p.3
1 Hans Eysenckâs contribution to our understanding of personality and psychological disorders
A personal view
Gordon Claridge
In his lifetime Hans Eysenck was such a controversial figure that any comment on his work is bound to be touched by individual bias; even the serious academic research discussed here, let alone his ventures into socially divisive topics and fringe areas of science. His ideas about personality were contentious and evoked strong opinions, even from those of us who knew and worked with him. It is appropriate to let an autobiographical theme run through this paper, in order to explain how Eysenck influenced my own thinking and my eventual conclusions about his contribution to personality theory and abnormal psychology. I have already covered part of the topic in several previous publications (Claridge, 1981, 1983, 2009a) and fully in an earlier Festschrift to Eysenck (1997). Given the lighter touch of the present piece, consulting, especially, the last of those papers will help the reader fill in details of evidence and arguments about the various themes to be introduced (Claridge, 1997a).
I first encountered Eysenck in the early 1950s when, as an undergraduate in the Psychology Department at University College London, I took his lecture course on personality. Even to my naive student eye it was obvious that the Department was not at the best place in its previous and subsequent prestigious history. I was aware that the famous â eventually to become infamous â Cyril Burt had just retired and the place had the taste of a collection of leftovers, spiced up by large dollops of Hullian learning theory, which seemed pointlessly irrelevant to the human psychology I had gone to university to study. I was probably on the point of dropping out. For good or ill Eysenck saved me from that fate.
Eysenckâs lectures were a revelation. With the lucidity and confidence that we all came to recognise as the hallmark of his public presentational style, he laid out an approach to individual differences that seemed awesomely complete: a statistically derived account of the descriptive features (dimensions) of personality; an explanation of their underlying biology, accessible through laboratory investigations (In my enthusiasm I obviously failed to notice his heavy reliance there on learning theory concepts and methodology!); and a connection to psychological disorders, envisaged as merely extreme positions on personality dimensions. In that last part of his theory he was foreshadowing his eventual vehement dismissal of the medical model (Eysenck, 1960), a stance that evoked much furore among the psychiatric profession but the arguments for which were already obvious to disciples such as myself â reflected in my scepticism about the content of a parallel series of lectures I was attending on mental âdiseasesâ, given by a clinical psychiatrist.
p.4
Despite considerable shifts in my research interests over the years, the power of this âEysenck epiphanyâ has lasted a lifetime and traces of it still pervade my thinking; especially about the dimensionality of personality and spectrum approach to disorder, and a belief that biology and genetics must play some part in shaping these variations. (Nothing new there to some contemporary observers but easy to forget that Eysenck pioneered both ideas in the modern era.) I have assimilated alternative perspectives, but that has always required a curious kind of effort not demanded by the Eysenck dogma. I sometimes liken the phenomenon to the experience I have noticed in lapsed Roman Catholics among my friends and colleagues: try as they might they can never fully shake off the last vestiges of their faith.
Eysenckâs influence on my thinking â and eventually my perceptions of the flaws in his work â were reinforced when he took me on as a research assistant. My remit was to test out in psychiatric patients the âcausalâ part of his theory, viz. using laboratory measures to explore correlates of the psychological disorders that formed the âcriterion groupsâ allegedly defining the end points of introversionâextraversion (IâE): anxiety based neuroses (dysthymics, as Eysenck called them) for introversion; and, for extraversion, various forms of hysterical disorder, as well as psychopathy. Neuroticism (N), being an orthogonal dimension and therefore assumed to be similarly high in both clinical groups, was not at the time considered relevant; predictions from the theory were solely about differences between the extreme manifestation of IâE.
A further word about dysthymiaâhysteria, the clinical manifestation, according to Eysenck (1957b), of extreme introversion and extraversion. The introverted end was unproblematic in being defined by a recognisable and relatively homogenous group of neurotic disorders. This was less true at the extraverted end, as eventually proved to be the case in subsequent thinking about those conditions, both within and outwith Eysenckian theory. Psychopathy was later hived off to define the third, psychoticism (P), dimension (Eysenck & Eysenck, 1976). Meanwhile âhysteriaâ, because of its dubious sexist connotations, dropped out of psychiatry altogether. The ragbag of disorders previously denoted by the term âhysteriaâ remained in the psychiatric classification, to be spread diagnostically across other abnormal reactions to stress: somatoform and dissociative. In the research discussed below, using the old nomenclature â that also included a category of hysterical personality disorder â and bundling in psychopathy as well, made the âextraverted disordersâ group quite mixed, diagnostically: they were often labelled âhysterico-psychopathsâ for the purpose of data analysis and reporting. To anticipate slightly, it actually turned out that there were very few differences between the various subgroups on important experimental measures, suggesting that the heterogeneity across that half of the sample was not as marked as might have been feared.
Despite the fact that Eysenck had promoted his theory as an explanation of both normal introversionâextraversion and clinical dysthymiaâhysteria, a comparison had never been properly made of the two criterion groups. Work attempting to connect the personality and clinical domains had been confined, albeit with a vigorous debate, to questionnaire studies (McGuire, Mowbray, & Vallance, 1963; Sigal, Star, & Franks, 1958; Slater, 1960; Storms & Sigal, 1958). Results there were ambiguous and Eysenck judged â to my good fortune â that a comprehensive laboratory based investigation of the questions raised was necessary.
p.5
The setting for the research was unusual and deserves mention. My job was based not in the Institute of Psychiatry but, by special arrangement with the army, at the Royal Victoria Military Hospital, Netley near Southampton. There, freed from the constraints of actually working in the IoP Psychology Department and away from Eysenckâs eagle eye, I established an outpost where I was free to do what I liked â and did: pursuing lines of research not immediately connected with testing his theory of neurosis. That âunofficialâ research involved studying an extra group of psychotic patients, unconnected at the time to the main hypotheses about neurosis and personality disorder. The work is not relevant to the present discussion, though it did introduce me to the topic of psychosis and eventually, among other things, shaped my thoughts about Eysenckâs psychoticism dimension, returned to later. The whole programme of research at Netley â as it was known for short â was published in my book Personality and Arousal (Claridge, 1967).
Netley, now demolished, was a unique hospital, ideal for the research contemplated. As well as a ready supply of easily matched healthy controls, it offered patients covering all of the diagnoses necessary for the research. Referring back to my earlier remarks, it is worth noting the particular availability of patients falling into the category of hysteria, especially those with the now defunct diagnosis of conversion hysteria. Reportedly infrequently seen at that time in civilian psychiatric practice, such individuals were very common in the military setting. Patients covered the whole range of expression of that once commonly diagnosed syndrome of pseudo-neurological disorder; viz. psychogenic blindness and deafness, anaesthesia and partial paralysis of limbs, and fugue (amnesic) states. Importantly â especially for that particular patient subgroup â all of the participants were young fit men, having been screened on entry to the army for obvious physical complications. Furthermore, the way the military services processed their psychiatric casualties meant that patients when tested were generally free of, or on minimal, medication.
The battery of behavioural procedures administered covered a wide variety of laboratory measures then regularly employed in Eysenckâs department to examine individual differences. They included tests of perception (e.g., Archimedes spiral after effect), motor performance (serial reaction time and Stroop interference), and auditory vigilance. To which we added a group of psychophysiological measures: EEG indices and tests of drug response. Among the latter the most salient â and figuring large in later research explicating certain features of Eysenckâs theory â was the sedation threshold. This was a procedure used to determine individualsâ tolerance of sedative drugs by injecting them intravenously with a barbiturate (commonly amylobarbitone sodium), continuing the infusion until they reached a defined end-point of loss of consciousness. It was introduced into psychiatry as a diagnostic tool by the American clinician Shagass (1954), who used EEG changes as his criterion for the threshold of sedation. Subsequently, in the 1950s and early 1960s there was a flurry of research on the technique, both by Shagass himself and by others exploring alternative criteria for determining the sedation threshold. Among them, at Netley, a psychiatrist colleague and I introduced a simple behavioural criterion (Claridge & Herrington, 1960).
p.6
Summarising the part of the work conducted at Netley on neurosis, two facts stood out. The first was that there was very good support for the predicted difference between the two criterion groups of anxiety based (dysthymic) patients and hysterico-psychopaths. (As noted above, there were few, if any, differences on objective laboratory measures among the subgroups of patients making up that diagnostically broad category.) The effects were especially evident for tests like auditory vigilance and sedation threshold, where dysthymics proved to have a significantly better vigilance performance and greater tolerance of the depressant drug (higher sedation threshold) than hysterico-psychopaths. The drug finding, incidentally, was scarcely a novel discovery since it had already been demonstrated by Shagass on a substantial sample of patients (Shagass & Jones, 1958). Still, it pleased Eysenck that his causal theory had been vindicated!
The other main finding from the study did not fit in with Eysenckâs dysthymiaâ hysteria story. Since according to theory the neurotic criterion groups were merely abnormal counterparts of introversionâextraversion, comparing them should have nothing to do with N; being independent of IâE the latter logically should have had no influence, even in clinical populations. This proved manifestly not true in our research, where factor analysis of the psychophysiological data demonstrated what we identified as two distinct components of âarousalâ, and led us to conclude that at the causal level both IâE and N contributed to dysthymiaâhysteria, as an interaction between the two dimensions. At an individual test level, this was dramatically illustrated in some highly replicable findings on drug tolerance differences to be found among non-clinical subjects, assessed for IâE and N. The results came from studies examining both nitrous oxide tolerance (Rodnight & Gooch, 1963) and ones using the regular barbiturate (sedation threshold) procedure (Claridge & Ross, 1973; Claridge, Donald, & Birchall, 1981).
In all of those experiments samples were subdivided into high and low scorers on the neuroticism scale. It emerged that IâE was related to sedative drug tolerance in opposite directions in the two subgroups of N. There was higher tolerance in neurotic introverts compared with neurotic extraverts, thus paralleling the patient data. But quite the reverse was true in low N subjects; in that case it was extraverts who had the greater tolerance. These findings were damaging for Eysenckâs theory, not only because of the novelty of the crossover effect itself, but also because it undermined his general prediction about the expected relationship between IâE and biological measures. Pointedly so, given the special âdrug postulateâ he had deliberately introduced to connect personality study to psychopharmacology: the proposal was that psychoactive drugs could be used as a special experimental tool to explore the biological correlates of introversionâextraversion (Eysenck, 1963).
p.7
In his book The Biological Basis of Personality, Eysenck (1967) acknowledged the possibility of interactions between IâE and N and recommended that data sets be routinely examined for such effects using so-called âzone analysisâ. This proposed comparing the performance scores for individuals subdivided according to the four combinations of IâE and N. At the same time he revamped his biological model, bringing it more into line with Western thinking and terminology about the nervous system. Pavlovian nervous typological theory had served him well â and it is instructive to remember that he had relied on it to pioneer the application of biology to the understanding of human personality. But its language was quaint and the constructs ambiguous and difficult to apply. It was time for it to be abandoned even in the face of continuing attempts to retain connections to the Pavlovian and neo-Pavlovian schools of thought (see Nebylitsyn & Gray, 1973).
Unfortunately, neither the methodological nor the conceptual changes Eysenck brought to his thinking produced much that was new in understanding personality differences. Zone analyses of data hinted that an explanation of the IâE/N interaction might lie in inverted-U effects (Claridge et al., 1981; Eysenck, 1967). But that was about all and zone analysis disappeared into a dusty box of procedures never to be opened again, except on occasions such as this. And the new model of the âconceptual nervous systemâ (acknowledgment to Hebb, 1955) elucidated little. The suggestion was that IâE reflected variations in ascending reticular (ARAS) arousal and N differences in limbic system activation. While this looked impressively âneurophysiologicalâ it proved to be more descriptive than explanatory.1
My personal view is that Eysenckâs 1967 book marked the end of his novel contribution to our understanding of individual differences, as envisaged within his original two-dimensional framework. This is not to say that the book is without purpose. On the contrary, it should be read by anyone who has doubts about possible biological and behavioural genetic involvement in healthy and unhealthy personality: it contains a vast amount of evidence still relevant to current debates. And work inspired by the IâE/N formulation certainly rumbled on; for example in applications to medical topics such as smoking and lung cancer (see Nias, 1997). But what ultimately killed off Eysenckâs own message was the revision to his theory suggested by Jeffrey Gray (1981). Grayâs proposal for two alternative dimensions of anxiety and impulsivity â drawn diagonally across IâE and N â made clinical sense. The scheme indicated a more explicit connection to disorder, at least on the anxiety side, and suggested a more obvious way of mapping the descriptive features onto a possible underlying biology. In particular, Grayâs reinforcement theory â anxiety and impulsivity representing sensitivity to, respectively, punishment and reward â seemed neater and less messy than Eysenckâs own formulation. The recent history of research within the âEysenckian schoolâ has, therefore, essentially been a history of pursuing and elaborating on Grayâs theory â see Pickering et al. (1997) and other papers in this issue.
Irrespective of the judged success or otherwise of that later work, the form of the Gray revision â and its subsequent versions â and the language in which they are couched raises an important question that needs to be addressed here. How far are such formulations theories of personality, in the full sense of the word? The issue can be articulated as a distinction between temperament and personality, one that in the adult field has not often been explicitly debated (though see Claridge, 2006; Claridge & Davis, 2003; Strelau & Zawadzki, 1997). Temperament is generally regarded as the âsimplerâ construct, reflecting emotional and motivational differences, grounded in quite genetically determined biological processes, readily observable in animals as well as in humans. The term has been more often applied in a developmental context, as a way of characterising apparently constitutional differences among children; giving rise to such formulations as the three-dimensional EAS scheme presented by Buss and Plomin (1984) who explicitly used strong heritability as a necessary criterion for a trait to be considered temperamental.
p.8
Judged in that light Eysenckâs theory and derivatives of it â and others in the same genre (e.g., Zuckermanâs, 1979, sensation seeking model) â are, in my view, more than anything accounts of temperament. Tracing Eysenckâs theory to its origins makes this obvious, where key parts of his thinking were shaped by the ancient theory of temperaments: as a basis for defining his descriptive dimensions and, at the causal level, by drawing upon Pavlovâs nervous type explanation of it, developed in his studies of dogs. Subsequently, in a sleight of Hans at which in argumentation he was notoriously adept, Eysenck tried to persuade us that his theory offered a comprehensive account of behaviours in many domains, even beyond strict personality differences, including social and political attitudes. Yet, as his 1967 revision reminds us, it remained at bottom (sic!) an account of psychological functions driven from below by relatively low brain structures. The impression (and perhaps the intention) given was that biology rules in a determinist, reductionist sense.2
Does this matter? It could be argued not, if pure temperament theories can elucidate behaviours that might be strongly determined by underlying hereditary traits and, therefore, seemingly âbiologicalâ in nature. In the abnormal field, primary psychopathy comes to mind (Hare, 1993; Viding, 2004). However, even there, understanding of behaviour can benefit from insights that more fully formed accounts of personality can provide. For example, the almost universal child abuse and neglect found in psychiatric patients is a factor that pure temperament theories are ill-equipped to deal with, because of the limited scope of their constructs. A researcher who has recognised this fact is Cloninger (2006) with his formulation of personality that proposes a two-layered structure: temperament, containing motivational elements, like novelty seeking, familiar in other similar theories; and character made up of higher order psychological traits, such as self-directedness and cooperativeness. Cloninger gives a ...