Reinforcement Sensitivity Theory (RST): The Neuroscience of Personality and Psychopathology
Convenor: Philip Corr, Swansea University
Purpose: The aim of this symposium is to discuss critically the Gray and McNaughton (2000) revision of Gray’s original highly influential Behavioural Inhibition System (BIS) model of anxiety, that has developed over the years to form a complex neuropsychological theory of emotion, motivation and personality -- now widely known as Reinforcement Sensitivity Theory (RST; for a summary, see Corr, 2008) with specific implications for understanding clinical disorders. Background: This revised theory postulates three major brain-behavioural systems: (a) the fight-flight-freeze system (FFFS), which mediates reactions to all aversive stimulus and controls avoidance, escape and attack responses, and is associated with the emotion of fear; (b) the behavioural approach system (BAS), which mediates reactions to all appetitive stimuli and controls approach behaviour, and is associated with the emotion of anticipatory hope/pleasure; and (c) the behavioural inhibition system (BIS), which is activated by goal-conflict (of any kind), and which serves as a goal-conflict resolution mechanism (e.g. when there is a conflict between approach and avoidance motivations) the BIS produces cautious, risk-assessing approach behaviour, and is associated with the emotion of anxiety. The complexity of the revised theory is greatly simplified with reference to a two-dimensional model of defensive intensity (i.e. strength of defensive reaction) and defensive direction (i.e. avoidance/fear or cautious approach/anxiety) (McNaughton & Corr, 2004).
Methods: This symposium provides a themed discussion of RST from active researchers in the field, and the Discussant (Dr Colin Cooper) has been chosen to provide a critique and overview of why RST fits into current thinking and research in personality and psychopathology. The contributors compose Professor Philip J. Corr, who describes the development of RST, how it relates to psychopathology, and how it may be used as an example of unifying experimental and correlations approaches in psychology; Dr Andrew Cooper presents data on the psychometric nature of the personality factors hypothesised by RST to underlie normal and abnormal expressions of emotion; Mr Adam Perkins presents data on a novel behavioural task that measures one-way (putatively fear-related) and two-way (putatively anxiety-related) avoidance in an attempt to provide experimental support for the differentiation of fear and anxiety; and Professor Alan Pickering discusses the results of compututional modelling to reveal the dopamine basis of BAS-related approch motivation.
Conclusions: This symposium will summarise the highly influential work of Jeffrey Gray and colleagues, that is widely recognised around the world as an important contribution to understanding how state activation in major emotion/motivational neural systems relates to trait differences in personality, and thus to predispositions to developing clinical disorders. The symposium will showcase some of the successes of RST, but also highlight the problems it faces and the future work that shall be required to achieve a mature neuroscience of personality and psychopathology.
The Reinforcement Sensitivity Theory (RST) of Personality: Conceptual and Empirical Foundations
P.J. Corr, Swansea University
Purpose: To sketch the conceptual and empirical foundations of revised Reinforcement Sensitivity Theory (RST; Gray & McNaughton, 2000; McNaughton & Corr, 2004) of personality in order to show its successes as well as the challenges it faces.
Background: RST represents a major theoretical account of the neural and psychological processes underlying the variations observed in the major dimensions of personality, principally Extraversion (E) and Neuroticism (N). Inspired by the individual differences research of Ivan Pavlov and guided by the animal learning studies (e.g., H. O Mowrer), the neurophysiology of reinforcement (e.g. J. Olds), as well as the psychometric and laboratory studies of H. J. Eysenck, Jeffrey Gray spent 40 years developing an elegant model of emotion and motivational systems, including individual differences in the functioning these systems (ex hyposthesi, personality). This talk traces the development of RST, from its official birth in 1970, through to the highly influential 1982 The Neuropsychology of Anxiety and onto its major revision on 2000 with the second edition of this book (co-authored with Neil McNaughton).
Methods: A literature review is provided, detailing the principal conceptual milestones and refinements of the model, illustrated by empirical data, especially from the author’s own laboratory.
Conclusions: It will be shown that RST offers a fecund account of the main state neuropsychological systems underlying emotion and motivation, as well as how these systems relate to trait differences in personality. It will be shown that RST offers a cogent account of the specificity of clinical disorders (e.g. panic vs. OCD) as well as the generality of such disorders as evidenced by comorbidity and the fuzziness of diagnostic categories seen in many clinical cases. Although in a continued state of development, it will be argued that RST offers an integrative model of the individual differences seen in the activity and outputs of the main brain-behavioural systems, and offers psychology a paradigm for linking personality predisposition and psychopathology in theoretically compelling fashion that holds important implications for the aetiology, maintenance and treatment of clinical disorders. It will also be argued that RST, in this specific context, shows how it is possible to unify the two major schools: experimental and correlational, which was one of the central themes in the author’s BPS 2007 Hans Eysenck Memorial Lecture.
Self-report measures of anxiety, fear and behavioural inhibition
A. Cooper, Goldsmiths College, University of London
Objectives: This presentation examines the relations between self-report measures of anxiety, fear and the Behavioural Inhibition System (BIS) using a confirmatory factor analytic approach. Structural models were also tested for measurement and invariance across gender. The presentation also aims to present preliminary data from the development of recently devised self-report measures relevant to the revised Reinforcement Sensitivity Theory (RST) of personality.
Design: This study employed a correlational design. Confirmatory factor analysis (CFA) was used to explore relations between the measures used. Model fit was tested for a number of nested CFA models representing relations between the anxiety, fear and behavioural inhibition measures. Once a suitable model was established, a further sequence of nested CFA models were used to test for structural and measurement invariance across gender.
Methods: 340 university students and local community members (167 males and 173 females) aged between 18 and 77 completed the Fear Survey Schedule (FSS), the Spielberger Trait Anxiety Inventory and the Carver and White Behavioural Inhibition Scale. Participants completed the measures as part of a battery of similar personality questionnaires.
Results: Using CFA, the results showed that while trait anxiety, behavioural inhibition and social fear (FSS) are distinct constructs, they are strongly positively related. Tissue damage fear (from the FSS) was, however, only weakly related to trait anxiety and behavioural inhibition. The best fitting model of these constructs was also shown to have measurement and structural invariance across gender.
Conclusions: The results highlight that, although personality measures of negative emotionality tend to be significantly positively related, there is sufficient evidence to suggest that different measures of anxiety, fear and behavioural inhibition potentially tap somewhat different processes. These results further suggest that the development of new self-report personality measures that include a more diverse range of processes related to threat sensitivity and avoidance may be useful in the context of the revised RST. A theoretical structure that has informed the development of a new set of RST scales will be outlined, and preliminary empirical data will be presented to show how the different levels of neural and psychological hierarchy in revised RST can be modelled and measured by questionnaire.
Reactions to threat and personality using a joystick operated runway task
A. Perkins, Swansea University
Objectives: To test Gray and McNaughton’s (2000) theory that fear is elicited by threats that need not be faced (i.e., avoided) and anxiety by threats that need to be faced (i.e. cautious approach).
Design: Scores on well-established questionnaire measures of Spielberger’s (STAI) trait anxiety and fear (Fear Survey Schedule, FSS) were compared to avoidance behaviour on a novel joystick operated runway task developed specifically for this study.
Methods: 108 participants were recruited from a university population. In session one, participants completed several psychometric questionnaires. In session two, participants completed the joystick operated runway task (JORT). In this task, a joystick was used to move a cursor along an on-screen runway, with high force applied to the joystick handle (relative to the participant’s maximum physical strength) equalling high speed along the runway. In Phase 1 of the task, the cursor stimulus was pursued by a threat stimulus that resulted in a burst of white noise to the participant when the threat stimulus caught up with the curser stimulus. Prior to this test phase, participants underwent a training phase with no threat of white noise and then a conditioning trial in which the cursor stimulus was caught and a burst of white noise was received. In phase 2 of the task, a second threat stimulus was added in front of the cursor stimulus so that participants were placed in a (theoretically) anxiety-inducing conflict situation: they had to move fast enough to avoid the pursuing threat stimulus but not so fast that they hit the threat stimulus in front (moving in the same direction). Specific hypotheses tested were: (a) that a positive association would exist between fear (FSS) scores and the increase in runway speed due to threat of white noise in Phase 1 of the JORT; and (b) that a positive association will exist between anxiety (STAI) scores and the magnitude of approach-withdrawal oscillations in Phase 2 of the JORT.
Results: Partially supportive of Gray and McNaughton’s (2000) hypothesis, in Phase 1, high scores on FFFS tissue damage fear boosted speed along the runway when pursued by a threat stimulus, but high scores on social fear had an opposite effect (this effect was only significant in males). Over the course of the experiment, the distance between the two threat stimuli in Phase 2 of the task was twice decreased and results showed that trait anxiety became a progressively stronger predictor of oscillation during conflict as this occurred, supportive of the Gray and McNaughton (2000) theory.
Conclusions: Reactions to threat, as measured on a joystick operated runway task, relate to scores on well-established questionnaire measures of personality partially as predicted. Results, therefore, offer some empirical support for the idea that tissue damage fear may index fear as portrayed by Gray and McNaughton (2000) and, additionally, confirm factor analytic studies of fear questionnaires that suggest tissue damage fear and social fear are distinct constructs.
Reward prediction errors and Reinforcement Sensitivity Theory: A neurobiological account
A. Pickering, F. Pesola, Goldsmiths College, University of London
Purpose: To illustrate a novel approach to developing and testing ‘bottom-up’ theories of personality, such as reinforcement sensitivity theory (RST).
Background: RST is a theory of personality which argues that inter-individual variation in the sensitivities of 3 fundamental brain-behavioural systems underlies major dimensions of personality, such as extraversion/impulsivity and neuroticism/anxiety. RST proposes that impulsivity (or extraversion) corresponds to variations in the sensitivity of the behavioural activation system (BAS). The BAS is thought to respond to rewarding stimuli thereby producing approach behaviour. Since RST was first proposed, knowledge of the neurobiology of the reward learning and motivation system has increased dramatically and so we reconsidered the theory in light of this new knowledge. In particular, it is widely argued that midbrain dopamine cells in the brain’s reward systems fire in response not to rewards themselves, but to reward prediction errors (RPEs; discrepancies between actual rewards and those predicted based on past experiences). We explored whether a personality trait, linked with brain dopamine systems, might correspond to sensitivity to RPEs rather than rewards (as proposed in traditional RST).
Methods: The study used a computational model of learning under reward to simulate probabilistic perceptual category learning. The model was neurally constrained by knowledge of the brain’s reward systems. Task performance was simulated under two rewarded conditions: in the symmetric payoff matrix, stimuli from each category being learned received equivalent points for correct categorisations and lower but equivalent points for errors; in the asymmetric payoff matrix one category received a much larger number of points for correct responses than the other with lower but equivalent points for errors. This motivational manipulation leads human participants, in the asymmetric condition, to shift their category boundaries in the direction of maximising winnings but sacrificing accuracy; participants in the symmetric condition place their category boundary at the point of maximal accuracy. The ability of the model to simulate these boundary shifts depends critically on the relative RPEs occurring on each trial. However, the simulations were found to be highly dependent on the absolute value of the RPEs, with larger RPEs inhibiting the ability to produce the boundary shift in the asymmetric condition but having no effect in the symmetric condition. Assuming that a personality trait corresponds to sensitivity to RPEs then the model predicts personality-behaviour correlations in real participants in the category learning task. We measured three dopamine-related candidate traits (extraversion, impulsive sensation seeking, and positive schizotypy) in 64 participants and correlated personality with category learning performance. Positive schizotypy, but not the other traits, was associated with the size of the category boundary shift in the asymmetric condition (but not in the symmetric condition).
Conclusions: RST may need to be expanded to reflect associations between personality traits and sensitivity to RPEs (rather than rewards); preliminary results suggest positive schizotypy might be the trait that is relevant to this revision.