Hallucinations and Reality Discrimination: Rethinking the Illusion of Reality

Ben Goertzel Intelligenesis Corp.

1. Introduction

Hallucinations have proved almost as vexing to psychological theorists as they are to the individuals who experience them. A variety of theories have been proposed, some psychological, some neurobiological, and some combining psychological and physiological factors (for reviews see Bentall, 1990, Slade and Bentall, 1988; Slade, 1994). However, none of these theories has proved entirely satisfactory.

Perhaps the most precise and comprehensive theory to date is that of Bentall and Slade (Bentall, 1990; Bentall and Slade, 1988; Slade, 1994), which holds that hallucinations are due to a lack of skill at "reality-testing" or reality discrimination.

According to this theory, individuals prone to hallucinate do not generate internal stimuli differently from the rest of us; they merely interpret these stimuli differently.

The Bentall and Slade theory provides a good qualitative understanding of a variety of clinical observations and experimental results regarding hallucinatory experience. However, the direct evidence for the theory is relatively scant. There are experiments which show that individuals with high Launay-Slade Hallucination scores have a bias toward classifying signals as real, instead of imagined (Bentall and Slade, 1988; Feelgood and Rantsen, 1994). But these experiments only indicates a correlation between poor reality discrimination and hallucination. They do not establish a causal relationship, which is what Slade (1994) and Slade and Benthall (1988) posit.

I will argue that, in fact, poor reality discrimination and hallucination produce one another, given their initial impetus by the cognitive trait that Hartmann (1991) calls "thin-boundariedness." In short, certain individuals place particularly permeable boundaries around entities in their minds, including their self-systems.

This is a result of individual differences in the nature of consciousness-embodying "perceptual-cognitive loops," as described in Goertzel (1994). The tendency to construct permeable boundaries, it is argued, encourages both hallucination and poor reality discrimination, which in turn are involved in positive feedback relationships with each other.

These ideas are harmonious with the dynamical- systems-theoretic psychology presented in Goertzel (1994) and Abraham and Shaw (1991). They are an alternate and, I suggest, more psychologically plausible explanation of the data which Bentall and Slade take in support of their theory of hallucinations.

2. Six Possible Explanations

Before embarking on our analysis of hallucination, some methodological clarifications may be in order. It is well- known that "correlation does not imply causation." However, attention is rarely drawn to the full realm of possible explanations for a correlation between two variables. Given a correlation between A and B, there are at least six distinct possible explanations, all of which must be taken seriously.

The three commonly- recognized explanations of a correlation between A and B are as follows:

1. A may cause B
2. B may cause A
3. A and B may both be caused by some other factor C.
The three additional explanations ensue from dynamical systems theory (Abraham and Shaw, 1991; Goertzel, 1994), which indicates that two variables may in fact "cause each other" due to positive feedback. They are as follows:

4. A and B may cause each other (with no initial prompting other than random fluctations)
5. A and B may cause each other, once initially activated by some other factor C
6. A and B may cause each other, in a process of mutual feedback with some other factor C
These possibilities are all observed in connectionist models on a routine basis (Rumelhart and McClelland, 1986), and in fact have a long history in Oriental psychology, going back at least to the notion of "dependent causation" in Buddhist psychology (Crook and Rabgyas, 1988).

Let us now consider all six of these possibilities in the context of reality discrimination and hallucination.

First, there is the possibility that unreliable reality discrimination causes hallucinations; i.e. that, as has been claimed, "hallucinations result from a dramatic failure of the skill of reality discrimination" (Slade and Benthall, 1988).

Then there is the possibility that things work the other way around: that hallucinations cause unreliable reality discrimination. This hypothesis is also quite plausible. For, consider reality discrimination as a categorization problem. One might reason as follows. In ordinary experience, there are substantial differences between internally generated- stimuli and externally-generated stimuli. Thus, it is easy to "cluster" stimuli into two categories: real versus imaginary. But, in the experience of a person prone to hallucinations, there is more of a continuum from internally-generated to externally-generated stimuli. The two categories are not so distinct, and thus categorization is not such an easy problem. Quite naturally, for such individuals, skill at distinguishing the two categories will be below par.

Third, there is the possibility that both hallucinations and unreliable reality discrimination are caused by some other factor, or some other group of factors. This of course raises the question of what these other factor(s) might be.

Fourth, there is the possibility that hallucinations and unreliable reality discrimination cause each other. In other words, the two may stand in a positive feedback relation to each other: the more one hallucinates, the worse one does reality discrimination; and the worse one does reality discrimination, the more one hallucinates. This possibility is inclusive of the first two possibilities.

Finally, there are combinations of the third and fourth options. The fifth possibility is that an external factor prompts a slight initial propensity toward poor reality discrimination and hallucination, which then blossoms, by positive feedback, into more prominent phenomena. And the sixth possibility is that poor reality discrimination and hallucinations to emerge cooperatively, by positive feedback, in conjunction with some other factor.

Here I will argue for the fifth possibility. My argument is perhaps somewhat speculative, but no more so than the arguments of Slade (1994) and Slade and Benthall (1988). It represents a much more natural interpretation of the data which they adduce in favor of their theory.

3. Consciousness Creates Reality

In this section, drawing on neuropsychological evidence, I will argue that the role of consciousness is to create percepts and concepts from stimuli, or in other words, to create subjective reality. But not everyone creates reality in the same way; the parameters of the process differ. In order to conceptualize individual differences in reality creation, I will introduce Hartmann's (1988) notion of "thin-boundaried" versus "thick- boundaried" individuals. This section is a summary of ideas presented at greater length, and with greater motivation, in (Goertzel, 1994).

The relevance of these considerations to hallucinations is quite simple. For, hallucinations are a matter of confusing internally generated stimuli for external reality. And consciousness, I will argue, is largely concerned with the construction of reality.

The relation between hallucinations and consciousness has been discussed in the past. In particular, Frith (1979) has used the limited channel capacity of consciousness to explain the occurence of hallucinations. Because of the limited capacity of consciousness, it is argued, only one hypothesis regarding the nature of a given stimulus can be consciously entertained at a given time. Hallucinations occur when preconscious information about a stimulus is not filtered out, so that consciousness becomes crowded with information, and the correct hypothesis (that the stimulus is internally rather than externally generated) is pushed out. It is worth emphasizing that the ideas presented here are quite different from these. Here we are arguing that individual differences in conscious information processing are crucial for hallucination. Frith, on the other hand, posits that the nature of conscious processing is invariant with respect to the propensity toward hallucination, while the type of information fed into consciousness varies.

"Consciousness constructs reality" is a strong statement, but it is essentially the same as many statements found in the neuropsychology literature, for instance Umilta's (1988) remark that "the formation of a given percept is dependent on a specific distribution of focal attention." Many neurospcyhologists have concluded that it is attention, i.e. directed consciousness, that actually causes the formation of percepts. And what are percepts but our inner version of external reality?

As a particular manifestation of this general idea, Treisman and Schmidt (1982) have argued for a two- stage theory of visual perception. First is the stage of elementary feature recognition, in which simple visual properties like color and shape are recognized by individual neural assemblies. Next is the stage of feature integration, in which consciousness focuses on a certain location and unifies the different features present at that location. If consciousness is not focused on a certain location, the features sensed there may combine on their own, leading to the perception of illusory objects.

Psychological experiments have verified the same phenomenon. For instance, Kawabata (1986) has shown that one makes a choice between the two possible orientations of the Necker cube based on the specific point on which one first focuses one's attention. Whatever vertex is the focus of attention is perceived as in the front, and the interpretation of the whole image is constructed to match this assumption. Similar results have been found for a variety of different ambiguous figures -- e.g. Tsal and Kolbet (1985) used pictures that could be interpreted as either a duck or a rabbit, and pictures that could be seen as eithr a bird or a plane. In each case the point of conscious attention directed the perception of the whole. And once consciousness has finished forming the picture into a coherent perceived whole, this process is very difficult to undo.

Treisman and Schmidt's division of perception into two levels is perhaps a little more rigid than the available evidence suggests. For instance, experiments of Prinzmetal et al (1986) verify the necessity of consciousness for perceptual integration, but also point out some minor role for consciousness in enhancing the quality of perceived features. But there are many ways of explaining this kind of result. It may be that consciousness acts on more than one level: first in unifying sub-features into features, then in unifying features into whole objects. Or it may be that perception of the whole causes perception of the features to be improved, by a sort of feedback process.

Having established that consciousness in some sense creates reality, the next question is how. In order to resolve this issue, I will borrow from Edelman (1990) the idea that consciousness consists of a feedback loop from the perceptual regions of the brain to the "higher" cognitive regions. In other words, consciousness is a process which cycles information from perception to cognition, to perception, to cognition, and so forth, in the process continually creating new information to be cycled around. In accordance with the parallel, distributed nature of consciousness (Dennett, 1991), it should be expected that a given brain will contain a number of such loops, bridging different parts of the cortex. These loops I call Perceptual- Cognitive Loops, or PCL's.

In the case of vision, for example, the PCL process may be conjectured to work as follows. First, the perceptual end does some primitive grouping procedures, and passes its results along to the cognitive end, asking for approval: "Did I group too little, or enough?" The cognitive end seeks to integrate the results of the perceptual end with its knowledge and memory, and on this basis gives an answer. In short, it acts on the percepts, by trying to do things with them, by trying to use them to interface with memory and motor systems. It gives the answer "too little coherentization" if the proposed grouping is simply torn apart by contact with memory -- if different parts of the supposedly coherent percept connect with totally different remembered percepts, whereas the whole connects significantly with nothing. And when the perceptual end receives the answer "too little," it goes ahead and tries to group things together even more, to make things even more coherent. Then it presents its work to the cognitive end again. Eventually the cognitive end of the loop answers: "Enough!" At the end of the process, one has an entity which is sufficiently coherent to withstand the onslaughts of memory.

The judgement of "enough" here is crucial. A maximally coherent percept is not desirable, because thought, perception and memory require that ideas possess some degree of flexibility. The individual features of a percept should be detectable to some degree, otherwise how could the percept be related to other similar ones? The trick is to stop the coherence- making process at just the right time. However, it should not be assumed that there is a unique optimal level of coherence. It seems more likely that each consciousness- producing loop has its own characteristic level of cohesion.

This is where the ideas of Hartmann (1991) become relevant. Hartmann has argued that each person has a certain characteristic "boundary thickness" which they place between the different ideas in their mind. Thick-boundaried people perceive a rigid division between themselves and the external world, and they create percepts and concepts which hold together very tightly. Thin- boundaried people, on the other hand, perceive a more permeable boundary between themselves and the external world, and create more flexible percepts and concepts. Thin- boundaried people are more likely to have hallucinations, and also more likely to have poor reality discrimination (Hartmann, 1988). Thin- boundariedness, I suggest, is the third major factor involved in the dynamics of reality discrimination and hallucination.

Based on several questionnaire and interview studies, Hartmann has shown that boundary thickness is a statistically significant method for classifying personalities. "Thin- boundaried" people tend to be sensitive, spiritual and artistic; they tend to blend different ideas together and to perceive a very thin layer separating themselves from the world. "Thick- boundaried" people, on the other hand, tend to be practical and not so sensitive; their minds tend to be more compartmentalized, and they tend to see themselves as very separate from the world around them. The difference between thin and thick- boundaried personalities, I suggest, lies in the "minimum cohesion level" accepted by the cognitive ends of PCL's. Hartmann himself gives a speculative account of the neural basis of this distinction, which is quite consistent with these ideas.

4. A Theory of Hallucination

Now we are ready to tie together the various threads that have been introduced, into a coherent theory of the interrelation between reality discrimination, hallucination, and boundary- drawing. I will argue that thin-boundariedness provides the initial impetus for both poor reality- discrimination and hallucination, which then go on to support and produce each other via a relation of positive feedback (Goertzel, 1994).

First, consider the thin-boundariedness for reality discrimination. As compared to an average person, an exceptionally thin-boundaried individual will have PCL's that place less rigid, more flexible boundaries around mental entities. This implies that external, real-world information will be stored and conceptualized more similarly to internal, imagined information. Such a person will naturally have more difficulty distinguishing real stimuli from non- real stimuli -- a conclusion which is supported by Hartmann's data. Thin- boundariedness is bound up with poor reality discrimination in a direct way. Next, to see the direct relationship between thin- boundariedness and hallucination, it is necessary to consider the iterative nature of PCL's. The key point is that a PCL does not merely classify data, it constructs data. It is responsible for developing stimuli into percepts in particular ways. Thus, if it has judged a certain stimulus to be "real," it will develop it one way; but if has judged the stimulus to be "imaginary," it will develop it another way. In a thin-boundaried person, there is less likely to be a rigid distinction between different ways of developing stimuil into percepts; so it is much more likely, that as a perceptual-cognitive loop iterates, internal stimuli will be developed as if they were external stimuli. What this means is that thin-boundaried people will be more likely to have internal stimuli that are as vividly and intricately developed as external stimuli -- i.e., as suggested by Hartmann's data, they will be more likely to hallucinate.

So, thin- boundariedness has the ability to lead to both hallucination and poor reality discrimination. We have already suggested, however, that the latter two traits have the propensity to support each other by positive feedback. Poor reality discrimination causes hallucination, according to the mechanisms identified by Bentall and Slade. And hallucinations may cause poor reality discrimination, by giving the mind a more confusing data set on which to base its categorization of internal versus external stimuli. The view which emerges from these considerations is consistent with the fifth possible relationship listed in Section 2: "A and B may cause each other, once initially activated by some other factor C." Thin- boundariedness sets off a process of mutual activation between the traits of poor reality discrimination and hallucination.

One might wonder whether the sixth relationship is not the correct one. Perhaps thin- boundariedness is itself produced, in part, by hallucinations or poor reality discrimination. But this conclusion does not seem to hold up. There is no clear causative link between hallucination and thin- boundariedness.

And the polarity of the influence of poor reality- discrimination on boundary- drawing seems quite open. In some individuals, poor reality discrimination might cause excessive thin- boundariedness; but in others, it might instead cause excessive thick- boundariedness. In yet others, it might do neither. The question is whether an individual tends to err on the side of making external stimuli "loose" or making internal stimuli "tight" - - or errs in a non- biased, neutral way.

5. Conclusions

In conclusion, we have proposed an alternate explanation of the correlation between hallucination and reality discrimination. Our explanation differs from that of Bentall and Slade in that it posits a circular causality between the two factors, initiated by a third factor of thin- boundariedness. While the Bentall and Slade theory is simpler, the present theory is psychologically more plausible, and does greater justice to the complexity of the human mind.

The question arises whether the present theory is empirically distinguishable from the Bentall and Slade theory. Such a distinction, it seems, could be made only by a study of the development of hallucinations in individuals prone to hallucinate, say schizophrenic individuals. The Bentall and Slade view, in which poor reality discrimination causes hallucination, would predict that poor reality discrimination should precede hallucinations, and should not increase proportionately to hallucinations. The present theory predicts that the two factors should increase together, gradually, over the course of development.

Regarding the clinical implications of the Bentall and Slade theory, the present study is equivocal. They argue that individuals prone to hallucination should be explicitly trained to distinguish internal from external stimuli. According to their theory, this should lead to a reduction of hallucinations. According to the present theory, the reduction of hallucinations would be one possible outcome of such training, but certainly not a necessary outcome. The clinical conclusions of Bentall and Slade emerge from our rethinking basically unmodified, but rendered slightly more dubious.


Abraham, Fred, Abraham, Ralph and Chris Shaw (1991). A Visual Introduction to Dynamical Systems Theory for Psychology. Santa Cruz: Aerial Press.
Bentall, R.P. (1990). The Illusion of Reality: A Review and Integration of Psychological Research on Hallucination. Psychological Bulletin v. 107(1), 82- 95
Bisiach, E. and A. Berti (1987). "Dyschiria: an Attempt at its Systematic Explanation," in Neurophysiological and Neuropsychological Aspects of Spatial Neglect, Ed. by M. Jeannerod, North- Holland, Amsterdam
Crook, J. and T. Rabgyas (1988). The Essential Insight, in Asian Contributions to Psychology, Ed. by A. Paranjpe, D. Ho and R. Rieber. New York: Praeger.
Feelgood, S. and A. Rantzen (1994). Auditory and Visual Hallucinations in University Students, Personality and Individual Differences 17- 2, 293- 296.
Frith, C.D. (1979). Consciousness, Information Processing, and Schizophrenia. British Journal of Psychiatry 134, 225- 235.
Goertzel, B. (1994). Chaotic Logic: Language, Thought and Reality from the Perspective of Complex Systems Science. New York: Plenum Press.
Hartmann, Ernest (1991). Boundaries in the Mind. New York: Basic Books.
Prinzmetal, W., D. Presti and M. Posner (1986). "Does Attention Affect Feature Integration?", J. Exp. Psychol.: Human Perception and Performance 12, 361- 369
Rumelhart, D. and J. McClelland (1986). Parallel Distributed Processing. Cambridge, MA: MIT Press.
Slade, P. (1994). Models of Hallucination, in The Neuropsychology of Schizophrenia, ed. by A. Bard and J. Curtis, pp. 245- 253.
Slade, P. and R. Bentall (1988). Sensory Deception: A Scientific Analysis of Hallucination. London: Croon Helm.
Treisman, A., and H. Schmidt (1982). "Illusory Conjunctions in the Perception of Objects," Cognitive Psychology 14, pp. 127- 141
Umilta, C. (1988). "Orienting of Attention," in Handbook of Neuropsychology v.1, Elsevier, NY