Executive control without conscious awareness:The cold control theory of hypnosis

Zoltán DienesDepartment of Psychology

University of Sussex

Josef PernerDepartments of Psychology

University of Salzburg

Running head: Cold control

Zoltán Dienes, Department of Psychology, School of Life Sciences, Sussex University, Brighton, BN1 9QG, UK. E-mail: dienes@biols.susx.ac.uk Tel: 00 44 1273 678550Fax: 00 44 1273 678058Home page: http://www.biols.susx.ac.uk/home/Zoltan_Dienes/

Josef Perner, Department of Psychology, University of Salzburg, Hellbrunnerstrasse 34, A-5020 Salzburg, Austriaemail: josef.perner@sbg.ac.atTel. 00 43 662 8044 5124Home page: http://www.sbg.ac.at/psy/people/perner_e.htm

1

Abstract

We argue for the theoretical possibility of unconscious executive control using the hierarchy of explicit knowledge developed in Dienes and Perner, (1999) and the Higher Order Thought (HOT) theory of Rosenthal (2002). Then we elaborate the cold control theory of hypnosis, according to which hypnotic response involves executive control without conscious intentions. Cold control theory allows an account of why some hypnotic tasks are more difficult than others, of individual differences in hypnotisability, and of the effects of expectation in hypnosis. More broadly, cold control theory is located in a framework allowing a number of theories of hypnotic responding (namely, empty heat and multiple selves).

2

Control and awareness intuitively seem intimately related (e.g. Norman & Shallice, 1986; Jacoby ,1991). Of course, some forms of control occur quite unconsciously (plausibly, for example, the detailed configuring of motor movements, Milner & Goodale, 1995). But there are some forms of control, like planning or overcoming strong response tendencies (the “Executive tasks” of Norman & Shallice) that are so commonly associated with conscious awareness that it would seem bizarre if they occurred without it. In fact, unconscious executive control is not possible in the theories of Norman and Shallice and Jacoby. In this article, we will argue for the theoretical possibility of unconscious executive control, based on the theories of Dienes and Perner (1999), Perner (1998, 2003), and Rosenthal (e.g.1986, 2002, 2004), and then argue that hypnosis provides an example of executive control without conscious awareness (cf Hilgard, 1977; Spanos, 1986; Oakley, 1999).

A fundamental explanatory problem in hypnosis is how activities that are normally performed voluntarily out of the hypnotic setting can be performed with the experience of involuntariness after hypnotic suggestion (see Lynn & Rhue, 1991; Fromm & Nash, 1992 for reviews). Of course, hypnotic phenomena present the researcher with many interesting problems to be explored, but a central if not defining issue is the experience of involuntariness, singled out as the “classical suggestion effect” by Weizenhoffer (1974). It is the fact that this experience of involuntariness can accompany the performance of otherwise mundane actions such as slowly raising one’s arm, holding one’s arm out straight and rigidly, acting like a child, and so on, when these actions are hypnotically suggested that makes the experience hypnotic rather than mundane. Other counter-intuitive hypnotic phenomena, like alterations in perception (positive and negative hallucinations), can be taken to be essentially examples of this process of creating the experience of involuntariness (cf Bentall, 1990; Frith, 1992), as we will explore. Hypnotic behaviour involves planning and strategy, and yet can be performed without conscious awareness of the contents of the plans, and without conscious awareness of any intentions to perform the behaviours (Hilgard, 1977; Sheehan & McConkey, 1982; Spanos, 1986; Oakley, 1999).

Kihlstrom (e.g. 1998) realized that understanding hypnosis required understanding the implicit-explicit distinction, and this distinction will in fact be our starting point. In this paper, we will first review the theory of Dienes and Perner (1999, 2002a,b, 2003; Perner & Dienes, 1999) describing a hierarchy of ways in which knowledge can be explicit. Then we will relate the different levels of explicitness to different types of control, following the ideas of Perner (1998, 2003), with executive control requiring the penultimate level of explicitness. The highest level of explicitness will be shown to be related to conscious awareness through Rosenthal’s (1986, 2002, 2004) higher order thought theory. The framework we develop will finally provide a number of ways of accounting for hypnotic phenomena, in particular, the experience of involuntariness.

A hierarchy of explicit knowledge

Dienes and Perner (1999, 2002a,b, 2003; Perner & Dienes, 1999) proposed an account of what it is to represent explicitly, and applied the notion of explicit representation to what it is to know a fact, thereby generating a hierarchy of ways in which knowledge can be explicit. The problem of what it is to represent, and hence what it is to represent explicitly, will not be discussed in detail in this paper (see the Dienes and Perner papers for discussion). But it will be necessary to indicate the sort of meaning of “explicit” we use. An English language representation like “The King

3

is dead” can be said to represent explicitly that the King is dead, whatever else the sentence might imply or presuppose (for example, that there existed a King, which is conveyed only implicitly). In the same way, mental representations have an explicit or literal representational content. By “explicit content” we do not mean, as a matter of definition, content that is conscious or reportable. We mean simply that the representation directly indicates some feature or states of affairs (just as, in some sense, “The King is dead” directly indicates that the King is dead). How explicit representation can in certain cases lead to conscious representation is an argument we develop, rather than assume.

Knowledge consists of having a mental state relation (specifically, one that implies knowing) towards a proposition. For example, I might see that the object in front of me is black. The proposition, the content of the knowledge, is that the object in front of me is black. The mental state relation by which I know is seeing, in this example. In order to know the content one does not have to explicitly represent that one knows the content; one could just explicitly represent the content. In fact, one might just explicitly represent part of the content of the full proposition that is known. Considering what minimal level of the content could be explicitly represented, and building up, explicitly representing more and more, until the whole state of affairs of knowing the proposition has been explicitly represented, provides the hierarchy of ways in which knowledge can be explicit.

Minimally, one might simply represent a feature of an object; for example, the feature “pencil” or “black”. After a person’s brief glimpse of some object, the only change in the person’s occurrent active representations might be the presence of an active “black” representation. Blackness has been explicitly distinguished from other colours, but that is all that has been made explicit. The person does not represent that they saw an object that was black, so they have no experience with the specific content “seeing that the object is black”. But the representation might enable the person to, for example, correctly guess what colour it had, especially if a forced choice decision is given. Correct colour guesses could even occur when the person is denying the presence of any object (cf Marcel, 1983). Such a representation constitutes a form of knowledge, but the knowledge is only minimally explicit (i.e. maximally implicit).

At the next level of the hierarchy, the person may explicitly represent that the feature is predicated of a particular individual (picked out by some referring expression like "the object in front of me"), explicitly representing the whole proposition: “The object in front of me is black”. At this stage, one can keep track of which individuals have which features (e.g. Bridgeman & Huemer, 1998).

At the third level of the hierarchy, one explicitly distinguishes the factual from the fictitious; i.e. one explicitly represents in which world the proposition is true (e.g. the real) in contrast to another world (e.g. the fictitious – a possible or even impossible world). At first blush, it may not appear obvious why one needs this stage of explicitness, but it is an important step. If a system cannot explicitly distinguish possibility from fact, any active representation would simply be taken to be true. And, indeed, one way of knowing a proposition p is simply to represent p and take it as true. But often when we as adult humans know p, we are simultaneously rejecting the possible match of not-p and reality; that is, we also know that not-p might have been true and is in fact false (O’Shaughnessy, 2000). Knowing by explicitly distinguishing fact vs possibility is when we know in a fact-explicit way, or “under the aspect of truth” as O’Shaughnessy puts it.

4

By having the capacity to label one model of an individual as possible or fictitious, the system can, without being deluded, consider models disengaged from reality (Perner, 1991). For example, in pretend play, a child pretending that a banana is a telephone considers one model of the object in which it is a banana and another model of the same object in which it is a telephone (Leslie, 1987). But the child is not confused or deluded; the child knows that one model is true only in a pretend or possible world, and thus that model must be explicitly labelled as such. Logically, the other model need not be labelled as true, it could just be taken as true. However, psychologically, it is highly likely that in setting up two contrasting models, one of which is labelled as fictitious, the other would, at least on occasion, be labelled as true; it is an ability that goes with the ability to label a model as fictitious.

Counterfactual reasoning also involves explicitly representing the possibility/factuality distinction. In the same way, the passage of time can be understood in terms of the same object in different models, one model true of the present, and the other (while possibly true) no longer true of the present; further, the space-time trajectory that actually happens is represented as the real one as opposed to others that might have been. The ability to represent factuality explicitly is an important development in the child, happening at about age 1.5 - 2 years old (Perner, 1991). For ease of expression we speak of fact-explicit for a representation of a proposition labelled as true in the real world, and of possibility-explicit when it is made clear that a proposition describes a mere possibility.

At the final level of the hierarchy, the mental attitude of knowing (seeing, etc) is explicitly represented: “I see that (in the real world) the object in front of me is black”. At this step, the knowledge is fully explicit knowledge; in previous steps, the knowledge was to some degree implicit. With fully explicit knowledge, the person knows that she knows.

We will now relate this hierarchy of explicitness of knowledge to different types of control, and then to conscious awareness.

Explicitness and control

Knowledge involves indicative representations; control in addition involves imperative representations. Indicative representations have the function of indicating a state of affairs; imperative representations have the function of bringing about a state of affairs, i.e. changing the world so that it conforms to how it is being represented (Millikan, 1984). For example, an active indicative representation of action A has the function of indicating that action A is occurring; but an active imperative representation of action A has the function of making action A happen.

As discussed above, one can explicitly mark the fact-possibility distinction for an indicative representation. Declarative knowledge is constituted by factuality-explicit (fact- or possibility-explicit) indicative representations: A declarative representation declares what is the case, i.e. they must represent the full proposition explicitly and typically that it is a fact rather than merely possible. Marking an active indicative representation as possible allows it to disengage from its normal function of indicating that something is the case; instead, it indicates that something is the case in a possible world. Similarly, one can explicitly mark the fact-possibility distinction for imperative representations, allowing them to disengage from their normal function. By marking an imperative representation as holding in a possible world, the active representation then does not immediately bring about its content.

5

Control can involve imperative representations like “If condition C then do action A!”. For example, you might say to yourself “If it is raining outside, I will bring an umbrella!” Though this representation contains active and occurrent representations of the condition C and the action A, C cannot be taken to be true (it is understood that it may or may not be true) and A is not necessarily to be immediately triggered. Thus, the imperative representation is possibility-explicit. Contrast a procedure or action schemata in which a C representation is linked to an A representation. Now when the C representation is active, it is taken to be true; it activates the A representation that immediately produces the action. That is, action schemata can be maximally implicit and are not in general possibility-explicit. Ultimately, any imperative possibility-explicit representation can only do anything by making contact with possibility-implicit action schemata.

Hilgard (1977) suggested a model of cognitive control in which action schemata (which he called cognitive control structures) compete amongst themselves such that the strongest at any given moment comes to control behaviour. An Executive Ego can override the strongest so that some other control structure actually controls behaviour. Hilgard presented this model as part of his neo-dissociation theory of hypnosis. Later, Norman and Shallice (1986) provided a very similar theory of cognitive control, motivated independently and without reference to hypnosis, that has come to be very influential, particularly in neuropsychology. They suggested that action schemata compete to control behaviour. The schema with the most activation is the one that wins. The level of activation of a schema is determined by the match of the schema’s trigger conditions with the conditions that actually obtain, and by the lateral excitation and inhibition between schemata (mutually incompatible schemata inhibit each other; co-operating schemata excite each other). This process by which a schema comes to be sufficiently active that it is the one that controls behaviour is called contention scheduling. In addition, there is a Supervisory Attentional System (SAS) that can send additional excitation or inhibition to a chosen schema, biasing its chances of winning. The SAS is attention demanding and is involved in conscious control, according to Norman and Shallice.

Norman and Shallice (1986) suggested five executive function tasks that the SAS was needed for: (1) planning or decision making; (2) troubleshooting; (3) learning new actions; (4) technically difficult actions; and (5) overcoming a strong pre-existing response. If contention scheduling were just left to itself, we would be entirely creatures of habit. If we always drive a certain route from home to work, that route is likely to be taken every time if contention scheduling were the only control process at work. But sometimes we can decide to do something new; for example, to make a detour at the traffic lights by turning left rather than right in order to buy milk at the supermarket. This new action requires the SAS (overcoming a strong pre-existing response). In general, the new action would only be accomplished if we were consciously aware of wanting to do it at the appropriate juncture. Hence, Norman and Shallice regarded the SAS as being intimately related to conscious awareness of what one is doing.

Perner (1998) pointed out that the two types of control (SAS and contention scheduling) relate to the implicit-explicit distinction. All the executive tasks (like following verbal instructions to implement new actions (3), or planning (1)) involve declarative or possibility-explicit representations of schema content “If C, do A!”. For example, technically difficult actions (4) presumably refers to strengthening difficult actions by declarative representations. Trouble-shooting (2) and overcoming responses (5) require one forms a declarative representation of what it is to be

6

inhibited as well as what it is to be achieved. All the executive function tasks involve possibility-explicit representations. In contrast, contention scheduling can simply use the possiblity-implicit representations constituting the action schema.

The SAS deals extensively in possibilities; it considers possible action rules (“Maybe ‘if C then A1’ but maybe ‘if C then A2’”) in exploring problem spaces (by considering the different possible outcomes of different actions) (Newell, 1990). Once the SAS has finalized on one action rule like ‘If C then A2’, the SAS uses a representation of the rule in which both C and A2 are just possibilities, even as the rule as a whole is meant to be actually implemented. The SAS might decide on a representation like “Do A!”. This representation does not apparently contain within it elements explicitly marked as possible; A is actually to be performed, full stop. But just as when we as adults know p, we also characteristically reject the possible match of not-p and reality, just so when we decide to do A by means of an imperative representation in the SAS (“Do A!”), we are characteristically simultaneously deciding to reject the possibility of not doing A. The SAS works with fact/possibility-explicit representations, and it is this constant dealing with possible actions that gives us freedom and makes us human (Dennett, 2003). We have a sense of free will partly because we represent that we could have done otherwise1.

In summary, there are two types of control distinguished by what is made explicit in the controlling representations. Executive control uses possibility-explicit representations; the content of the possiblity-explicit representation determines what schema comes to control behaviour. Contention scheduling uses representations that do not explicitly distinguish fact and possibility.

Explicitness and conscious awareness

We will now show the relation between explicitness and conscious awareness by use of Rosenthal’s (e.g. 1986, 2002) higher order thought theory. Rosenthal provided an account of when a mental state is conscious, e.g. when is seeing a case of conscious seeing and when is it unconscious? Blindsight patients can report highly accurately whether an object is moving up or down, even while they claim to have no visual experience whatsoever (Weiskrantz, 1986, 1997). Their accurate reports indicate they do see that, e.g., an object is moving up. But their verbal reports indicate that they do not consciously see that an object is moving up. The data indicate we need a distinction between seeing and consciously seeing, or more generally between being simply conscious of and being consciously aware of something (Carruthers, 2000).

A mental state (e.g. of seeing) makes us conscious of some state of affairs, in the minimal sense of “conscious of” that applies to the seeing that occurs in a blindsight patient’s blind field. In contrast, we are consciously aware of a state of affairs when we behold it with specifically a conscious mental state. Rosenthal argues that a mental state is a conscious mental state when we are conscious of being in that mental state. Consistently, it sounds bizarre to say we could consciously see but we were not conscious of seeing. In Rosenthal’s account, we are conscious of mental states by having thoughts about those states. A thought about being in a mental state is a second order thought, because it is a mental state about a mental state. For example, the first

1 Further the fact that the mental states that led to the intention are often themselves unconscious, leads us to experience our intentions as uncaused; hence, it seems to us we freely made the decision (Rosenthal, 2003).

7

order state could be seeing that “the object in front of me is black”. This makes us conscious of the object in front of me being black. The second order thought “I see that the object in front of me is black” makes us conscious of the first order state of seeing, and hence makes the seeing a conscious mental state. By virtue of the second order thought we consciously see that the object in front of me is black. In summary, according to higher order thought theory, a mental state is a conscious mental state when the person has a higher order thought to the effect that they are in that mental state (for elaboration see Rosenthal, 2002; for review, criticism and discussion of higher order theories of consciousness, see chapters in Gennaro, forthcoming).

The necessary and sufficient condition for conscious awareness in higher order thought theory is the presence of a suitable higher order thought (HOT). But a HOT about knowledge is just the most explicit level of the Dienes and Perner (1999) hierarchy of ways in which knowledge can be explicit. Conscious awareness corresponds to the highest level of explicitness, mental state explicitness. If a black object is exposed so briefly that only an active representation “black” is formed, the person is not consciously aware of seeing the black object. It is only when the representation, a HOT, “I see that the object in front of me is black” is formed that the person consciously sees that the object is black (Cheesman & Merikle, 1984; Merikle, 1992).

A second order thought (e.g. “I see that the cat is black”) makes one aware of the first order thought (“the cat is black”) and hence makes the first order thought a conscious thought. The second order thought itself is not a conscious thought until one becomes conscious of it - by a third order thought (“I know that I am seeing that the cat is black”). It is the third order thought that makes one consciously aware or introspectively aware that it is me who is seeing. We will make use of the distinction between second order and third order thoughts later when discussing hypnosis. Consider the intention “Lift the left arm!”. This is not a conscious intention unless there is a second order thought “I am intending to lift my left arm”. This second order thought makes one conscious of the intention, but not consciously aware of having an intention. To be consciously aware (or introspectively aware) of intending, there needs to be a third-order thought “I know that I am intending to lift my left arm”.

An important point to note is that we argued in the previous section that executive tasks required only possibility-explicit representations. But conscious awareness requires fully mental state explicit representations. The theory allows, counter-intuitively, executive function tasks entirely controlled by representations which do not provide conscious awareness. We call executive control without a HOT (without conscious intentions) “cold control”. In the next section we use these ideas to build up a hierarchy of voluntary action (Perner, 2003).

A hierarchy of voluntary action

Perner (2003) distinguished between a higher-level executive-control system, and a lower-level fact-implicit-control system, corresponding to the supervisory attentional system and contention scheduling, respectively. The higher system can trigger the lower system and also monitor its successful performance (cf Wegner, 1994). For example, the higher system might use a possibility-explicit imperative representation “When I finish the next paragraph, I will make some green tea”. This sets up the lower system (Gollwitzer, 1999). Indeed, seconds later, when one has just finished the paragraph, one might find oneself getting up to make the tea. However, other on-going contention scheduling processes that occur while finishing the

8

paragraph may have in the meantime activated other schemata, so tea making is not initiated. Monitoring by the higher-level system can correct these disturbances and ensure that tea making goes as planned. Either or both of these actions by the higher system (triggering or monitoring) can lead to the representation, the HOT, “I am intending this action”. Sense of agency comes from the second order thought “I am intending this action” and also the third order thought “I know I am intending this action”. The function of the second order thought of intending is (i) to track initial executive control and (ii) to track consistency of lower level control with any previous executive control that may have set the lower level control up.

We can now describe a hierarchy of voluntary action. At the bottom there is non-intentional action (Perner, 2003): Contention scheduling produces an action not set up by the higher system. Here contention scheduling produces an action the person would say they do not mean to do, i.e. action slips (Reason, 1979). For example, if one went upstairs to change one’s clothes, but ended up putting on one’s pyjamas (James, 1890), that would be a non-intentional action. In intentional action, the higher system sets up the lower system; when appropriate conditions are met, the schema executes the appropriate action. In this case, the person would say they did mean to perform the action.

Intentional action divides into two types (Perner, 2003). In absent-minded intentional action, the lower system runs unmonitored, producing actions appropriate to plan. For example, one decides to drive from home to work, and does so successfully, relying only on contention scheduling. One can say that is exactly what one intended. The action was performed absent-mindedly, but it was still intentional. In fully executive-controlled intentional action, the higher level triggers and continuously monitors execution of action schemas, over-riding them when necessary. For example, one could drive to work, and check at each stage that one is really taking the right route.

Fully executive-controlled intentional action itself in principle divides into two types: Either with or without HOTs. Executive control is unambiguously voluntary when accompanied by a HOT “I am intending this action”, and also a third-order thought “I know I am intending this action”, making the person consciously aware of the intention. Executive control without HOTs, cold control, allows the possibility of executive function tasks performed unconsciously, without a sense of agency or volition.

The every day use of the term “intention” does not clearly distinguish the first-order possibility-explicit imperative representation that controls the action schema (“Do X!”, “If C do A!”) as opposed to the higher order thoughts about that representation (e.g. “I am intending to do X”). For clarity, we use the term “intention” to refer to the first-order imperative representation (just as “seeing” refers to the first-order visual representation and not to the higher order thought that one is seeing). Thus, intentions are genuinely causal. The second order thought about intending has the function of tracking this causal process. The second order thought, as a representation, can misrepresent, and hence occasionally gets things wrong (cf Wegner, 2002)2, and this allows an explanation of hypnotic phenomena: the cold control theory of hypnosis, or executive control without accurate HOTs.

2 But that does not mean that conscious will is an illusion (contrast Wegner, 2002). Most of the time the higher order thoughts accurately track and make us aware of the underlying causal intentions (executive control) and/or the consistency of actions made by contention scheduling (interacting with the environment) with executive control (compare Wegner & Wheatley, 1999).

9

Cold control: Executive control without a HOT

The cold control theory of hypnosis states that successful response to hypnotic suggestions can be achieved by forming an intention (possibility-explicit imperative representation in the SAS) to perform the action or cognitive activity required, without forming the higher order thoughts about intending that action that would normally accompany the reflective performance of the action. The first part of the theory claims that hypnosis typically involves the SAS (i.e. executive control). We will first consider the evidence for this, and then consider the consequence of not forming suitable HOTs about intending.

Hypnotic suggestions can involve the subject engaging in executive function tasks. For example, a standard suggestion used in stage hypnosis, and that can be reproduced in the lab (Evans, 1980), is the suggestion to forget e.g. the number “four.” The subject will count e.g. “1, 2, 3, 5, 6” fingers on a hand. This must involve executive control (overcoming a strong pre-existing habit), but the person denies awareness of why they count unusual numbers of fingers on their hands. (According to the logic of Jacoby, 1991, the ability of the subject to exclude “four” from its habitual production implies conscious awareness of four; this is just what the subjects themselves deny having.) Sackheim, Nordlie, and Gur (1979) found that a highly hypnotizable subject given strong motivation instructions for blindness performed significantly below chance in reporting the emotion shown in photographed faces. Bertrand and Spanos (1985) gave subjects a list of three words in three different categories, and highly hypnotizable subjects, when suggested, could selectively forget one word from each category. Subjects recalled on a category-by-category basis, and must have inhibited the to-be-forgotten word when recalling each category. Spanos, Radtke, and Dubreuil (1982) found that highs suggested to forget certain words in any type of task given to them produced those words at a below baseline level in a word association test. This is executive control, because the existing associations that would be produced by contention scheduling must be suppressed. Similarly, Smith, Morton, and Oakley (unpublished, cited in Kopelman & Morton, 2001) gave subjects a free association task. Subjects were then given a hypnotic suggestion that they would forget the task just performed. When the free association task was repeated with the same words, the hypnotised subjects (unlike non-hypnotised subjects) gave different words as free associates than their previous productions, thereby over-riding their existing habits. Strikingly, Raz, Shapiro, Fan, and Posner (2002), Raz, Landzberg, Schweizer, Zephrani, Shapiro, Fan, and Posner (2003), and MacLeod and Sheehan (2003) found that (at least some) highs could eliminate the Stroop effect when given the hypnotic suggestion that they could not read the words. Remarkably, the habit of reading was apparently suppressed. Challenge suggestions also require executive control. In a challenge suggestion, the subject is asked to try to perform some action, like bending arm, while being told the arm is rigid and unbendable. People often respond to this suggestion to trying to contract both triceps and biceps simultaneously. But in general, contention scheduling ensures the smooth performance of actions by inhibiting contradictory actions, and so does not lead to a muscular stalemate.

In general, virtually any arbitrary behaviour can be hypnotically suggested despite the fact that such behaviour might be novel to the person, at least novel in context, and many hypnotic suggestions require the person to ignore some salient aspect of the situation (e.g. analgesia or amnesia suggestions). That is, at least many hypnotic responses are under executive control.

10

For most of us most of the time, it seems occurrent possibility-explicit representations automatically lead to appropriate HOTs. The difficulty of using possibility-explicit representations without HOTs is illustrated by Wegner’s (1994) task of asking people to not think of white bears for a specified time. People find this extraordinarily difficult. In this task, a possibility-explicit imperative representation is formed by the SAS “Do not produce representations of white bears”. This representation can be used to guide the lower system, and also monitor its success. But if a second order thought is automatically formed “I am intending not to produce representations of white bears”, the HOT about intending makes the content of the possibility-explicit representation, which includes the concept white bear, the content of a conscious mental state. That is, engaging in the task to not think of white bears, itself leads to the HOT that makes one consciously think of the concept of white bears.

It might be objected that the HOT of intending makes one consciously think of the concept of white bears, rather than to consciously think of white bears per se, and it is the latter that is demanded by Wegner’s task. That is, one need not fail the Wegner task simply by virtue of having a HOT of intending. We asked a class of undergraduate psychologists to “not think of the concept of white bears” for 30 seconds; one student out of 53 rated themselves as passing this task3. The straight Wegner task – of not thinking of white bears – entails that one avoid forming HOTs about other mental states, like images, associatively triggered by the concepts in the intention4. That is, either task illustrates that it is extraordinarily difficult to prevent conscious awareness of the concepts that figure in one’s occurrent possibility-explicit representations.

Maybe highly hypnotizable subjects are especially good at dissociating HOTs of intending from intentions. In support of this idea, Bowers and Woody (1996) found that after hypnosis, highs could NOT think of their favourite car for two minutes more effectively than lows, a finding replicated by King and Council (1998)5. Prima facie, the highs could engage in executive control without corresponding HOTs. If highs are especially good at not forming HOTs of intending about intentions, they should also be good at not thinking of the concept of their favourite car, a prediction yet to be tested as such. But the fact that highs can pass the forget-four task implies that highs need not become consciously aware of the labels for concepts that figure in their intentions, i.e. their intentions can indeed remain unconscious.

This raises the question of how the normal ascent from possibility-explicit representations to mental-state explicitness is prevented in responding to hypnotic suggestions. It is very difficult to dissociate HOTs from first order states. Consider HOTs of perception. It is very difficult to now form the HOT “I am seeing a pink elephant” when in fact you are not. Conversely, it is very difficult to look at this paper 3 Another group was asked to “not think of white bears” for 30 seconds; this yielded a success rate of 6 students out of 50. The difference in success rates between trying not to think of white bears and trying not to think of the concept of white bears was significant, χ2 = 4.15, p < .05. That is, at least some participants distinguish between these two tasks.4 Wegner (1994) postulates a monitoring process that constantly looks for mental contents indicative of failure of control, and it is the action of the monitor that leads to the dramatic failures of thought suppression. Consistently, we postulate that HOTs of intending are normally triggered by both the first order imperative representation initiating control and by the monitoring of its success or failure. If the monitor did its work unconsciously – without HOTs at all – attempts at (conscious) thought suppression would not fail.5 It may not be important that the difference between highs and lows occurred only after hypnotic induction. Woody, Bowers, and Oakman (1992) point out that in general highs can respond in the same way with or without hypnotic induction so long as they notice the contextual appropriateness to do so.

11

and not have the HOT “I am seeing the paper in front of me”. It may be easier to control HOTs of intention (rather than perception). Arguably, we often perform acts consistent with intentions without ever bothering to represent them as intentional. The link between intention and HOTs of intending may be somewhat weaker than the link between perception and HOTs of perceiving, allowing HOTs of intention to be relatively loosely triggered by relevant actions (Wegner and Wheatley, 1999; Wegner, 2002). This may be partly due the fact that the SAS can delegate control to the lower system. So when one does something, it is not easy to check whether it was intentional by checking the SAS for a description of the act. There may be none, because the lower system’s specific action was but one way of implementing the SAS’s more general intention in relation to the environmental flux of stimulation, e.g., switching gears in the flux of traffic. In any case, for simplicity we will presume that all that is required in hypnotic response is dissociating HOTs of intending from actual intentions.

How is ascent from fact-explicitness to mental-state explicitness prevented? According to HOT theory, HOTs are just thoughts and so their occurrence will be sensitive to the same influences as other thoughts. That is, consistent with socio-cognitive approaches to hypnosis (e.g. Spanos, 1986, 1991), formation of a HOT about intentions might be prevented by activation of beliefs and expectations inconsistent with it.

The expectation that the act will occur involuntarily might inhibit HOTs of intending from occurring. Similarly, a belief system about hypnosis producing involuntary actions might inhibit contradictory beliefs (like "I am intending this action"). For example, Lynn, Nash, Rhue, Frauman, and Sweeney (1984) showed how beliefs about ability to resist suggestions determined general ability to resist suggestion. Subjects informed in a lecture (before a hypnosis session) that a good hypnotic subject could or could not resist suggestions found that in the subsequent hypnosis session, they could or could not resist according to the information previously presented to them.

Kirsch and Lynn have especially emphasized the importance of expectation in hypnotic responding (e.g. Kirsch & Lynn, 1999). Kirsch (1985, 1991) argued that hypnotic experiences are the outcome of expectations, based on the following evidence. First, it is well known how generally responsive hypnotic subjects are to demand characteristics; for example, if subjects are told that good hypnotic subjects will extend a cataleptic arm when hypnotised, that is what highly hypnotised subjects will do (Orne, 1959). Second, for suggestions given with no induction, the correlation between expectation of response and response is 0.53 (Braffman & Kirsch, 1999). Third, post-induction, the expectation of number of suggestions that will be experienced correlates highly with number of hypnotic suggestions experienced, r = .64 (Council, Kirsch, & Hafner, 1986). Fourth, subjects pass more suggestions after an induction rather than without an induction only to the extent that they expect to (Braffman & Kirsch, 1999). And finally, subjects’ expectations can be experimentally enhanced with dramatic consequences. When a suggestion for seeing red was given, a hidden red light bulb switched on dimly. In later suggestions, no subject tested as a low (Wickless & Kirsch, 1989). In sum, expectations play an important role in hypnotic responding.

One general counter-argument against accounts of hypnosis that rely on expectation is that subjects are often surprised by their hypnotic responses (e.g. Zamansky & Clark, 1986). Does not surprise indicate something counter-expectational happened? The expectation of X can cause X to happen, but the extent

12

to which the expectation causes X (this causing reflecting physical properties of the representation of the expectation) need not match the amount of X expected (the content of the representation of expecting); that is, our expectations can be effective in a way that is personally surprising. For example, a dowser’s expectation that the rod will turn downwards makes it turn downwards, but an expectation that it will go down by a small amount might lead it going down to a large (and hence surprising) amount; or the expectation that it will go down at some time makes it go down at a particular time.

But there is a more powerful argument against hypnotic responsiveness being directly caused by expectations. Kallio and Revonsuo (2003) point out that in everyday life we can fully expect to, for example, see our keys where we left them on the table, but in clear viewing conditions this does not cause us to see our keys on the table if they are not there. Wagstaff, Toner, and Cole (2002) found that when non-hypnotised subjects were 100% confident they would not see something on a sheet of paper (all the previous pieces of paper had been blank), they all did still see the “8” that was on it. And surely evolution has led us to see what the data rather than our expectations indicate, when the data are clear. In everyday life, expecting to see our keys does not in itself make us see them when they are not there.

Cold control theory offers a solution. Expectations need only affect the formation or otherwise of higher order thoughts of intending. So in order to hallucinate keys we would need to have the intention to imagine keys being there. Expectation can lead us to not have HOTs about intending the imagery, but in order to hypnotically hallucinate, the intention to imagine has to be there for some other reason, i.e. that it fits in with other intentions, plans and strategies. Cold control theory does not need to postulate that expectations affect first order perceptual or other first order states; expectations need only affect higher order thoughts about intending. Thus, hallucinations and the other phenomena of hypnosis will only occur when they are strategically appropriate (White, 1941; Barber, 1969; Sarbin & Coe, 1972; Spanos, 1986), because relevant intentions will only be formed when strategically and contextually appropriate.

While expectations seem the most natural candidate for preventing the formation of HOTs of intending, they may not be the only effective means for preventing formation of HOTs about intention. Cold control theory would allow for any other mechanism by which HOTs are avoided, or even a special state in which HOTs can be readily avoided, though we are not personally partial to a state explanation of hypnosis (see Kirsch & Lynn, 1995, for the arguments against state theory; Kallio & Revonsuo, 2003, provide arguments sympathetic with state theory).

We will now consider whether cold control theory sheds any light on differences in difficulty in different hypnotic suggestions and also any light on individual differences in hypnotisability.

Why are some hypnotic suggestions easier than others? Hypnotic suggestions can be roughly divided into simple motor suggestions (“Your arm is becoming so light it is rising in the air”), challenge suggestions (“Try to bend your arm” in a rigid arm suggestion), and cognitive suggestions (amnesia, hallucination, the ‘forget four’ suggestion, etc). In general, more people can reliably pass motor suggestions (about 90% of people) than challenge suggestions (about 50% of people), and more people pass challenge suggestions than cognitive suggestions (from 50% to 10% of people, depending on the suggestion) (e.g. Hilgard, 1965; Perry, Nadon, & Button, 1992; Kallio & Ihamuotila, 1999). Cold control theory provides two ways of accounting for different degrees of item difficulty: first, different orders of higher order thoughts

13

being involved; and second, different degrees of effort involved in implementing first order intentions. We consider each in turn.1) The order of the thought: SOTs vs TOTs.

Rosenthal (e.g. 2004) suggested that most of the time when we have higher order thoughts we have simply second order thoughts. Relatively occasionally, when we introspect we have third order thoughts. And maybe fourth and fifth order thoughts can be left to the philosophers. Given there is some means by which accurate higher order thoughts can be prevented, it would make sense if accurate third-order thoughts (TOTs with the content: “I know I am intending to do X”) were easier to prevent than second-order thoughts (SOTs with the content: “I am intending to do X”). Given this plausible assumption, the individuals most skilled at preventing accurate HOTs (highly hypnotizable subjects) would be able to avoid both accurate TOTs and SOTs about intending; less skilled individuals (mediums and lows) may only be able to avoid accurate TOTs about intending.

Assume lows can only prevent accurate TOTs of intending but not SOTs. The second order thought (e.g. “I am intending to not say four”) is still there making them consciously think of the content of the relevant intention (“do not say four”), hence making them have a conscious thought about “four”. Even if they then had an inaccurate TOT (“I think I am not intending to not say four”), they may not be introspectively aware of intending to not say four, but they would still be thinking of four in a conscious mental state. So they could not do the ‘forget four’ task. But lows could do other tasks like arm levitation. They would have an unavoidable SOT (“I am intending to raise my arm”) making them consciously think of raising their arms, but by virtue of preventing accurate TOTs (“I know I am intending to raise my arm”) they would not be introspectively aware of intending the arm to rise, so the action would appear involuntary in a sense.

Only being able to avoid accurate TOTs for a particular suggestion would mean the subject might have both an accurate SOT about intending (“I am intending to lift my arm”) and a TOT that misrepresents that SOT (“I think I am not intending to lift my arm”). This would be a peculiar state and perhaps fragile. The TOT would go hand in hand with a disposition for the subject to verbally express it, perhaps by saying “I think I am not intending to lift my arm”. But given the performance conditions for saying that one thinks one is not intending are the same as for saying that one is not intending, the TOT may be also expressed as “I am not intending to lift my arm”, because of automatic linguistic habits in expressing TOTs (Rosenthal, 2004, chapter 10). Conversely, the SOT “I am intending to lift my arm” would go hand in hand with a disposition to unthinkingly express it as “I am intending to lift my arm” (this expression would have to be unthinking, because if one thought about it one would have an accurate TOT). So there would exist conflicting dispositions to say whether one intended to lift the arm. Demand characteristics and the subject’s motivation to fulfil them may then determine what is actually said, and whether the inaccurate TOT could be maintained.

A more convincing experience would require the avoidance of even accurate SOTs of intention. If one had simply the intention to move one’s hand, but no second order thoughts about having that intention, then there would be no verbal expression of the intention. If the person had an inaccurate SOT “I did not intend to move my hand”, that SOT would go hand in hand with verbally expressing it: “I did not intend to lift my hand”. If the person reflected on that statement, there would be a compatible TOT “I think I am not intending to lift my hand”. It would seem to the person that

14

they did not intend it. The TOT and the SOT would not produce conflicting verbal dispositions, and they would thus support rather than compete with each other.

Assume highly hypnotizable subjects can avoid even accurate SOTs of intending. Thus, they can do the ‘forget four’ task, because the second order thought about intention can be avoided. Further, when performing even motor suggestions they could avoid SOTs about the motor suggestion so they need not be consciously thinking about the action in any way. Zamansky and Clark (1986) asked high and low hypnotizable subjects to engage in imagery inconsistent with the hypnotic suggestions given (e.g. for a rigid arm suggestion, to imagine bending the arm). Highs were just as responsive to suggestions when engaged in imagery inconsistent with the suggestion as when having consistent imagery. In contrast, the performance of lows was severely degraded by contradictory imagery, strongly supporting the notion highs can suppress SOTs of intention, but lows only TOTs of intention. In order to implement the required executive control, lows need to be consciously thinking about the action to be performed. Highs do not need to be consciously thinking of the action to be performed. Similarly, Hargadon, Bowers, and Woody (1995) found that highs were just as responsive to an analgesia suggestion when involved in counter-pain imagery as in an image-less condition where imagery and even suggestion-related thoughts were proscribed. Highs do not need accurate SOTs to respond effectively to suggestions. These results directly falsify theories of hypnosis that postulate response is based simply on absorption in response-consistent thoughts and imagery (Arnold, 1946; Barber, Spanos, & Chaves, 1974; Baars, 1988, 1997) but corroborate cold control theory.

The ability to avoid accurate SOTs would facilitate the performance of any hypnotic task which needs to be performed with intentions whose contents remain unconscious. Consider, for example, the task used by Spanos et al (1982) in which subjects were to forget the use of a specific set of words in any context. One way to perform this task is to form an intention the content of which involves reference to the specific words that are to be forgotten. But the content would have to remain unconscious or else the words would be part of a conscious mental state and hence not forgotten. It is not obvious how else the task could be performed. It is an executive task so an intention must be formed; yet the intention should remain unconscious. Thus, similar cognitive tasks that involve not being consciously aware of some specific stimulus should be especially difficult, and more difficult than for example motor tasks where there is not a problem in being consciously aware of any concepts or stimuli in order to successfully respond.2) First-order effort.

Positive hallucination can be one of the more difficult hypnotic suggestions. According to Hilgard (1965), about 50% of people pass the taste hallucination (experiencing a sweet or sour taste in the mouth) and about 50% pass the mosquito hallucination (hearing or feeling a mosquito), but only about 10% of people hallucinate a voice. How can cold control theory account for hallucinations? Positive hallucinations could be produced by the executive-controlled production of relevant imagery; the lack of suitable HOTs about intending the imagery might lead the person to experience the image as a perception because the image is not experienced as intended (cf Bentall, 1990; Frith, 1992). There are several possibilities for why this might be a relatively difficult task.

First, to experience the image as a perception requires a HOT representing one as perceiving the target of the image. Such a HOT of perception may require

15

preventing any HOTs of imagining from occurring. Thus, hallucinations may involve preventing accurate SOTs of intending.

Second, it could be that successful positive hallucination involves suppressing the conscious perception of the real world where it contradicts the suggested world, and thus the suppression of accurate SOTs about e.g. seeing (for visual hallucinations) is also required. Such suppression might play a role in some cases, but in general highs readily report that their hallucinations are transparent, and hence the real world is still perceived (“trance logic”: Orne, 1959; Kihlstrom, 1985; Spanos, 1986). So this consideration does not explain the difficulty of positive hallucinations.

A third explanation is that there exists differences in ability to prevent accurate HOTs of intention depending on the amount of cognitive effort required in executing the first-order intention. Performing a simple motor action may be less cognitively demanding than forming an image. Thus, it may be more difficult to suppress higher order thoughts about forming images than performing motor actions. Images of tastes and simple noises (like the sound of a mosquito) might be easier to form than images of voices. Similarly, ignoring intensely painful stimuli, or not perceiving a stimulus that has been primed by instructions (as in a negative hallucination) may be especially demanding.

The greater skills of highs in suppressing HOTs may allow them to suppress HOTs even when large amounts of cognitive effort are required in implementing the first-order intention. This could be tested by acquiring effort ratings, or measuring interference of each task with, say, random number generation, for different types of tasks outside of the hypnotic setting and determining the correlation with assessed effort and difficulty of the corresponding hypnotic suggestion.

The idea that the more effort goes into performing a task the harder it is to suppress HOTs of intention may also explain why challenge suggestions are more difficult than simple motor suggestions. A simple motor suggestion that one’s arm is so light it is rising requires the effort needed to lift one’s arm. But the challenge to lift one’s arm – while being told the arm is so heavy that it cannot be lifted – involves both an attempt to lift the arm and an attempt to stop the lifting from happening. For some people this will involve some considerable muscular effort using antagonistic muscles to prevent the lift. The greater effort involved in successful responding to the challenge suggestion rather than the simple motor suggestion may be one reason why it is harder to suppress HOTs of intending with challenge rather than simple motor suggestions. We will consider another way of successfully responding to challenge suggestions later, one that involves no muscular effort at all.

There is one type of suggestion that appears to pose special difficulty for cold control theory: The hidden observer (Janet, 1889; Prince, 1909; Hilgard & Hilgard, 1975; Hilgard, 1977, 1992; Nogrady, McConkey, Laurence, & Perry, 1983; Spanos, 1986; Kirsch & Lynn, 1998). In a typical hidden observer suggestion, the subject may for example first be given an analgesia suggestion while their arm rests in ice-cold water. The subject is also told that there is a hidden part of them that really knows everything that is going on and when, e.g. their shoulder is touched, this “hidden observer” can talk to the experimenter. If a high hypnotizable subject is asked for pain report without the shoulder being touched, many highs will report virtually no pain. When the shoulder is touched, however, the “hidden observer” reports high pain ratings and may even report that it has felt this pain all along. Without the hand on the shoulder the subject reports low pain ratings, and may report that little pain had been felt all along. About 50% of highs can show a hidden observer effect (depending on task demands).

16

The problem the hidden observer poses for cold control theory is that the hidden observer can give pain ratings. Does this not contradict the notion that suggestion is the reduction of pain by executive control without HOTs? Using cold control, analgesia could be produced by setting up a possibility-explicit imperative representation “Do not attend to the noxious stimulus”, and not having a HOT about doing so. This could make it seem like one was not in pain, but such cold control does not entail any HOTs about being in pain anywhere in the mind. That is, cold control does not appear to explain the hidden observer. Two responses are possible: First, the hidden observer relies on a mechanism different from cold control (formation of different "I"s; this will be discussed below); and second, the hidden observer cue simply acts as a cue for the strategic reversal of the suggestion (see Spanos, 1986, and Kirsch & Lynn, 1998, for supportive discussion). According to the second account, when the shoulder is not touched, the subject uses cold control to reduce pain by the use of cognitive pain control strategies (Miller & Bowers, 1986); when the shoulder is touched the subject ceases to use the strategies and pain is experienced (Spanos, 1986). Because suggestions like analgesia (and others used in the hidden observer response) are difficult, obtaining a hidden observer response in subjects is difficult. In addition, responding with a hidden observer requires fairly quick changes in strategy, so a rapid flexibility in avoiding or enabling HOTs may be a desirable skill for a subject to show a hidden observer response. In addition, eliciting a hidden observer response in which the hidden observer claims it always felt pain implies forgetting that one was not feeling previously; such amnesia involves avoiding SOTs of remembering, and that is difficult.

In summary, cold control theory enables us to get a handle on the order of difficulty of different hypnotic suggestions in a non ad hoc way. Suggestions that require avoiding accurate SOTs of intending will be more difficult than suggestions that require only the avoidance of accurate TOTs of intending; and the more effort involved in performing the task, the harder it may be to avoid accurate higher order thoughts, so the more difficult the task will be as a hypnotic suggestion. Another consistent explanation is in terms of the skills required in implementing the first-order intention. Everyone can perform motor actions; not everyone has compelling vivid imagery. Indeed, imagery vividness is correlated with hypnotizability (Sheehan, 1982; Wilson & Barber, 1981; but contrast Jamieson & Sheehan, 2003).

We cannot claim to have explained the rank order of difficulty of all hypnotic responses (e.g. why do only about a third of people pass amnesia suggestions compared to 50% of people who pass the mosquito hallucination?), but cold control theory does provide a means for thinking about why some suggestions are harder than others within the context of a single theory, a single mechanism for producing hypnotic responses.

Difference between high and low hypnotisable subjects on non-hypnotic tasks. Cold control theory can generate predictions about how highs and lows may differ in various tasks outside of the hypnotic context. The fundamental skill postulated by cold control theory is dissociating HOTs of intention from intentions. So the ability to produce actions in any context that feel like they happen by themselves should be the main correlate of susceptibility. Indeed, the best correlate of hypnotisability is suggestibility without a hypnotic induction; the correlation goes from about .65 (Hilgard & Tart, 1966) to 0.85 (Barber & Glass, 1962).

Hypnotisability should also correlate with other sorts of tasks, as we will now argue. We will make the assumption that executive control is easier with relevant HOTs to support it; the HOTs can add to and support the controlling activation. Perhaps this is why HOTs so readily accompany executive control normally. Therefore, to be the sort of person who easily engages in executive control without

17

HOTs, it would help to be good at executive control. Thus, one might expect highs on average to be better than lows at executive control. Are they?

Finding correlates for hypnotisability is notoriously difficult, and such non-hypnotic measures that do correlate with hypnotisability do so only moderately, if not sporadically. Nonetheless, despite the chequered pattern, a useful broad-brush way of summarising many of the results is in terms of executive function ability (Crawford, Brown, & Moon, 1993).

Maintaining attention is an example of an executive function task, because it involves successfully overcoming distraction. One of the most frequently replicated correlates of hypnotisability (with an r of about .30) is self-reported absorption in imaginative activities, typically measured by Tellegen’s absorption questionnaire (Tellegen & Atkinson, 1974; see Roche & McConkey, 1990, for a review). The absorption questionnaire asks for example whether the subject becomes absorbed in seeing shapes in a fire or the clouds, or in reading a book. Similarly, in interviews J. Hilgard (1970, 1974), found that highly hypnotisable people, unlike lows, frequently became absorbed in imaginative experiences, for example, drama or listening to music. Wilson and Barber (1981) found that their group of very highly hypnotisable women (unlike the low hypnotisable women) had an ability to concentrate on imaginative fantasies to such an extent they could, for example, bring themselves to orgasm by fantasy alone: A fine demonstration of executive control.

Other self-report measures of ability to focus attention have also indicated differences between highs and lows. Crawford, Brown, and Moon (1993) and Lyons and Crawford (1997) distinguished between absorption, which they suggested involves loss of awareness of one’s surroundings, and a focussed attention where one is aware of one’s surroundings without being disturbed by them (e.g. concentrating easily on reading while in a noisy room). A questionnaire based on subjects’ rating how often they typically found themselves focussing attention in this way discriminated highly hypnotizable from low hypnotizable subjects (a finding replicated by Kallio, Revonsuo, Hämäläinen, Markela, and Gruzelier, 2001). Jamieson and Sheehan (2002) found a significant correlation between a similar questionnaire and hypnotisability. Van Nuys (1973) found that the number of intrusive thoughts that were reported by subjects during meditation was correlated negatively with hypnotisability (a finding replicated by e.g. Spanos, Rivers, & Gottlieb, 1978, and Spanos, Stam, Rivers, & Radtke, 1980; conversely, Jamieson & Sheehan, 2002, obtained a null result on a similar task looking at intrusive thoughts while listening to music).

Karlin (1979) also investigated ability to selectively attend to relevant information and ignore distractions. He found that highs as compared to lows could more successfully follow one of two stories spoken in the same voice. Subjects’ attentional skill was determined by summing their score on a recall test and their rating of their ease of understanding the target story. Karlin, Morgan, and Goldstein (1980) replicated and extended this finding by demonstrating successful response to a suggestion for hypnotic analgesia was related to performance on the attention task; Crawford, Brown, and Moon (1993) failed to replicate a difference between highs and lows using Karlin’s procedure (and cite two other unpublished failures to replicate). But whenever significant results are found for Karlin’s task, they are with highs outperforming lows. Karlin proposed that highs could plausibly use cues for determining the probability of words to bias perception of those words. This interpretation was supported by Dienes (1987), using an objective measure of performance, not relying on people’s ratings of how easy the task was. Dienes played

18

highs and lows two simultaneous lists of words defined by a theme (e.g. one list could be animal names and the other male first names); highs performed better than lows at shadowing the chosen list. That is, highs could select on the basis of representational content (semantic category) to a greater degree than lows. Selecting by representational content (content control) is a characteristic of executive control (Perner, 1998). There was no difference between highs and lows when shadowing could be based on a clear perceptual cue (male vs female voice), but there was a difference when the perceptual cue was less clear (two different male voices), indicating a superior ability of highs over lows to select under difficult conditions. Consistently, Stava and Jaffa (1988) found no relationship between an easy selection task (reading one story and ignoring a spoken one) and hypnotisability.

Naish (1985) also found that highs could bias perceptual schemata more strongly than lows. In a signal detection task (detecting 300ms tones amongst 360ms tones), when subjects were told the signal would be more likely, the change in bias was higher for highs than lows (and, based on verbal report, perception changed not just response tendency). Naish’s results were numerically replicated by Sigman, Phillips, and Clifford (1985; experiment 2).

Another measure of inhibiting habitual response is the Stroop test. While some studies have found no significant difference between highs and lows on Stroop interference (Kaiser, Barker, Haenschel, Baldeweg, & Gruzelier, 1997; Aikins & Ray, 2001; Kallio et al, 2001), Dixon, Brunet, and Laurence (1990) and Dixon and Laurence (1992) found highs showed greater Stroop interference than lows, apparently suggesting that highs have less executive control than lows. However, Dixon and colleagues argued that the greater Stroop effect for highs did not indicate a weakness of executive control, but a strength of automatic control. Crucially, Dixon & Laurence (1992) found that highs had greater strategic control over reversing the Stroop effect than lows. When a colour word systematically predicted an opposite colour patch, highs could reverse the Stroop effect (i.e. become faster to incongruent rather than congruent word- colour patch pairs) at shorter time intervals between the word and the colour patch than lows could (despite the fact that highs were overcoming a greater Stroop effect to begin with).

On another measure of inhibition of distracting information, David, King, and Borckardt (2001) and David and Brown (2002) investigated the relationship between negative priming and hypnotisability. In the negative priming task, subjects were asked to ignore or attend successive words in a list. Negative priming was shown by longer reaction times to later categorise the to-be-ignored words rather than the to-be-attended words. Across three studies, David and his colleagues found that individual differences in negative priming correlated positively with hypnotisability, providing further evidence for greater executive ability in highs rather than lows.

In terms of frontal measures of cognitive flexiblity, Kallio et al (2001) and Aikins and Ray (2001) found no difference between highs and lows on verbal fluency, but Aikins and Ray found that highs outperformed lows on the Wisconsin Card Sorting Test.

Negative evidence of a relationship between hypnotisability and frontal ability is hinted at by Woody and Farvolden (1998). No data or statistics are presented, but they report highs showed greater proactive interference and source amnesia than lows.

Baddeley (1986) took random number generation to be a task that loaded highly on executive function ability. For subjects to be successful at randomising their output, they need to inhibit habitual responses to avoid clichéd and hence non-random responding. Indeed, Graham and Evans (1977) found that highs were better

19

than lows at random number generation. Evans (1991) reports replications by three other teams of this relationship while Crawford, Brown, and Moon (1993) found no significant relationship between hypnotisability and random number generation, and cite three other null findings, but none with a significant difference in the opposite direction. In sum, significant results have been found on the random number generation task only in cases where highs outperform lows. Relatedly, the relation of hypnotisability to various creativity tests (e.g. P. G. Bowers, 1978; Shames & P. G. Bowers, 1992) may be based on the ability of highs to overcome habitual responses.

Highs appear to respond to visual illusions differently to lows; for example, highs report seeing more reversals of a Necker cube, more cases of autokinetic movement (illusory movement of a light in a dark environment), and a stronger Ponzo illusion (e.g. Miller, 1975; Wallace, Knight, & Garrett, 1976; Wallace, 1979; Crawford, Brown, & Moon, 1993; contrast Jamieson & Sheehan, 2002). While Miller, Wallace and Crawford et al. interpreted these findings as indicating greater attentional or executive skills for highs rather than lows, the findings are difficult to interpret without knowing what subjects were trying to do. If highs had good executive control skills, they should be able to e.g. maintain one perspective of the Necker cube for a long time if that is what they were trying to do, and conversely make the perspective change frequently if that was what they were trying to do. Explicitly manipulating instructions would be an interesting study for future research.

Finally, there is conflicting evidence whether highs can co-ordinate the processing of multiple simultaneous information sources better than lows. Sigman, Phillips, and Clifford (1985) found that highs compared to lows could correctly hear pairs of words played simultaneously. Conversely, Green and Lynn (1995) asked people to simultaneously perform a letter discrimination and number writing task; highs made more errors than lows. Stava and Jaffa (1988) found no relationship between hypnotisability and ability to simultaneously read one story and listen to another, or to simultanesouly listen to a story and copy digits.

In this area, one particular methodological artefact needs to be considered: To what extent do highs do well on tasks in general because they already know that they are highs and that the research has to do with hypnosis; or conversely, they score more highly on a hypnotisability measure because they knew they already did well on a task in the same context (Council, Kirsch & Hafner, 1986)? In other words, to what extent are the positive associations above due to hypnotisability and executive function task measurements taking place in the same context? The relationship with hypnotisability for a few executive measures has been determined when hypnotisability and executive function testing take place in different contexts, and in these cases that have been tested the relationship remains; namely, absorption (initially a mixture of null and positive findings, see Kirsch & Council,1992, for a review; see Crawford et al, 1993 and the meta-analysis by Lyons, 1992, cited therein, for the conclusion that there is a relation between absorption and hypnotisability in unrelated contexts; also Barnier & McConkey, 1999); other self report measures of ability to focus attention (Kirsch & Council, 1992; Crawford et al, 1993); and Necker cube reversals and autokinetic movement (Crawford et al, 1993).

In sum, the behavioural evidence tantalisingly suggests a relation between hypnotisability and executive control, consistent with cold control theory. Crawford and Gruzelier (1992) have argued that the neuropsychophysiological evidence taken as a whole also supports the association of good frontal (executive) abilities and hypnotisability. Key evidence for the claim comes from finding differences between highs and lows in EEG theta (3-7 Hz) power in frontal regions (amongst other locations) in various studies. Crawford and Gruzelier interpret the high theta in high

20

hypnotisable subjects as reflective of an ability to ignore environmental stimuli. More recently, Graffin, Ray, and Lundy (1995) found that highs compared to lows show more theta activity in frontal but not posterior areas, and also interpreted frontal theta as indicating “the continuous concentration of attention”.

Note that the relationship between executive control and hypnotisability is not strong; correlations as low as 0.2 are typical. Any theory that actually required highs to be strong on executive control would have difficulty explaining the weak relationships found. Highs are not highs because they can, for example, concentrate well. Cold control theory neeed not make such assumptions: Highs are highs because they can dissociate HOTs of intending from intentions and this does not demand being especially good at executive control. Being good at executive control is just a likely correlate of hypnotisability, because one can allow oneself to prevent relevant HOTs if one is good at executive control without HOTs.

The existence of different ways to respond to hypnotic suggestions. There is more than one way to successfully respond to hypnotic suggestions (McConkey & Barnier, in press). This point was brought out forcefully by Sheehan and McConkey (1982) who, after a hypnosis session, played subjects back video of the session and asked for commentary on what the subject was experiencing or thinking. Different cognitive styles emerged; for example, some subjects tended to concentrate on the literal message of the hypnotist, whereas other more independent subjects were willing to change the exact suggestion given to suit their preferences, for example hallucinating blue rather than the suggested orange. Subjects could also be more or less constructive in how much they embellished strategies to experience the suggested effect; for example, in order to achieve the hallucination suggested, one subject first visualised the object on her hand and then moved it to the required location. Subjects can also differ in how involuntary a response seems. These different cognitive styles are all consistent with cold control; subjects can vary in the exact intentions they formulate to achieve a suggested effect, and which and how many of these intentions they can act on while preventing appropriate HOTs of intending. The better subjects are at cold control, the greater the variety of intentions they should be able to implement without relevant HOTs; indeed, Sheehan and McConkey report that highs were more likely to report constructive (embellishing) styles of cognizing than lows.

As described above, cold control can be used preventing only TOTs of intending or preventing even SOTs. The distinction between the abilities to prevent TOTs vs SOTs may underlie the possibility of distinct types of subjects, rather than a simple continuum of hypnotisability (Oakman & Woody, 1996). Although cold control is a single theoretical mechanism, it is consistent with hypnotisability being multidimensional in various ways (cf. Woody, Drugovic, & Oakman, 1997). For example, subjects may vary not only in their ability to prevent SOTs or TOTs but also in how committed they are to being a good hypnotic subject; how willing they are to exhibit hypnotic behaviour given subjective experience of involuntariness that is compelling to different degrees; and subjects may vary in their capacities to implement relevant first order intentions (e.g. forming vivid images).

Summary. Cold control theory postulates that hypnotic responding is based on executive control (the SAS) without higher order thoughts about intending. It thus explains why many hypnotic responses can be executive tasks. It also gives us a handle on the order of difficulty of hypnotic suggestions (e.g. whether the suggestion requires SOT or just TOT avoidance), on individual differences in hypnotisability (e.g. the relation between executive ability and hypnotisability), and on why expectations seem to have much larger effects in a hypnotic rather than non-hypnotic

21

context. It also allows subjects different routes to achieving a given hypnotic response.

Other accounts suggested by HOT theory

Cold control theory postulates that hypnotic experiences arise due to intentions without HOTs about intending. But HOT theory itself can produce two other explanations of hypnosis. First, HOT theory proposes a distinction between HOTs and first order states, and the converse of first order states without HOTs (cold control theory) is HOTs without first order states (empty heat theory). Second, in HOT theory, mental states seem to belong to a person because the HOT represents that person – the “I” – as being in the state. But if somehow there could be multiple “I”s, multiple selves, one of those “I”s may not be aware of the experiences of the other Is (Kihlstrom, 1997). We will consider empty heat and multiple-selves theories in turn.

Empty heat: HOTs without first order states. According to Rosenthal (2000), one can have a second order thought that one is in a certain first order state, without actually being in that state. Rosenthal gives the example of someone going to the dentist, and they expect to feel pain. As the dentist applies the drill, the person hollers in pain. The dentist tells them they could not have felt any pain there because he personally had removed the relevant nerve previously. The person, somehow more correctly attunes their second order thoughts to first order mental states, and says, “You are right! I am just feeling vibrations!” Rosenthal argues that the person was initially in a first order mental state of feeling vibration, while having a second order thought about feeling pain. The person consciously felt the pain, because that was the content of the second order thought. Mistaken HOTs could produce many of the experiences brought about by hypnotic suggestion, in particular hallucinations, by representing one as being in a state one is not actually in. Woody and Szechtman (2000) propose an empty heat theory of hypnosis when they say “some of the subjective impressiveness and convincingness of hypnotically suggested phenomena, such as hallucinations, may be due to vivid feelings of knowing, rather than vivid sensory qualities per se” (p. 10).

Consistently, Naish (1983) found that in a signal detection task, highly hypnotisable subjects changed their bias to a greater extent than lows when told the signal was more likely; further, based on their verbal report, this resulted in a change in the subjective experience of highs. That is, highs were more likely to be influenced by expectations to see a target when it wasn’t there. Maybe highly hypnotisable subjects are just those people who can produce HOTs of perception in the absence of first-order perceptual states.

If hypnotic suggestions operate directly on HOTs, then hypnotic hallucinations would not involve actual first order perceptual states. So hypnotic hallucinations would not, for example, facilitate implicit perceptual tasks, or involve the use of visual pathways in the brain, or at least they would involve only areas concerned with higher order thoughts of perception. On the other hand, if, as cold control theory postulates, hypnotic hallucinations operate via preventing HOTs about intentions to imagine, hypnotic hallucinations would involve brain pathways involved in perception to the extent that those pathways are used by the imagination. Kosslyn and Thompson (2003) provided a meta-analysis of PET, fMRI and SPECT studies of visual imagery, showing that imagery can activate early visual cortex, as early as V1 or V2. Conversely, Weiskrantz (1997) rules out V1-V5 as supporting higher order thoughts about seeing (on the grounds that removal of non-visual cortex can leave an

22

animal completely blind; thus, the mechanisms for generating HOTs about seeing must lie outside of the visual cortex6). If we accept both these claims, finding a context in which hypnotic hallucinations causes activity in primary visual cortex would support cold control theory but falsify empty heat theory in that context. This is the sort of logic we now pursue in two relevant example cases.

Kosslyn et al (2000) asked highly hypnotisable subjects to either see a coloured pattern in colour, or to see a grey-scale pattern in colour. PET scanning indicated that the left and right fusiform areas were active in highs either seeing genuine colour or hallucinating colour, but not when veridically seeing grey-scale. To be consistent with empty heat theory, the fusiform area would need to be responsible for the formation of higher order thoughts of seeing. However, Dehaene et al (2001) found that both conscious and unconscious (masked) words produced activity in the fusiform area, so activity in this area is not sufficient for higher order thoughts of perception. Similarly, Driver, Vuilleumier, Eimer, and Rees (2001) reported activation in the fusiform area for extinguished (i.e. not consciously seen) visual stimuli. That is, empty heat theory is falsified as an account of the colour hallucinations reported in Kosslyn et al7. On the other hand, for cold control theory to account for the results, imagination would need to be sufficient to induce activity in the fusiform areas. Indeed, when subjects were instructed without hypnosis to “remember and visualise” the colour pattern, the same degree of activation of the right fusiform region was found as when subjects were hallucinating. But the “remember and visualise” instructions led to less activation in the left fusiform region than when hallucinating, challenging cold control theory. In interpreting the latter result, however, one should bear in mind Kosslyn et al’s concern that the subjects did not “drift into hypnosis” and hallucinate in the “remember and visualise” condition. The wording was chosen to “lead the subjects to attend to the visible stimulus and alter it rather than to substitute a complete hallucination”. That is, the demand characteristics entailed forming a less convincing image in the “remember and visualise” condition rather than the hallucinate condition. It is thus not surprising that this was reflected in less relevant activity in the fusiform area for the “remember and visualise” condition than the hallucination condition. Cold control theory predicts that if subjects capable of producing activation in the left fusiform gyrus when hypnotically hallucinating colour are tested out of hypnosis, activation will be produced in the left fusiform gyrus when subjects are asked to intentionally produce the same vivid experience as when hallucinating – but these results are not yet in.

A more ambiguous conclusion follows from the next example, involving the hypnotically modulated experiences of pain. When a painful thermal stimulus was applied, Hofbauer, Rainville, Duncan, and Bushnell (2001) found that hypnotic suggestions for increased sensory pain compared to suggestions for analgesia resulted in greater activity in the somatosensory cortex, S1 in particular. This would fit empty heat theory if S1 were the area that produced higher order thoughts about being in pain. Indeed, damage to S1 can lead to “numbsense”, i.e. sensitivity to tactile sensory stimuli without conscious awareness (Rossetti & Rode, 2001), consistent with S1

6 It might be thought that an alternative explanation for such lesions removing conscious vision is that when a first-order representation is not adequately connected to its consumers, it ceases to be a representation at all (cf Millikan, 1984). However, from the point of view of HOT theory, the malformedness of the first order representation is only relevant to impairing conscious seeing to the extent that it influences the ability of the first order state to trigger higher order thoughts.7 Though ideally one would want to have evidence of the role of the fusiform area in the unconscious perception of specifically colour.

23

being the site that produces higher order thoughts about tactile sensations8. However, damage to V1 produces blindsight, but it is unlikely that V1 is the area that produces higher order thoughts about seeing (see Weiskrantz, 1997; Rees, Kreiman, & Koch, 2002, for arguments). Thus, the question of whether empty heat theory is the best theory in this case remains open. Hofbauer et al’s results are also consistent with cold control theory. Directing attention away from a painful stimulus reduces S1 pain-evoked activity (Bushnell et al, 1999). That is, attending towards or away from the stimulus, without being aware of intending to do so, could produce the hypnotic modulations in felt pain and also in S1 activity.  Empty heat theory could in principle apply to challenge suggestions, like being asked to try to bend one’s arm, while being told one’s arm is as rigid as an iron bar. In this case, to pass the suggestion, the subject must fail to move. On cold control theory the subject might intend “Go through the motions of trying to move but do not move” while being unaware of intending this. Empty heat theory offers an alternative strategy. The subject does not intend to move at all, but forms the second order thought (SOT) (with no first order intention) “I am intending to move” and/or the third order thought (TOT) “I think I am intending to move”. But without the intention to move, the subject will not move (even while the subject believes that he or she is trying), so the suggestion is passed. With the cold control strategy, muscular effort is exerted in both trying to move and in resisting that attempt. With the empty heat strategy, no muscular effort need be exerted at all. We do not know of systematic data on this, but we anecdotally often observe subjects exerting effort in successfully responding to challenge suggestions; at least for those subjects, cold control theory provides the better explanation.

Empty heat theory could explain the positive relation between expectation and response to suggestion reviewed in relation to cold control theory. Mistaken second order thoughts should be most likely to occur if one strongly expects the first order state to occur, given that second order thoughts are after all just thoughts, and so should be responsive to the same provocations as other thoughts.

Multiple selves. Kihlstrom (1997) suggested that hypnotic subjects could create an additional "Hypnotic I". Because the hypnotic I's intentions (causes of hypnotic responding) are not linked to the normal "I" the person does not experience himself as intending the actions to occur. This theory corresponds to Hilgard’s (1977, 1986) neo-dissociation theory in that Hilgard postulated that in hypnosis the Executive Ego was split in two so that there are two conscious streams, one controlling the hypnotic responses, and the other unaware of this control.

How could separate “I”s come about? Rosenthal (2003) shows how HOT theory accounts for our normal sense of the unity of consciousness. There are a number of reasons why HOTs normally lead to a sense of unity. Whenever we have a HOT it represents oneself as being in a certain state. That is, it uses a mental token of

8 Though we do not know whether damage to S1 can lead to “numbpain”, i.e. first order qualitative states representing specifically noxious stimuli, in the absence of awareness of being in such states. Rees, Kreiman, & Koch (2002) argued that visual awareness depends on prefrontal and parietal cortex rather than just specifically visual cortical areas and Rees (2001) also speculates conscious awareness depends on dorsal frontoparietal cortex in a stimulus independent way. While we can only speculate at this time, our best guess would be that higher order thoughts about pain are not produced in S1, but probably do involve the prefrontal cortex. Consistently, Szechtman, Woody, Bowers, and Nahmias (1998) found that the brain region that distinguishes auditory hallucination and actual listening on the one hand, and imagination on the other, lies not in the auditory cortex but in the anterior cingulate; the latter region rather than the former is therefore plausibly involved in forming HOTs of actually hearing.

24

“I” to refer to the individual that is thinking the HOT to be the same individual that is in the first order state targeted by the HOT. The HOT makes us conscious of not only the target state, but also the self that has it. But how can we know whether the thinker of different successive HOTs is the same individual, the same self? A default assumption might be that in the absence of any reason to distinguish the selves, they can be taken to be the same self. But we can also identify our self in more informative ways, by using a battery of ragbag properties; we can represent our self as being the person that has certain memories, personality attributes, bodily characteristics, types of views, and so on.

In order to have multiple “I”s, we need a different battery of properties to identify the different selves (Rosenthal, 2003). As the set of memories we have is one property by which we identify ourselves, precluding memories of the experiences of the different selves would help ensure the different selves remain different. This is the process that Hilgard refers to as setting up an “amnesic barrier”. In hypnosis, one self believes it will passively do whatever is suggested; another self intentionally can make things happen. When a second order thought that one intended an action is formed, this HOT in itself simply refers to the self that has the first order intention as being the same self that is having the HOT. This does not identify the self in an informative way per se. When the person is having HOTs identifying themselves as the self who is passively responding to hypnotic suggestions, the self of the SOT of intending will not be identified as the same self. In fact, the SOT can be forgotten whenever one has a HOT about the set of states that go together to make the current self. But one can also have HOTs about the sets of states that go together to make the person who actively causes what the hypnotist wants, and the self of the SOT will be identified with this self.

On this account, the subject does not have to avoid the formation of accurate HOTs of intending or perceiving in order to respond successfully to hypnotic suggestion; instead the subject groups together different states, and uses these to identify different selves. Accurate HOTs of intending or perceiving belong to only one of these selves, so the other self experiences actions as merely happenings. The prediction is that it should be possible to talk to the self that owns the HOTs of intending or perceiving and so does know what is going on really. This is exactly what Hilgard (1977) observed, a phenomenon he dubbed the “hidden observer”. Hilgard found that, for example, when a subject seemed to be successfully responding to a hypnotic deafness suggestion, if the subject was told there was a part of him that really could hear all the sounds that went on, the subject was then able to report what the sounds and conversations were, even though he previously reported being unaware of the very same sounds and conversations.

As reviewed in the section on cold control theory, the notion of a hidden observer has generated a lot of research and controversy (see e.g. Kirsch & Lynn, 1998, for an overview, and comments by Kihlstrom, 1998, and Woody & Sadler, 1998). Spanos, Gwynn, and Stam (1983) argued that the hidden observer was an artefact of suggestion; of course the subject will behave in an appropriate way if you say what the demands are. For example, when subjects undergoing painful stimulation relieved by hypnotic analgesia were told that there was a hidden part of them that knew what was going on really, that part gave higher pain ratings when it was contacted. But when subjects were told there was a hidden part of them that was so hidden it would feel even less pain, that part gave even lower pain ratings when contacted. These results do not exactly contradict Hilgard’s neo-dissociation theory or the multiple-selves story sketched above (why not set up a third self with whatever

25

characteristics are suggested). But they do indicate how responsive highs are to demand characteristics, and highlight how easy it would be to confirm a theory by simply suggesting to subjects whatever the theory requires. Nogrady et al (1983) found that subjects still gave duality responses (e.g. in age regression, feeling that one was both a child and an adult at the same time) when there were low demand characteristics for multiple selves. But could such claims simply reflect incomplete responding, i.e. experiences that are not entirely convincing? For example, they might come about not because of multiple selves, but because of incompatible second and third order thoughts, and different reporting instructions (e.g. hidden observer vs standard) favour verbally expressing to the hypnotist or experimenter one HOT rather than the other. However, in contexts in which hidden observer responses occur more often in highs rather than mediums, the “incomplete responding” explanation is unlikely (Nogrady et al, 1983; Kirsch & Lynn, 1998).

While it may be open how best to interpret the processes producing hidden observer responses, one fact about hidden observer responses seems uncontroversial enough: People often pass all sorts of hypnotic suggestions without demonstrating a hidden observer on request. While the level of hidden observer responding can vary dramatically according to the overt demands for it and the type of suggestion (see Kirsch & Lynn, 1998, for a review), in many studies about 50% of highs show hidden observer responding. That would be roughly 5% of the population, whereas a majority of people are responsive to at least some hypnotic suggestions. Bowers (1992) and Kirsch and Lynn argue that one cannot use phenomena so rare as hypnotic amnesia or the hidden observer to support the notion of multiple selves (separated by amnesic barriers) as an explanation of hypnotic responding in general.

Are hypnotised subjects genuine?

Cold control theory rests on the evidence of verbal reports of highly hypnotisable subjects. The claim that executive tasks can be performed without conscious awareness is based on the fact that, for example, a hypnotised subject might consistently fail to say “four” while denying knowing why their counting is going wrong. In one sense, the simplest explanation of the reports of hypnotised subjects is that they are faking, merely complying with demand characteristics in an insincere way. While the vast majority of researchers in hypnosis believe most subjects have genuine experiences, Wagstaff (1981) argued that compliance (i.e. faking) might still account for at least a majority of hypnotic responding. In this section we review a couple of lines of evidence indicating we should take seriously what hypnotised subjects say about the content of their conscious experiences.

Presumably if a person is insincerely complying, a convincing performance will be put on only if there is a second person to be insincere towards. Kirsch et al (1989) introduced the surreptitious observation design in order to determine the extent to which such compliance happens in highly hypnotisable subjects. Highly hypnotisable subjects and lows who had been instructed to simulate being highs were played a tape of hypnotic suggestions. They were told that the purpose of the experiment was to see if hypnotic responsiveness was influenced by the presence or absence of an observer. Half of the subjects were watched by the experimenter; the other half were left by themselves to respond and record their own responsiveness. Of course, even when subjects were apparently alone they were being watched. Kirsch et al determined whether subjects believed the manipulation by seeing what the simulators did. If being watched vs being left alone created different demand characteristics, this should

26

be reflected in the simulators’ behaviour. Indeed, there was a massive effect on the performance of simulators; they performed near ceiling when watched and near floor when left alone. Crucially, highs were unaffected by the manipulation. Their responsiveness remained near ceiling whether being watched or left alone. These results were replicated by Spanos et al (1993) and also by Perugini et al (1998) using difficult hallucination suggestions. In sum, the surreptitious observation studies suggest that virtually none of the responses of highs occurred because of faking. Highs consistently respond to hypnotic suggestions presumably because they are motivated to have the experiences it seems to them that they really have.

Another strategy for determining whether highs deliberately fake hypnotic responses was employed by Kinnunen, Zamansky, and Block (1994). These authors first showed that they could use skin conductance responses (SCRs) to determine when people were lying in a controlled situation (people were asked to select one of a number of items, but to lie about having selected that item). Highs (“reals”) and simulators were taken through a number of hypnotic suggestions and asked about the genuineness of their experience (e.g. “Were you really unable to bend your arm?” “Do you remember what was on the card?” etc). The SCRs for highs for these questions were not significantly larger than for neutral questions (in fact, they were less than for neutral questions); whereas for simulators, the SCRs to these questions were significantly larger than for neutral questions. That is, simulators showed the physiological response expected if they were lying, but reals did not. Kinnunen, Zamansky, and Nordstrom (2001) also found that highs met the SCR criterion of truthfulness in reporting the genuineness of their experience, but subjects urged to comply did not.

On a final point, we ultimately have no means other than verbal report to determine the contents of specifically conscious mental states. While verbal reports are highly fallible means of determining the content of mental states in general, it is widely regarded as almost self-evident that the content of any conscious state we are in that is intentional (in the sense of propositional) and conceptual can be expressed verbally (see Dienes, in press, Rosenthal, 2004, for recent related discussion; cf Sheehan & McConkey, 1982; McConkey, 1991).

Comparison with other theories.

Cold control theory’s emphasis on executive function brings it in line with the two most prominent theories of hypnosis in the 1970s and 1980s, namely Hilgard’s neo-dissociation theory and the socio-cognitive approach, the latter argued for vigorously by Spanos amongst others. The 1990s saw theories emerge in which hypnotic responding involved contention scheduling rather than executive control (e.g. Woody and Bowers, 1996; Kirsch & Lynn, 1997). We briefly compare cold control with these different theories, as well as with recent theories in which executive function is important (Oakley, 1996; Crawford & Gruzelier, e.g. 1992).

In Hilgard’s neo-dissociation theory (e.g. 1977, 1986, 1992), the "executive ego" (SAS) was postulated as being the cause of any hypnotic response, consistent with cold control theory. The potential incompatibility between the theories is whether there is some part of the person that is aware of the first order states the hypnotized person denies having. According to cold control theory, when a highly hypnotisable person produces, for example, hypnotic analgesia, there simply are no higher order thoughts about the pain (nor HOTs about intending to engage in cognitive strategies to relieve the pain). According to Hilgard, the monitoring and executive functions of the Executive Ego (SAS) are split into two. “The two parts differ only

27

in that they are separated by an amnesic barrier” (Hilgard, 1986, p. 234). This might imply that the two parts are quite capable of their own higher order thoughts, and this makes Hilgard’s theory a multiple-selves theory (Kihlstrom, 1997), different from cold control. However, Hilgard (1986) also said that there are “two experiences going on simultaneously; of one the subject is aware, of the other he is unaware” (pp 236). The streams are also referred to as subconscious vs conscious. There are two interpretations of these phrasings. On one, the stream about which one lacks awareness, the sub-conscious stream, is a stream of first order states for which there are no accurate higher order thoughts. This is the “no HOTs” interpretation. On the other interpretation, the lack of awareness refers only to the lack of awareness of the stream by one of the selves; it is the hypnotised self that is unaware, though another self is aware of the first order contents.

The “no HOTs” interpretation makes neo-dissociation theory compatible with cold control. Many suggestions could be carried out by forming intentions, but failing to be aware of those intentions, i.e. by cold control (cf footnote 5 in Kihlstrom, 1998). Dissociative responses could also come about in negative hallucinations by having only first order states of perceiving in the absence of accurate higher order thoughts of perception, a perceptual analogue of cold control (i.e. cold perception). Kihlstrom (1998) prima facie accepted the “no HOTs” interpretation of neo-dissociation theory in reviewing implicit-explicit distinctions generally (e.g. in blindsight) as supporting evidence for Hilgard’s theory. Blindsight consists of first order visual states without higher order thoughts of seeing; there is no evidence that blindsight involves multiple selves (Kirsch & Lynn, 1998).

The prime evidence for neo-dissociation theory is the hidden observer, and the hidden observer can express higher order thoughts. This strongly implies that Hilgard did not intend the “no-HOTs” interpretation of his theory. Cold control theory and neodissociation theory are then clearly different theories. But they have in common the postulate that hypnotic responding involves executive functions.

The other great strand of theorising about hypnosis has been broadly called the social psychological (e.g. Sarbin & Coe, 1972; Spanos, 1986) or sociocognitive (Spanos, 1991) perspective. These approaches view hypnotic behaviour as fundamentally similar to other more mundane forms of social behaviour, behaviour to be explained by personal (rather than sub-personal) level states like expectations, beliefs, purposes, and attributions. These approaches typically date their intellectual pedigree as starting with the work of White (1941), who famously characterised hypnotic responding in terms of ‘goal-directed striving’: “Hypnotic behaviour is meaningful goal-directed striving, its most general goal being to behave like a hypnotised person as this is continuously defined by the operator and understood by the subject” (p. 483). The body of empirical work showing that hypnotic responses are indeed contextually appropriate, flexible, planned and goal-directed (e.g. Spanos, 1986) inspired a central notion in cold control theory, namely that executive functions are involved (consistent also with Hilgard’s neo-dissociation theory). Cold control can also be simply described at the personal level, i.e. as the use of intentions without awareness of having those intentions, a description entirely consistent with the social psychological approach (and also, on a certain reading, with neo-dissociation theory). Where cold control theory goes beyond social psychological approaches is in a specific commitment to higher order thought theory (Rosenthal, 1986); and social psychological approaches go beyond cold control theory in having a specific commitment to there being no special state of hypnosis, with garden variety kinds of expectations and beliefs being the important cognitive variables that lead people to make an attribution error regarding the causes of their behaviour. We have seen above how a commitment to higher order thought theory leads to predictions concerning e.g. the order of difficulty of hypnotic suggestions, and also to considering brain imaging to distinguish cold control and empty heat theories. Approaches - like the social psychological and cold control -

28

need not be inconsistent to inspire different research questions, different types of answers, and different agendas.

Norman and Shallice’s (1986) influential notion that the supervisory attentional system was responsible for behaviour that feels voluntary and contention scheduling for behaviour that feels automatic motivated two key theories in the 1990s: That of Bowers and Woody (Bowers, 1992; Woody & Bowers, 1994), who saw their theory as a development of Hilgard’s neodissociation theory; and that of Kirsch and Lynn (1997), who saw their theory as a development of the socio-cognitive approach.

Bowers and Woody (1994) described hypnosis as a weakening of frontal lobe function so that contention scheduling could control behaviour (hence the feeling of involuntariness). In dissociation terms, the dissociative split did not render the supervisory attention system in two (as in Hilgard’s theory, or a version of it), it split one or more action schemata from the supervisory attentional system, so that the schemata could become directly triggered by hypnotic suggestion. Thus, they call their theory the dissociated control theory. However, uniquely associating actions experienced as involuntary with contention scheduling creates a problem. Hypnotic responding cannot be based simply upon contention scheduling, because hypnotic responses can involve performing executive function tasks, as reviewed above. Our primary criticism of dissociated control theory, and its primary difference from cold control theory, is that the theory fails to get to grips with the highly strategic and, when necessary, executive nature of hypnotic responding. In cold control theory, like dissociated control theory, control is split off from consciousness, but the supervisory attentional system is still involved.

Bowers and Woody (1996) used Wegner’s white bear task to provide support for their theory. They argued that the absence of the ironic conscious awareness of bears (or cars, in their study) indicated that the highly hypnotisable subjects simply did not have the intention to not think about bears. How could this be possible? Highs could simply forget they were asked to think about bears. It would be as if no one had ever mentioned bears. The task would then be easy. However, this explanation fails to relate the white bear task to very similar seeming tasks, like the forget-4 task or analgesia. In these latter tasks, the subject has to also not think about certain things, and the good performance of highs on all these tasks is plausibly explained by appealing to one method for performing them all. The forget-4 and analgesia tasks cannot be performed by simply forgetting what the instructions were; in each case, the material to-be-ignored is being cued to be attended to all the time. There needs to be something actively done to stop the material from entering consciousness. Since this cannot be done by contention scheduling, we argue it is done by an intention in the SAS. If it is accepted that the white bear task is performed intentionally, then a failure to become consciously aware of the concept of white bear on the task must involve the avoidance of higher order thoughts about intending not to think about white bears. In any case, the success of highs on the white bear task does not necessitate the conclusion that the task is performed non-intentionally.

A further difference between cold control and dissociated control theories is that dissociated control theory is committed to there being a special state of hypnosis, specifically a state in which frontal lobe function is weakened. Consistent with dissociated control theory, Jamieson and Sheehan (in press) found that hypnotic induction impaired Stroop performance for highs (but not lows) (cf Kaiser et al, 1997). However, the Stroop impairment for highs may have resulted from their expectations regarding hypnosis (e.g. as a state in which one is drowsy) as suggested by the findings by Raz et al. Consistent with the idea that highs in hypnosis actually have good executive abilities which can be used when expectations allow, when highs were given hypnotic suggestions to eliminate the Stroop effect altogether, remarkably they could do this (Raz et al, 2002, 2003). And

29

inconsistent with the notion that hypnotic induction weakens frontal abilities in highs, Dienes (1987) found no effect of hypnosis on the ability of highs to perform difficult selective attention tasks. Similarly, Woody & Farvolden (1998) did not find any effect of hypnosis on the tendency of highs to have source amnesia or to be affected by proactive interference. Kallio et al (2001) also found no specific effect of hypnosis on highs for Stroop, vigilance, simple and choice reaction time, and category fluency, but a significant effect for letter fluency. Gruzelier and Warren (1993) had also previously found hypnotically produced reductions in letter fluency for highs. How highs under hypnosis would perform on letter fluency when they are lead to expect that hypnosis would improve performance on the task remains an open question. In sum, however we explain the letter fluency results, there is not strong evidence that highs are frontally impaired in hypnosis. If they were too severely frontally impaired, they would not be able to respond to the more demanding suggestions requiring executive involvement9.

Below we suggest that the same phenomenon we call hypnosis also widely manifests itself as spirit possession in other contexts. In the case of spirit possession, there is not in general the cultural belief that in being possessed one has to act in a passive lobotomised fashion. For example, Mullin (2001) describes the State Oracle of Tibet. The State Oracle is a monk consulted about difficult decisions of state. During consultations, the monk is taken over by the deity Pehar Gyalpo and may spontaneously dance with heavy ceremonial dress on and then give advice and deliberation concerning important decisions. Later the monk will not consider himself as having intentionally performed any of the actions or as having intentionally weighed up relevant pros and cons in forming a decision. Despite the involuntary nature of the performance, nobody expects the monk to be passive, acquiescent to others, lacking initiative, or in any way frontally impaired while performing. If the same mechanism underlies spirit possession as hypnosis, cold control provides a more plausible explanation than functional lobotomy.

Kirsch and Lynn (1997) also regarded hypnotic responding as being based on contention scheduling, together with the generalized intention to perform an action whenever it is suggested. They proposed that this implementation intention (Gollwitzer, e.g. 1999) produces an automatic response (the suggested action) whenever its satisfaction conditions (i.e. the suggestion) are obtained. That is, once the implementation intention has been formed (“Do what is suggested”), control can be handed over to contention scheduling. In this way, Kirsch and Lynn wished to avoid any need for a special state or special mechanisms like dissociation, and yet still have hypnotic response depend on contention scheduling and hence be experienced as automatic. However, one cannot automatically perform tasks requiring executive control, not even with generalized implementation intentions: One must specifically form the right possibility-explicit representation and this goes beyond contention scheduling. It is true that when a possibility-explicit imperative representation is formed in the SAS, that representation only

9 Work on hypnosis and creativity is also inconsistent with the claim that hypnosis reduces executive function. When a person has poor frontal function, control of behaviour is determined by contention scheduling. Contention scheduling results in the repetitive performance of habitual actions, actions appropriate to constraints relevant when the habits were formed, but not necessarily to any newly specified set of constraints. The supervisory attentional system is particularly involved in choosing novel outputs and ensuring that novel outputs are appropriate to newly specified constraints. That is, people with poor frontal function should perform poorly on tests of creativity, since creativity is all about producing something novel and appropriate. Shames and Bowers (1992) review the relationship between hypnosis and performance on a creativity test (the Torrance test of creative thinking, scored on fluency, flexibility, originality and elaboration; Torrance, 1990), and found evidence that a hypnotic induction (with highs or lows) either improved creative performance or did not alter it. Either way, the claim that frontal function was impaired by hypnosis was not supported.

30

ever does anything when it passes control to appropriate action schemata, the lower level system. The lower system, now set up to perform the action, can then perform it (unless contention scheduling and environmental triggers dissipate the effect of the SAS intention). But, a specific SAS intention is needed for specific executive tasks. If I need to not say ‘four’, then I need a specific intention “do not say ‘four’” given the constant push of contention scheduling to say ‘four’ in the context (of e.g. counting). The generalised intention to do whatever the hypnotist says does not obviate the need for further specific intentions as specific executive function tasks crop up in responding to hypnotic suggestions.

Crawford and Gruzelier (e.g. 1992) and Guzelier (1998) have proposed a neurophysiological account of hypnosis. They postulate that highs have better executive skills than lows (e.g. Crawford, Brown, & Moon, 1993) and hence can deploy their attention in different ways. Gruzelier (1998) and Gruzelier and Warren (1993) argue that the better ability of highs to focus attention allows them during an induction to exhaust their frontal abilities, and hence end up frontally impaired in a hypnotic state. By contrast, Crawford (e.g. Crawford, Knebel, & Vendemia, 1998) sees hypnotic responding to e.g. pain as dependent on the effective functioning of the supervisory attentional system in highs. Crawford’s idea is, of course, consistent with cold control theory, but, as noted above, cold control theory does not rely on any superior ability of highs over lows in frontal tasks, and the evidence for hypnotisability differences in frontal task performance indicates an effect probably too weak to actually form the basis of the capacity for hypnotic responding.

Oakley (1999) proposed that self-awareness could be separated from executive control, and hypnosis involved the executive system becoming extended to include the hypnotist; the executive system then manipulates the information available to self-awareness. The explanations for many effects may run along similar lines forOakley’s theory and cold control theory, which have in common postulating the involvement of the SAS, together with a lack of awareness of this involvement. A difference between the two theories is that Oakley sees hypnotic effects – for example, positive and negative hallucinations - as occurring at a late stage in processing. In cold control theory, the subject needs to engage in a strategy that produces a required effect, e.g. disattending from pain, which can produce effects early in the system. Further, cold control does not need the metaphor of the executive system becoming extended to include the hypnotist. Finally, cold control involves a specific commitment to higher order thought theory.

Discussion

Why does hypnotic behaviour exist? At first blush, hypnotic experience and behaviour are an unlikely product of natural selection. Surely natural selection favours accurate perception without hallucinations, and awareness of one’s own control over one’s actions (or at least not a systematic misrepresentation of the intentional causes of actions)? One possibility is that hypnosis was not specifically selected for at all. For whatever reason higher order thoughts evolved, only a certain amount of accuracy was actually required for them to do their job, and the residual amount of slack between intentions and higher order thoughts about intending – the slack that allows hypnotic experience – was simply tolerated by selective processes. But another answer is suggested by the observation that hypnotic-like experiences are extremely common cross-culturally and seem to serve definite functions (Lewis, 1973, 2003). We are taking the intentional control of cognition and behaviour without awareness of intention as the essential nature of hypnotic experiences. These types of experiences

31

occur largely associated with religious rituals and in the form of spirit possession (Lewis, 2003). Presumably, for our ancestors tens of thousands of years ago, hypnotic experiences also occurred in divine and spiritual contexts. If our distant human ancestors performed actions, or saw images etc that it seemed to them they did not produce, the obvious conclusion might have been that a spirit or divine force must have caused them.

One speculative function of such possession experiences is simply to support religious beliefs. If there were selective pressure on people to have religious beliefs, as some have argued (e.g. Alper, 1996), then the experience of being taken over by a spiritual force would help strengthen spiritual beliefs (Oesterreich, 1930), and hence could be selected for as well. (Relatedly, the compelling nature of possession experiences could also function to enhance placebo cures by convincing everyone that great spiritual powers are at work.) Note that cold control would be the perfect way of achieving this end. The strategic nature of cold control allows the experience to correspond to whatever is required by the religious beliefs held, and to make sure the experiences occur in appropriate contexts. The lack of accurate higher order thoughts provides the necessary self-deception (you must cause a behaviour/cognition but not know that you did so, so that it can be attributed to divine/spiritual intervention). The claim that hypnotic-like possession experiences can bolster religious beliefs is supported by large number of possession instances in religions struggling to get power; once religion has power, there is less need for such experiences to give authority to the religion, and indeed the experiences occur less often (Lewis, 2003).

Lewis (1971, 2003) reviews the various sociological functions of possession experiences. For example, a possessed person can perform behaviours for which they are not held responsible. A socially marginalised person (like in many cultures, a woman in a struggling marriage) can demand (with the voice of a mighty spirit) - and acquire - from the husband the gifts necessary for the spirit to be exorcised. In general, a person who speaks with voice of a spirit, acquires the authority of the spirit. If this performance is convincing to others, then the person (male or female) can climb the social ladder to achieve a status they could never have been accorded otherwise, including becoming the most senior political figure. Lewis (2003) documents just how culturally and globally widespread this phenomenon is. Again cold control provides the perfect mechanism. The behaviour and experience can be planned as to be as contextually appropriate as possible, depending on the assumptions of the particular culture. The self-deception afforded by inaccurate higher-order thoughts enables the performance to be convincing to oneself, which in turn makes it more convincing to others.

On this story, hypnosis is just one particular cultural expression of a more general phenomenon, and many of the particular characteristics of hypnotic behaviours are historical accidents frozen in time. The association of hypnosis with sleep (long discredited in the academic world), or the notion that the hypnotised person is passive, apparently lobotomised as it were, are simply particular cultural beliefs.

Cold control theory uses higher order thought (HOT) theory, levels of explicitness, and dual control to account for hypnotic phenomena; in turn, hypnotic phenomena provide support for Rosenthal’s (1986) HOT theory, for Dienes and Perner’s (1999) levels of explicitness and for the dual control model (Norman & Shallice, 1986; Perner, 1998). HOT theory specifically makes a distinction between first order states, e.g. intentions, and awareness of those states. This distinction seems crucial to understanding hypnosis, given one accepts that hypnotic responses are

32

intentional. Hypnotic responses have all the hallmarks of strategic actions. Thus, first-order phenomenological theories of volition (e.g. Gallagher, 2002) are faced with a problem in accounting for the existence of hypnotic phenomena. According to those theories, the experience of volition is part and parcel of the first order states; thus, there cannot be intended actions not experienced as volitional. That is, hypnotic phenomena indicate we need a distinction between first order intentions and awareness of those intentions. But hypnotic phenomena only support some HOT theories and not others; specifically, hypnotic phenomena only support HOT theories that allow higher order thoughts to be wrong about their target states. According to Carruthers (2000), HOTs are essentially part of (some) first order states, and hence always accurate. Carruthers only explicitly considered perceptual states, but if the same idea is applied to intentions, how could one experience intended actions as automatic on such a theory? In order to explain why intentions in the SAS are typically conscious, intentions in the SAS need to be upstream of the mechanisms by which higher order thoughts are formed. If HOTs are always accurate, and intentions in the SAS are the sort of states for which HOTs are formed, then it is not possible for people to perform intentional actions that they experience as non-intentional.

Concluding note

While we call cold control theory a “theory”, in truth we regard it more as a means of theoretically orienting in the right direction rather than as a final explanation of hypnosis. But we hope our arguments convince psychologists, philosophers, and cognitive scientists generally that hypnosis offers a rich domain for testing ideas of consciousness and control.

References

Aikens, D., & Ray, W. J. (2001). Frontal lobe contributions to hypnotic susceptibility: A neuropsychological screening of executive function. International Journal of Clinical and Experimental Hypnosis, 49, 320-329.

Alper, M. (1996). The “God” part of the brain: A scientific interpretation of human spirituality and God. Rogue Press: New York.

Arnold, M. B. (1946). On the mechanism of suggestion and hypnosis. Journal of Abnormal and Social Psychology, 41, 107-128.

Baars, B. (1988). A cognitive theory of consciousness. Cambridge: Cambridge University Press

Baars, B. (1997). In the theatre of consciousness. Oxford: Oxford University Press.

Baddeley, A. (1986). Working memory. Oxford: Oxford University Press.

Barber, T. X. (1969). Hypnosis: A scientific approach. New York: Van Nostrand.

Barber, T. X., & Glass, L. B. (1962). Significant factors in hypnotic behaviour. Journal of Abnormal and Social Psychology, 64, 222-228.

33

Barber, T. X., Spanos, N. P. & Chaves, J. F. (1974). Hypnosis, imagination, and human potentialities. New York: Pergamon.

Barnier, A., & McConkey, K. M. (1999). Absorption, hypnotisability, and context: Nonhypnotic contexts are not all the same. Contemporary Hypnosis, 16, 1-8.

Bertrand, L. D., & Spanos, N. P. (1985). The organization of recall during hypnotic suggestions for complete and selective amnesia. Imagination, Cognition, and Personality, 4, 249-261.

Bentall, R. P. (1990). The illusion of reality: a review and integration of psychological research on hallucinations. Psychological Bulletin, 107, 82-95.

Bowers, K. S. (1992). Imagination and dissociation in hypnotic responding. International Journal of Clinical and Experimental Hypnosis, 40, 253-275.

Bowers, K. S., & Woody, E. Z. (1996). Hypnotic amnesia and the paradox of intentional forgetting. Journal of Abnormal Psychology, 105, 381-390.

Bowers, P. G. (1978). Hypnotizability, creativity and the role of effortless experiencing. International Journal of Clinical and Experimental Hypnosis, 26, 184-202.

Braffman, W., Kirsch, I. (1999). Imaginative suggestibility and hypnotizability: An empirical analysis. Journal of Personality and Social Psychology, 77, 578-587.

Bridgeman, B., & Huemer, V. (1998). A spatially oriented decision does not induce consciousness in a motor task. Consciousness and Cognition, 7, 454-464.

Bushnell MC, Duncan GH, Hofbauer RK, Ha B, Chen J, and Carrier B. (1999). Pain perception: is there a role for primary somatosensory cortex? Proceedings of the National Academy of Sciences of the USA, 96, 7705-7709.

Carruthers, P. (2000). Phenomenal consciousness naturally. Cambridge: Cambridge University Press.

Cheesman J. & Merikle, P. M. (1984). Priming with and without awareness. Perception & Psychophysics, 36, 387-395.

Council, J. R., Kirsch, I., & Hafner, L. P. (1986). Expectancy versus absorption in the prediction of hypnotic responding. Journal of Personality and Social Psychology, 50, 182-189.

Crawford, H. J., Brown, A. M., & Moon, C. E. (1993). Sustained attentional and disattentional abilities: Difference between low and highly hypnotisable persons. Journal of Abnormal Psychology, 102, 534-543.

Crawford, H. J., & Gruzelier, J. H. (1992). A midstream view of the neuropsychophysiology of hypnosis: Recent research and future directions. In E.

34

Fromm, & M. R. Nash, (Eds), Contemporary hypnosis research (pp 227 – 266). The Guilford Press: New York.

Crawford, H. J., Knebel, T., & Vendemia, J. M. C. (1998). The nature of hypnotic analgesia: Neurophysiological foundation and evidence. Contemporary Hypnosis, 15, 24-35.

David, D., & Brown, R. J. (2002). Suggestibility and negative priming: Two replication studies. International Journal of Clinical and Experimental Hypnosis, 50, 215-228.

David, King, and Borckardt (2001). Is capacity for negative priming correlated with hypnotisability? A preliminary study. International Journal of Clinical and Experimental Hypnosis, 49, 30-37.

Dehaene S., Naccache L., Cohen L., Bihan D. L., Mangin J. F., Poline J. B., Riviere D. (2001) Cerebral mechanisms of word masking and unconscious repetition priming. Nature Neuroscience, 4, 752-758.

Dennett, D. C. (2003). Freedom evolves. London: Penguin Press.

Dienes, Z. (1987). Selective attention: Relevance to hypnosis and hypnotizability. Thesis submitted in partial fulfillment of the requirements for Master of Arts (Hons), School of Behavioural Sciences, Macquarie University.

Dienes, Z. (in press). Assumptions of subjective measures of unconscious mental states: Higher order thoughts and bias. Journal of Consciousness Studies.

Dienes, Z., & Perner, J. (1999) A theory of implicit and explicit knowledge. Behavioural and Brain Sciences, 22, 735-755.

Dienes, Z., & Perner, J. (2002a) A theory of the implicit nature of implicit learning. In French, R M & Cleeremans, A. (Eds), Implicit Learning and Consciousness: An Empirical, Philosophical, and Computational Consensus in the Making? Hove, UK: Psychology Press (pp 68-92).

Dienes, Z., & Perner, J. (2002b). The metacognitive implications of the implicit-explicit distinction. In Chambres, P., Izaute, M., & Marescaux, P.-J. (Eds), Metacognition: Process, function, and use. Dordrecht, Netherlands: Kluwer Academic Publishers (pp 171-190).

Dienes, Z., & Perner, J. (2003). Unifying consciousness with explicit knowledge. In Cleeremans, A. (Ed.) The unity of consciousness: binding, integration, and dissociation. (pp 214-232). Oxford University Press.

Dixon, M., Brunet, A., & Laurence, J.-R. (1990). Hypnotizability and automaticity: Toward a parallel distributed processing model of hypnotic responding. Journal of Abnormal Psychology, 99, 336-343.

35

Dixon, M., & Laurence, J.-R. (1992). Hypnotic susceptibility and verbal automaticity: Automatic and strategic processing differences in the Stroop-colour naming task. Journal of Abnormal Psychology, 101, 344-347.

Driver, J Vuilleumier, P Eimer, M and Rees G (2001). Functional magnetic resonance imaging and evoked potential correlates of conscious and unconscious vision in parietal extinction patients. NeuroImage, 14, S68 – S75.

Eastwood, J. D., Gaskovsky, P., & Bowers, K. S. (1998). The folly of effort: ironic effects in the mental control of pain. International Journal of Clinical and Experimental Psychology, 46, 77-91.

Evans, F. J. (1980). Posthypnotic amnesia. In G. D. Burrows & L. Dennerstein (Eds), Handbook of hypnosis and psychosomatic medicine (pp 85-103). Amsterdam: Elsevier/North-Holland.

Evans, F. J. (1991). Hypnotizability: Individual differences in dissociation and the flexible control of psychological processes. In S. J. Lynn, & J. W. Rhue, (Eds), Theories of hypnosis: Current models and perspectives (pp 144 – 169). The Guilford Press: New York.

Frith, C. D. (1992). The cognitive neuropsychology of schizophrenia. Hove, UK: Psychology Press.

Fromm, E., & Nash, M. R. (Eds) (1992). Contemporary hypnosis research. The Guilford Press: New York.

Gallagher, S. (2002) Self-agency and self-ownership. Paper presented at the 2002 Congress of the European Society for Philosophy and Psychology, Lyon, France, July 10-13. (http://www2.canisius.edu/~gallaghr/)

Gennaro, R. W. (Ed.) (forthcoming) Higher-Order Theories of Consciousness. John Benjamins Publishers.

Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of simple plans. American Psychologist, 54, 493-503.

Graffin, N. F., Ray, W. J., & Lundy, R. (1995). EEG concomitants of hypnosis and hypnotic susceptibility. Journal of Abnormal Psychology, 104, 123-131.

Graham, C., & Evans, F. J. (1977). Hypnotizability and the deployment of waking attention. Journal of Abnormal Psychology, 86, 631-638.

Green, J. P., & Lynn, S. J. (1995). Hypnosis, dissociation, and simultaneous task performance. Journal of Personality and Social Psychology, 69, 728-735.

Gruzelier, J. (1998). A working model of the neurophysiology of hypnosis: A review of evidence. Contemporary Hypnosis, 15, 3-21.

36

Gruzelier, J., & Warren, K. (1993). Neuropsychological evidence of reductions on left frontal tests with hypnosis. Psychological Medicine, 23, 93-101.

Hargadon, R., Bowers, K. S., & Woody, E. Z. (1995). Does counterpain imagery mediate hypnotic analgesia? Journal of Abnormal Psychology, 104, 508-516.

Hilgard, E. R. (1965). Hypnotic susceptibility. New York: Harcourt, brace, & World.

Hilgard, E. R. (1977). Divided consciousness: Multiple controls in human thought and action. New York: Wiley-Interscience.

Hilgard, E. R. (1986). Divided consciousness: Multiple controls in human thought and action, second edition. New York: Wiley-Interscience.

Hilgard, E. R. (1992). Dissociation and theories of hypnosis. In E. Fromm, & M. R. Nash, (Eds), Contemporary hypnosis research (pp 69 – 101). The Guilford Press: New York.

Hilgard, E.R. & Hilgard, J.R.(1975) Hypnosis in the Relief of Pain. William Kaufman.

Hilgard, E. R., & Tart, C. T. (1966). Responsiveness to suggestions following waking and imagination suggestions and following induction of hypnosis. Journal of Abnormal Psychology, 71, 196-208.

Hilgard, J. R. (1970). Personality and hypnosis: A study of imaginative involvement. Chicago: University of Chicago Press.

Hilgard, J. R. (1974). Imaginative involvement: Some characteristics of the highly hypnotizable and the nonhypnotisable. Intnerational Journal of Clinical and Experimental Hypnosis, 22, 138-156.

Hofbauer, R. K., Rainville, P., Duncan, G. H., & Bushnell, M. C. (2001). Cortical representation of the sensory dimension of pain. Journal of Neurophysiolology, 86, 402-411.

Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30, 513-541.

James, W. (1955/1890). The principles of psychology. Dover Publications.

Jamieson, G. A., & Sheehan, P. W. (2002). A critical evaluation of the relationship between sustained attentional abilities and hypnotic susceptibility. Contemporary Hypnosis, 119, 62-74.

Jamieson, G. A., & Sheehan, P. W. (in press). An empirical test of Woody and Bowers’ dissociated control theory of hypnosis. International Journal of Clinical and Experimental Hypnosis

Janet, P. (1889). L’automatisme psychologique. Paris: Felix Alcan

37

Kaiser, J., Barker, R., Haenschel, C., Baldeweg, T., & Gruzelier, J. H. (1997). Hypnosis and event-related potential correlates of error-processing in a stroop-type paradigm: A test of the frontal hypothesis. International Journal of Psychophysiology, 27, 215-222.

Kalio, S., & Ihamoutila, M. (1999). The Finnish norms for the Harvard Group Scale of Hypnotic Susceptibility Form A. International Journal of Clinical and Experimental Hypnosis, 47, 227-235.

Kallio, S., & Revonsuo, A. (2003). Hypnotic phenomena and altered states of consciousness: A multilevel framework of description and explanation. Contemporary Hypnosis, 20, 111-164.

Kallio, S., & Revonsuo, A., Lauerma, H., Hämäläinen, H., & Markela, J., & Gruzelier, J. (2001). Anterior brain functions and hypnosis: A test of the frontal hypothesis. International Journal of Clinical and Experimental Hypnosis, 49, 95-108.

Karlin, R. A. (1979). Hypnotisability and attention. Journal of Abnormal Psychology, 88, 92-95.

Karlin, R. A., Morgan, D., & Goldstein, L. (1980). Hypnotic analgesia: A preliminary investigation of quantitated hemispheric electroencephalographic and attentional correlates. Journal of Abnomral Psychology, 89, 591-594.

Kihlstrom, J. F. (1985). Hypnosis. Annual Review of Psychology, 36, 385-418.

Kihlstrom, J. F. (1997). Consciousness and me-ness. In J. Cohen & J. Schooler (Eds.), Scientific approaches to consciousness. (pp. 451-468). Mahwah, NJ, USA: Lawrence Erlbaum Associates, Inc.

Kihlstrom, J. F. (1998). Dissociations and dissociation theory in hypnosis: Comment on Kirsch and Lynn (1998). Psychological Bulletin, 123, 186-191.

King, B. J., & Council, J. R. (1998). Intentionality during hypnosis: An ironic process analysis. International Journal of Clinical and Experimental Hypnosis, 46, 295-313.

Kinnunen, T., Zamansky, H. S., & Block, M. L. (1994). Is the hypnotised subject lying? Journal of Abnormal Psychology, 103, 184-191.

Kinnunen, T., Zamansky, H. S., & Nordstrom, B. L. (2001). Is the hypnotized subject complying? International Journal of Clinical and Experimental Hypnosis, 49, 83-94.

Kirsch, I. (1985). Response expectancy as a determinant of experience and behaviour. American Psychologist, 40, 1189-1202.

Kirsch, I. (1991). The social learning theory of hypnosis. In S. J. Lynn, & J. W. Rhue, (Eds), Theories of hypnosis: Current models and perspectives (pp 439 – 465). The Guilford Press: New York.

38

Kirsch, I., & Council, J. R. (1992). Situational and personality correlates of hypnotic responsiveness. In E. Fromm, & M. R. Nash, (Eds), Contemporary hypnosis research (pp 267 – 291). The Guilford Press: New York.

Kirsch, I., & Lynn, S. J. (1995). The altered state of hypnosis: Changes in the theoretical landscape. American Psychologist, 50, 846-858.

Kirsch, I., & Lynn, S. J. (1997). Hypnotic involuntariness and the automaticity of everyday life. American Journal of Clinical Hypnosis, 40, 329-348.

Kirsch, I. & Lynn, S. J. (1998). Dissociation theories of hypnosis. Psychological Bulletin, 123, 100-115.

Kirsch, I. & Lynn, S. J. (1999). Automaticity in clinical psychology. American Psychologist, 54, 504-515.

Kirsch I., Silva C. E., Carone J. E., Johnston J. D., Simon B. (1989). The surreptitious observation design – an experimental paradigm for distinguishing artifact from essence in hypnosis. Journal of Abnormal Psychology, 98, 132-136.

Kopelman M., Morton J. (2001). Psychogenic amnesia - functional memory loss. In Davies G., Dalgleish T. (Eds), Recovered Memories: Seeking the Middle Ground. Chichester, Wiley. (pp 219-243).

Kosslyn, S. M., & Thompson, W. L. (2003). When is early visual cortex activated during visual mental imagery? Psychological Bulletin, 5, 723-746.

Kosslyn, S. M., & Thompson, W. L., Constantini-Ferrando, M. F., Alpert, N. M., Spiegel, D. (2000). Hypnotic visual illusion alters colour processing in the brain. American Journal of Psychiatry, 157, 1279-1284.

Lewis, I. M. (1971). Ecstatic religion: A study of shamanism and spirit possession. Routledge: London.

Lewis, I. M. (2003). Ecstatic religion: A study of shamanism and spirit possession, Third Edition. Routledge: London.

Lynn, S. J., Nash, M. R., Rhue, J. W., Frauman, D. C., & Sweeney, C. A. (1984). Nonvolition, expectancies, and hypnotic rapport. Journal of Abnormal Psychology, 93, 295-303.

Lynn, S. J., & Rhue, J. W. (Eds) (1991). Theories of hypnosis: Current models and perspectives. The Guilford Press: New York.

Lyons, L. C., & Crawford, H. J. (1997). Sustained attentional and disattentional abilities and arousability: Factor analysis and relationship to hypnotic susceptibility. Personailty and Individual Differences, 26, 1071-1084.

MacLeod, C. M., Sheehan, P.W. (2003). Hypnotic control of attention in the Stroop task: A historical footnote. Consciousness and Cognition, 12, 347-353.

39

Marcel, A. J. (1983). Consciousness, masking, and word recognition. Cognitive Psychology, 15, 198-237.

McConkey, K. M. (1991). The construction and resolution of experience and behaviour in hypnosis. In S. J. Lynn & J. W. Rhue (Eds.), Theories of hypnosis: Current models and perspectives (pp 542-563). New York: Guilford Press.

McConkey, K. M., & Barnier, A. (in press). High hypnotisability: Unity and diversity in behaviour and experience. In M. Heap, R. J. Brown, & D. A. Oakley (Eds.), High hypnotisability: Theoretical, experimental, and clinical issues. London: Bruner-Routledge.

Merikle, P. M. (1992). Perception without awareness: Critical issues. American Psychologist, 47, 792-795.

Miller, M. E., & Bowers, K. S. (1986). Hypnotic analgesia and stress inoculation in the reduction of pain. Journal of Abnormal Psychology, 95, 6-14.

Miller, R. J. (1975). Response to the Ponzo illusion as a reflection of hypnotic susceptibility. International Journal of Clinical and Experimental Hypnosis, 13, 148-167.

Millikan, R. G. (1984). Language, thought, and other biological categories. Cambridge, MA: MIT Press.

Milner, A.D. & Goodale, M.A. (1995) The visual brain in action. Oxford: Oxford University Press.

Mullin, G. H. (2001). The fourteen Dalai Lamas: A sacred legacy of reincarnation. Clear Light Publishers: Santa Fe, New Mexico.

Naish, P. L. N. (1985). The “trance” described in signal detection terms. British Journal of Experimental and Clinical Hypnosis, 2, 139-142.

Nogrady, H., McConkey, K.M., Laurence, J.R. & Perry, C.(1983) Dissociation, duality and demand characteristics in hypnosis. Journal of Abnormal Psychology 92:223-235.

Norman, D., & Shallice, T., 1986: Attention to action: Willed and automatic control of behavior. In Davidson, R., Schwartz, G., and Shapiro, D., (eds.) Consciousness and Self Regulation: Advances in Research and Theory, Volume 4. Plenum: New York, NY. pp. 1-18.

Oakley, D.A. (1999). Hypnosis and consciousness: A structural model. Contemporary Hypnosis, 16, 215-223.

Oakman, J. M., & Woody, E. (1996). A taxometric analysis of hypnotic susceptibility. Journal of Personality and Social Psychology, 72, 981-991.

40

Oesterreich, T. K. (1930). Possession, demoniacal and other, among primitive races, in antiquity, the Middle Ages, and modern times. London.

Orne, M. T. (1959). The nature of hypnosis: Artifact and essence. Journal of Abnormal Psychology, 58, 277-299.

O’Shaughnessy, B. (2000). Consciousness and the world. Oxford: Oxford University Press.

Perner, J. (1991) Understanding the representational mind. MIT Press.

Perner, J. (1998). The meta-intentional nature of executive functions and theory of mind. In P. Carruthers & J. Boucher (Eds.), Language and Thought: Interdisciplinary themes (270-283). Cambridge: Cambridge University Press.

Perner, J. (2003). Dual control and the causal theory of action. In N.Eilan & J.Roessler (Eds.). Agency and self-awareness. Oxford: Oxford University Press (pp 218-243.

Perner, J., and Dienes, Z. (1999) Higher order thinking. Behavioural and Brain Sciences, 22, 164-165.

Perry, C., Nadon, R., Button, J. (1992). The measurement of hypnotic ability. In E. Fromm, & M. R. Nash, M (Eds) (1992). Contemporary hypnosis research (459 – 490). The Guilford Press: New York.

Perugini EM, Kirsch I, Allen ST, Coldwell E, Meredith JM, Montgomery GH, Sheehan J (1998). Surreptitious observation of responses to hypnotically suggested hallucinations: A test of the compliance hypothesis. International Journal of Clinical and Experimental Hypnosis, 46, 191-203.

Prince, M. (1909). Experiments to determine co-conscious (subconscious) ideation. Journal of Abnormal Psychology, 3, 33-42.

Raz, A., Landzberg, K. S., Schweizer, H. R., Zephrani, Z. R., Shapiro, T., Fan, J., & Posner. M. I. (2003). Posthypnotic suggestion and the modulation of Stroop interference under cycloplegia. Consciousness and Cognition, 12, 332-346.

Raz, A., Shapiro, T., Fan, J., Posner, M. I. (2002) Hypnotic Suggestion and the Modulation of Stroop Interference. Archives of General Psychiatry, 59, 1155-1161.

Reason, J. (1979).Actions not as planned:The price of automatization. In G. Underwood & R. Stevens (Eds.) Aspects of consciousness: Vol. 1. Psychological issues. London: Wiley.

Rees, G. (2001). Seeing is not perceiving. Nature Neuroscience, 4, 678-680.

Rees, G., Kreiman, & Koch, C. (2002). Neural correlates of consciousness in humans. Nature Reviews Neuroscience, 3, 261-270.

41

Roche, S. M., & McKonkey, K. M. (1990). Absorption: Nature, assessment, and correlates. Journal of Personality and Social Psychology, 59, 91-101.

Rosenthal, D. M. (1986). Two Concepts of Consciousness. Philosophical Studies, 49,. 329-359.

Rosenthal, D. M. (2000). Consciousness, Content, and Metacognitive Judgments. Consciousness and Cognition, 9, 203-214.

Rosenthal, D. M. (2002). Consciousness and Higher-Order Thought. In the Macmillan Encyclopedia of Cognitive Science, pp. 717-726. Macmillan Publishers Ltd.

Rosenthal, D. M. (2003). Unity of Consciousness and the Self. Proceedings of the Aristotelian Society, 103, 325-352.

Rosenthal, D. M. (2004). Consciousness and Mind. Oxford: Clarendon Press.

Rossetti, Y., & Rode, G. (2001) Numbsense and blindsight. In De Gelder B., De Haan E., Heywood C. Eds. Out of Mind: Varieties of unconscious processing. Oxford University Press.

Sackheim, H. A., Nordlie, J. W., & Gur, R. C. (1979). A model of hysterical and hypnotic blindness: Cognition, motivation, and awareness. Journal of Abnormal Psychology, 88, 474-489.

Sarbin, T. R., & Coe, W. C. (1972). Hypnosis: A social psychological analysis of influence communication. New York: Holt, Rinehart, & Winston.

Szechtman, H., Woody, E., Bowers, K. S., & Nahmias, C. (1998). Where the imaginal appears real: A positron emission tomography study of auditory hallucinations. Proceedings of the National Academy of Sciences, 95, 1956-1960.

Shames, V. A., & Bowers, P. G. (1992). Hypnosis and creativity. In E. Fromm, & M. R. Nash, (Eds), Contemporary hypnosis research (pp 334 – 363). The Guilford Press: New York.

Sheehan, P. W. (1982). Imagery and hypnosis: forging a link, at least in part. Research in Communications in Psychology, Psychiatry, and Behaviour, 7, 257-272.

Sheehan, P. W., & McConkey, K. M. (1982). Hypnosis and experience: the exploration of phenomena and process. Hillsdale, N.J.:Erlbaum.

Sigman, A., Phillips, K. C., & Clifford, B. (1985). Attentional concomitants of hypnotic susceptibility. British Journal of Experimental and Clinical Hypnosis, 2, 69-75.

Spanos, N. (1986). Hypnotic behaviour: A social-psychological interpretation of amnesia, analgesia, and "trance logic." Behavioural and Brain Sciences, 9, 449-502.

42

Spanos, N. P. (1991). A sociocognitive approach to hypnosis. In S. J. Lynn & J. W. Rhue, (Eds) Theories of hypnosis: Current models and perspectives (324-361). The Guilford Press: New York.

Spanos, N.P., Burgess, C. A., Roncon, V., Wallacecapretta, S., & Cross, P. (1993). Surrepitiously observed hypnotic responding in simulators and in skill-trained and untrained high hypnotisables. Journal of Personality and Social Psychology, 65, 391-398.

Spanos, N. P. Gwynn, M., & Stam, H. J. (1983). Instructional demands and ratings of overt and covert pain during hypnotic analgesia. Journal of Abnormal Psychology, 92, 479-488.

Spanos, N. p., Radtke, H. L., & Dubreuil, D. L. (1982). Episodic and semantic memory in post-hypnotic amnesia: A re-evaluation. Journal of Personality and Social psychology, 43, 565-573.

Spanos, N. P., Rivers, S. M., & Gottlieb, J. (1978). Hypnotic responsivity, meditation, and laterality of eye movements. Journal of Abnormal Psychology, 87, 566-569.

Spanos, N. P., Stam, H. J., Rivers, S. M., & Radtke, H. L. (1980). Meditation, expectation and performance on indices of non-analytic attending. International Journal of Clinical and Experimental Hypnosis, 28, 244-251.

Stava, L. J., & Jaffa, M. (1988).Some operationalizations of the neodissociation concept and their relationship to hypnotic susceptibility. Journal of Personality and Social Psychology, 54, 989-996.

Tellegen, A., & Atkinson, G. (1974). Openess to absorbing and self-altering experiences (‘absorption’), a trait related to hypnotic susceptibility. Journal of Abnormal Psychology, 83, 268-277.

Van Nuys, D. (1973). Meditation, attention, and hypnotic susceptibility: A correlational study. International Journal of Clinical and Experimental Hypnosis, 21, 59-69.

Wagstaff, G. F. (1981). Hypnosis, compliance, and belief. New York: St Martins Press.

Wagstaff, G. F., Toner, S., & Cole, J. (2002). Is response expectancy sufficient to account for hypnotic negative hallucinations? Contemporary Hypnosis, 19, 133-158.

Wallace, B. (1979). Hypnotic susceptibility and the perception of after-images and dot stimuli. American Journal of Psychology, 92, 681-691.

Wallace, B., Knight, T. A., & Garrett, J. B. (1976). Hypnotic susceptibility and frequency reports to illusory stimuli. Journal of Abnormal Psychology, 85, 558-563.

Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review, 101, 34-52.

43

Wegner, D. M. (2002). The illusion of conscious will. Cambridge, MA: MIT Press.

Wegner, D. M., & Erskine, J. (2003). Voluntary involuntariness: Thought suppression and the regulation of the experience of will. Consciousness and Cognition, 12, 684-694.

Wegner, D. M., & Wheatley, T. P. (1999). Apparent mental causation: Sources of the experience of will. American Psychologist, 54, 480-492.

Weiskrantz, L. (1986). Blindsight: A case study and implications. Oxford: Oxford University Press.

Weiskrantz, L. (1997) Consciousness lost and found. Oxford: Oxford University Press.

Weizenhoffer, A. M. (1974). When is an ‘instruction’ an ‘instruction’? International Journal of Clinical and Experimental Hypnosis, 22, 258-269.

White, R. W. (1941). A preface to a theory of hypnotism. Journal of Abnormal and Social Psychology, 36, 477-505.

Wickless, C., & Kirsch, I. (1989). The effect of verbal and experiential expectancy manipulations on hypnotic susceptibility. Journal of Personality and Social Psychology, 57, 762-768.

Wilson, S. C., & Barber, T. X. (1981). Vivid fantasy and hallucinatory abilities in the life histories of excellent hypnotic subjects (“somnambules”):Preliminary report with female subjects. In E. Klinger (Ed.), Imagery: Vol 2. Concepts, results, and applications (pp. 133- 152). New York: Plenum Press.

Woody, E., & Bowers, K. S. (1994). A frontal assault on dissociated control. In S. J. Lynn & J. W. Rhue (Eds), Dissociation: Clinical and theoretical perspectives (pp 52 – 79). New York: Guilford Publications.

Woody, E., Drugovic, M., & Oakman, J. M. (1997). A re-examination of the role of nonhypnotic suggestibility in hypnotic responding. Journal of Personality and Social Psychology, 72, 399-407.

Woody, E., & Farvolden, P. (1998). Dissociation in hypnosis and frontal executive function. American Journal of Clinical Hypnosis, 40, 206-216.

Woody, E., & Sadler, P. (1998). On re-integrating dissociated theories: Comment on Kirsch and Lynn (1998). Psychological Bulletin, 123, 192-197.

Woody, E., & Szechtman, H. (2000). Hypnotic hallucinations: Towards a biology of epistemology. Contemporary Hypnosis, 17, 4-14.

44

Zamansky, H. S., & Clark, L. E. (1986). Cognitive competition and hypnotic behaviour: Whither absorption? International Journal of Clinical and Experimental Hypnosis, 34, 205-214.

45