Hypnotic suggestibility, cognitive inhibition, and dissociation

Abstract

We examined two potential correlates of hypnotic suggestibility: dissociation and cognitive inhibition. Dissociation is the foundation of two of the major theories of hypnosis and other theories commonly postulate that hypnotic responding is a result of attentional abilities (including inhibition). Participants were administered the Waterloo-Stanford Group Scale of Hypnotic Susceptibility, Form C. Under the guise of an unrelated study, 180 of these participants also completed: a version of the Dissociative Experiences Scale that is normally distributed in non-clinical populations; a latent inhibition task, a spatial negative priming task, and a memory task designed to measure negative priming. The data ruled out even moderate correlations between hypnotic suggestibility and all the measures of dissociation and cognitive inhibition overall, though they also indicated gender differences. The results are a challenge for existing theories of hypnosis.

Introduction

In response to suggestion, highly hypnotizable participants report and display a wide variety of interesting responses (Hilgard, 1965). They report hallucinations, negative hallucinations (not seeing or hearing a stimulus that is present), and clinically significant degrees of pain reduction. They display selective amnesia, partial paralyses, and vastly reduced Stroop interference (e.g. Raz, Shapiro, Fan, & Posner, 2002). With the exception of the Stroop effect modulation, one might suspect that these are merely enacted responses performed in compliance with the perceived demand characteristics of the experimental situation. However, behavioural and physiological data converge to indicate that these reports and responses reflect genuine changes in experience (e.g. Kinnunen et al., 1994, Kirsch et al., 1989, Oakley, 2008).

One of the most notable characteristics of hypnosis is the stability of individual differences in responsiveness to suggestion. Test–retest correlations of .75 have been reported for a standard measure of hypnotic suggestibility over a 25-year interval (Piccione, Hilgard, & Zimbardo, 1989). Despite the prominence and reliability of individual differences in responding, correlates of hypnotic suggestibility are notoriously difficult to find. This is disappointing because theories of hypnosis suggest that various predictors of hypnotic suggestibility should be detectable. For example, dissociation theories (e.g. Hilgard, 1986, Woody and Bowers, 1994) regard hypnosis as involving a fractionation in the normal cognitive control systems. Dissociation theories are consistent with the association of hypnotisability with dissociative disorders in clinical populations (Bryant et al., 2001, Frischholz et al., 1992, Stutman and Bliss, 1985). But these theories also predict that in non-clinical as well as clinical populations people who are highly responsive to hypnotic suggestion should be more liable than low suggestible people to dissociate in everyday life (see Kirsch & Lynn, 1998, and Woody & Sadler, 2008, for comprehensive reviews of these theories). Surprisingly, the self-reported tendency to dissociate in everyday life has not been found to reliably predict hypnotic responding (e.g. Kirsch and Council, 1992, Nadon et al., 1991; though for an exception see Butler & Bryant, 1997). Other theories postulate that hypnosis involves a particular deployment of attention in ways unavailable to low suggestible people (e.g. Crawford and Gruzelier, 1992, Raz et al., 2005). Indeed, one of the most frequently replicated correlates of hypnotic suggestibility is self-reported absorption in imaginative activities (e.g. Barnier and McConkey, 1999, Lyons and Crawford, 1997, Tellegen and Atkinson, 1974, Wilson and Barber, 1981; contrast Jamieson & Sheehan, 2002). Even so, when methodological artefacts are controlled, absorption only accounts for about 1% of the variance in hypnotic responding (Kirsch & Council, 1992). Because these low correlations counter influential theories, we will explore further the relation between hypnotic suggestibility and dissociation, and between hypnotic suggestibility and cognitive inhibition – an aspect of attentional functioning theoretically important for hypnosis.

Dissociation is ‘a disturbance or alteration in the normally integrative functions of identity, memory, or consciousness’ (DSM), involving a lack of integration of mental processes with ongoing conscious experience or a consciously experienced sense of disconnection (e.g. Cardeña, 1994). Dissociation is the basis of two of the most popular theories of hypnosis, neo-dissociation theory (Hilgard, 1986) and dissociated control theory (Woody & Bowers, 1994). According to neo-dissociation theory, hypnotic responding consists of splitting the central executive so that one part controls hypnotic responding without the other part being aware of it. According to dissociated control theory, hypnotic responding consists of splitting an action schema from central executive control, so that it is activated directly by the hypnotist’s words. Both theories postulate that high hypnotic suggestibility results from an ability to produce the relevant fractionation. If either theory is right, there ought to be a substantial association between tendencies to dissociate and hypnotic suggestibility. Although the empirical data fail to demonstrate an association between dissociation and suggestibility (Kirsch & Council, 1992), this failure may be due to characteristics of the scale that is most commonly used to assess tendencies to dissociate.

Dissociation is most commonly assessed on the Dissociative Experience Scale (DES; Bernstein & Putnam, 1986). The DES was developed by people who worked with clients with dissociative disorders and is focused on people who dissociate frequently. As a result, the scale is highly skewed when administered to non-clinical populations, with most people scoring at the extreme low end of the scale. The extreme skew of the DES with non-clinical populations may be the reason for the low correlations with hypnotic suggestibility (e.g. Nadon et al., 1991).

Wright and Loftus (1999) reported a revised form of the DES (the Dissociative Experience Scale-Comparative; DES-C) specifically geared toward decreasing the skew of the scale when assessing dissociation in a normal population. Skew was reduced by amending the response scale for the DES. In the original, participants had to estimate the number of times that they had each of 28 dissociative experiences. Instead, Wright and Loftus asked their participants how often they had each experience compared with other people. This small change made the overall score roughly normally distributed with their samples of university students.

We used the revised DES-C as a predictor of hypnotic suggestibility. A meaningful association with hypnotic responding would provide empirical support for neo-dissociation theory and dissociated control theory. Conversely, an only trivial association would count against neo-dissociation and dissociated control theories.

To respond successfully to hypnotic suggestions, participants apparently focus attention on the experience that is suggested and ignore competing ideas (Barber et al., 1974, Hilgard, 1970, Kirsch et al., 1999). Thus, it is natural to suggest that hypnotic suggestibility involves an ability to block or inhibit potentially competing information. For example, neo-dissociation theory (Hilgard, 1986) proposes that highly hypnotisable subjects (henceforth ‘highs’) have a special means, namely dissociation, for blocking information from reaching consciousness. Other theorists have argued hypnotisability can be seen as consisting of the ability to disattend information (cf. Crawford et al., 1993, Crawford and Gruzelier, 1992). Crawford and Gruzelier argue that highs are especially able both to disattend from some stimuli and to focus on others. Gruzelier (1998) and Gruzelier and Warren (1993) argued that the ability of highs to sustain this attentional state allows them to exhaust their frontal abilities during an induction, and hence end up frontally impaired in a way that constitutes the hypnotic state. By contrast, Crawford (e.g. Crawford, Knebel, & Vendemia, 1998) saw hypnotic responding to stimuli such as pain as dependent on the continued effective functioning of the executive and inhibitory system in highs. Similarly, Baars (1988) thought hypnosis consisted of keeping contradictory information out of awareness. The view that highly hypnotisable subjects have a special ability in cognitive inhibition (or anything else) can be contrasted with the view that they simply have appropriate attitudes, interpretations, and beliefs (cf. Spanos, 1986). It is thus theoretically important to establish whether highs do have any special ability to inhibit information.

The relation between inhibitory ability without a hypnotic induction and hypnotic suggestibility has been studied most directly using the Stroop effect, random number generation, pro-active interference, and more recently with a memory priming paradigm. Studies using the Stroop test have produced conflicting findings. Without hypnotic induction or suggestions being used, most studies have found no significant difference between highs and lows on Stroop interference (Aikens and Ray, 2001, Egner et al., 2005, Kaiser et al., 1997; Kallio, Revonsuo, Hämäläinen, Markela, & Gruzelier, 2001). Dixon, Brunet, and Laurence (1990) and Dixon and Laurence (1992) found significantly more Stroop interference in highs than lows; however, Rubichi, Ricci, Padovani, and Scaglietti (2005) found significantly less Stroop interference in highs rather than lows. On a related task, Iani, Ricci, Gherri, and Rubichi (2006) found that highs and lows without an induction were not detectably different in terms of the effect of irrelevant flanking items on the classification of a central letter.

Random number generation requires inhibitory ability. For subjects to be successful at randomising their output, they need to inhibit clichéd responding (such as counting upwards) and inhibit recent responses to avoid them being disproportionately likely. Graham and Evans (1977) found that highs were better than lows at random number generation. Replicability is unfortunately low: Evans (1991) reports replications by three other teams of this relationship while Crawford et al. (1993) found no significant relationship between hypnotic suggestibility and random number generation, and cite three other null findings, but none with a significant difference in the opposite direction. In sum, results are sporadic but significant results have only been found with highs outperforming lows, suggesting highs may have a greater inhibitory ability than lows. On the other hand, they may reflect the greater motivation of highs rather than lows to perform difficult tasks in a hypnosis lab.

Farvolden and Woody (2004) tested pro-active interference in highs and lows. Participants were trained on one set of paired associates (AB) then on three study-test trials of a second set (AC). On the first test trial of the second set, highs made more errors in recalling C to the cue A than lows did. This may be due to an inability by highs to inhibit the AB pairings, though overall ability to recall was not statistically controlled in testing the difference between highs and lows on the AC trial.

A further way of assessing cognitive inhibition is with a negative priming task, in which participants are instructed to attend to some stimuli and ignore others (Neill, Valdes, & Terry, 1995). The increase in latency to identify previously ignored stimuli compared to control stimuli is referred to as negative priming and interpreted as an indication of cognitive inhibition. David and Brown (2002) used a memory priming task fitting this specification. In an initial phase participants were instructed either to attend to or to ignore a set of words. In the test phase, participants classified the same words into categories. David and Brown argued that to-be-ignored words were actively inhibited, making them slower to later be classified. They obtained significant correlations (about .40) between suggestibility and response latency to classifying the to-be-ignored words, a finding they interpreted as showing highs were good at inhibition.

Although the David and Brown (2002) study appears to support the hypothesis that hypnotic suggestibility is associated with cognitive inhibition, there are two problems with their interpretation of the data. First, the statistical analysis did not control for response latencies to stimuli to which they had been instructed to attend during the priming task. Indeed, Kirsch and Braffman (2001) reported a small but significant association between hypnotic suggestibility and simple reaction time. Negative priming refers to differences in processing between responses to control stimuli and previously ignored stimuli. Therefore, by failing to control of response times to control stimuli, the association they obtained may not have involved negative priming at all. Instead, it may have been a replication of the association of suggestibility with simple reaction time, as reported by Kirsch and Braffman (2001). The second problem is that without a control group who only classified the words without pre-exposure to them, it is impossible to know whether any difference between to-be-attended and to-be-ignored words results from inhibition of the to-be-ignored words or positive priming for the to-be-attended words. Strictly, cognitive inhibition was not assessed in the David and Brown (2002) study.

In sum, while various theories depend on the notion of highs having a special ability to inhibit information, there are no convincing data indicating that highs are better than lows at cognitive inhibition. If cognitive inhibition is a stable ability as such, it should express itself on a range of relevant tasks. Therefore, we investigated the relation between hypnotic suggestibility and cognitive inhibition using three measures of inhibitory ability. First, we used the David and Brown (2002) task to determine whether it provides a measure of inhibition or positive priming and whether either such attentional measure correlates with hypnotic suggestibility. Second, we used a spatial negative priming task in which participants searching for a shape are slower to indicate the shapes location when it occurs in a location in which a to-be-ignored distracting shape occurred on the previous trial (see Tipper, 2001). We used this task because it was one of the earliest and is one of the most well investigated measures of negative priming. Third, we used a latent inhibition paradigm introduced by Lubow and Kaplan (1997) because it has been shown to relate to individual differences in expected ways (i.e. it correlated negatively with schizotypy; Lubow, Kaplan, & De la Casa, 2001, as predicted by the diminished latent inhibition sometimes shown in acute schizophrenia, e.g. Baruch, Hemsley, & Gray, 1988), and thus it itself is a meaningful measure of individual differences. In this paradigm, participants search for a fixed target shape amongst distractors all of the same shape. Latent inhibition is shown by participants later being slower when the previous distractor shape becomes a target shape, compared to when the target shape is either novel or remains the old target. This paradigm also allows a measure of individual differences in pre-attentive processing as shown by the novel pop out effect, i.e. the tendency for attention to be drawn to a novel item in a familiar background.

As well as performing the three inhibition tasks, all participants also completed the normalised dissociative experiences questionnaire of Wright and Loftus (1999). To prevent associations being inflated because participants knew the variables were being jointly investigated, contextual cues linking hypnotic suggestibility screening and measuring dissociation and inhibition were minimised. Council, Kirsch, and Hafner (1986) showed that participants aware that a study was aimed at investigating the relation between hypnosis and other variables produced stronger correlations than participants unaware of the connection between the measures. Thus, hypnotic suggestibility screening took place in one location and was presented as a study of hypnosis, and the other measures were taken in another location at a different time and presented as a study of processing speed.

Our main interest was the overall relation between hypnotic suggestibility and the other measures but we also tested correlations separately for each gender. We had no theoretical basis for expecting gender differences in the relation between hypnotic suggestibility and cognitive inhibition. However, Lubow et al. (2001) found that latent inhibition correlated with schizotypy only for females, indicating that the role of cognitive control in producing subjective experiences may be different for males and females. Relatedly, Townshend and Duka (2005) found the effect of alcohol on cognitive inhibition was different for females and males and Wang et al. (2009) also found evidence for different brain circuits for cognitive inhibition in men and women.