PNESs of three individuals. All had a dual diagnosis of PNES and epilepsy. In all three, PNESs (but not epileptic seizures) were associated with decreased theta band activity focused in the left posterior parietal cortex. Given that theta band activity of this kind is thought to indicate tight synchronization between frontal and parietal regions, such as is required during cognitively demanding tasks (Sauseng, Klimesch, Schabus, & Doppelmayr, 2005), its sudden absence may be associated with a dramatic loss of cognitive capability and perhaps even loss of awareness. In one patient examined with implanted electrodes, the focus of this theta change and the focus of his epileptic seizures were in neighboring regions—a finding that calls into question whether the two types of seizures really are caused by entirely different mechanisms.
Of course, a key feature of both PMDs and PNESs is likely to be heterogeneity, so researchers are probably seeking not one explanation but several. In the case of PNESs, some events may turn out to have a physiological basis (e.g. cardiovascular events: Parra, Iriarte, & Kanner, 1999; Vossler, 1995; Zaidi, Clough, Cooper, Scheepers, & Fitzpatrick, 2000), others may represent other types of neurological events, and yet others may turn out to be well-characterized psychophysiological events such as panic attacks. Similarly, some PMDs may arise as a consequence of peripheral factors, such as localized pain, whereas others may be indicative of a more central neurological disturbance. The disordered movements may have been reinforced further through learning via subcortical mechanisms. Again, acknowledging the heterogeneous basis of these disorders is likely to be crucial to further progress.
Limitations
One major limitation of this study is that our conclusions were based almost entirely on evidence from two disorders: movement disorders and seizures. Many other complaints have also been attributed to psychological causes, from chronic pain and fatigue to visual and somatosensory disturbances. Most of the interpretational issues identified here apply equally to these other conditions but not necessarily its more substantive conclusions. The next step in this research is to extend the approach to other illnesses that are commonly explained within a psychogenic framework.
A second limitation is that although we adopted a systematic search strategy in some sections, in other sections with fewer studies or wide variations in study method and design, we used a more narrative approach. Also, even when we adopted a systematic approach, it fell short of the full standards required for a comprehensive systematic review. For example, we did not search all possible sources for study reports, nor did we take into consideration the influence of publication bias. A third limitation is that, in our review, we did not address the issue of deliberate feigning. An urgent future priority is to develop criteria that can be used to distinguish more effectively between genuine and feigned illness (see Nicholson, Stone, & Kanaan, 2011, for discussion).
Conclusion
Given our current limited understanding of medical disease and its markers, it is perhaps not surprising that not all physical complaints can be associated with a specific, well-recognized disease process. In these circumstances, it is tempting to offer a psychological explanation. However, in this review, we have argued that such explanations are not currently supported by the evidence, at least not in the domains considered here. Further, these explanations may be harmful for the patient. To find better treatments for these illnesses, researchers and practitioners may need to retire those overworked psychological explanations that are commonly invoked in the face of uncertainty and instead adopt a completely fresh perspective. Such an approach may lead to a much deeper understanding of this perplexing collection of illnesses.
Footnotes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
