20th WCP: The Role of Mental Variation in Cognitive Science: Structured Imagination and Conceptual Combinations

(1) Preliminary Remarks

(a) Introduction

What is the role of mental variation in cognitive science? I will attempt to answer this question, as it often happens in philosophy, by dividing it into two separate questions: (1) What role does mental variation already (implicitly) play in cognitive science? and (2) Would cognitive science benefit by inquiring (explicitly) into the role of mental variation?

I will attempt to show that mental variation already plays an important, although not always explicit, role in cognitive science. Additionally, I will suggest that explicating the role of mental variation in cognition may be seen as a vital component of maintaining the strength of certain approaches and "schools" of cognitive science. To illustrate this, let me present the relevance of variation in disputes between the connectionist and more traditional theories of cognition.

The dispute between more traditional (propositional and "language of thought" based) understanding of cognitive science and recent connectionist theories may be seen as centred, in part, on the role of variation (manipulation) in cognition. More traditional account would appeal to the fact that the correspondence between formal languages and propositional content may be seen as dependent upon the combinatorial structure of strings of symbols. Thus, one may claim that sentences have parts with both fixed and adjustable meanings and that we form new meaningful sentences by rearranging these parts in new combinations. In this case, the ability to vary parts of structured sentences may be seen as quite significant: this ability would help us account for our understanding of propositions we never encountered before.

Instead of appealing to the paradigms of cognition which involve either strings of symbols or some sort of propositional structure, connectionists would rather appeal to some type of neural architecture. They argue that the properties of some sort of functional networks in cognition represent the properties of neural activity much closer than the properties of any sequential symbol-processing system. The problem of variation, however, did not go away. It still remains to be solved, only transferred to the level of motor processes and image transformation.

(b) Mental variation, Structured Imagination, and Conceptual Combinations

I want to begin my treatment of structured imagination and conceptual combinations in cognitive science by noting that one would not encounter the notion of mental variation directly and explicitly if one were to inquire into either of them. However, I would like to argue that the way in which cognitive science understands "structured imagination" and "conceptual combinations" implies tacit use of principles and structures of a complex cognitive strategy — which has been much used in phenomenology, logic, and general epistemology namely, the strategy of mental variation.

It should be noted, however, that I am not claiming that inquiry into "free variation" by phenomenologists should be seen as identical to the inquiry into "structured imagination" and "conceptual combinations" by cognitive science. There are notable differences between the three notions. Nevertheless, behind both of the notions there is, I believe, an elementary and primitive activity of the mind, namely, the activity of mental variation. This activity can be described in simple terms: varying, in one's mind, some aspects or elements of a compound structure, while keeping some other aspect or elements of that structure constant.

In the course of examination of this activity, we will see that neither the underlying structure nor the elements being varied need always be specified or made explicit in thinking. On the contrary, the activity of mental variation appears to be a mostly automatic and unconscious mechanism both in real-life and in experimental circumstances.

Before I present the notion of structured imagination as developed in Creative Cognition by Finke, Ward and Smith, let me sketch my basic understanding of contemporary cognitive science. Cognitive science can be characterized as a relatively new field of research, straddling the disciplines of philosophy, psychology, computer science, linguistics, neurology, and neuropsychology. Cognitive science, in general, attempts to test and advance the "functional" and "computational" models of thought. As Johnson-Laird remarked, in The Computer and the Mind, the advance of the "age of computability" has resulted in the following conjecture: "Perhaps the mind stands to the brain in much the same way that the program stands to the computer" (Johnson-Laird 1988, 8). This conjecture, based solely on a powerful and heuristic analogy, has led to a large number of computational theories about various functions and properties of the mind.

In order to achieve its primary goal, the scientific explanation of mental phenomena, cognitive science adopts the naturalist and empiricist approach to mental phenomena. Alvin Goldman, in his Philosophical Applications of Cognitive Science, clearly and succinctly sums up the empirical grounds of cognitive science: "How people ordinarily represent or understand mental concepts (a central problem in philosophy of mind) is ultimately an empirical problem requiring theoretical and experimental investigations appropriate to cognitive science" (Goldman 1993, 63).

(2) Structured Imagination and Mental variation

The first problem that cognitive science encounters in its treatment of imagination is conceptual or definitional. In other words, cognitive science needs to sketch the limits of the concept of imagination. In order to do this, cognitive scientists adopt, overtly or covertly, several assumptions about imagination. Here are some of the more important characteristics ascribed to imagination in Creative Cognition:

(1) Imagination involves the generation and experiences of ideas that go beyond what is currently known (114). However, it is important to add that the imagination does not go outside the bounds of known reality.

(2) Imagination results in some tangible product — in contrast to fantasy, dreaming, and so on.

(3) Imagination is not a mysterious process through which new ideas spring into being full blown and with no obvious link to existing cognitive structures. On the contrary, imagination is influenced by existing knowledge frameworks. That is why the authors of Creative Cognition use the term "structured imagination."

(4) Imagination is different from mental imagery in being both more global and more restrictive than mental imagery. This is how the authors of Creative Cognition put it:

Imagination is the process by which people mentally generate novel objects, settings, events, and so on. It is more global than mental imagery in the sense that although these imagined entities might take on the form of mental images, they need not. Imaginative products can also exist in the form of verbal description. Imagination is also more restrictive than mental imagery in the sense that it must involve the generation of something new, whereas certain manifestations of mental imagery can be purely recollective (Finke, Ward, Smith 1993, 115).

As we have seen from this sketch, the understanding of imagination in Creative Cognition is based on several assumptions. The most fundamental one, however, is the assumption that imagination is necessarily productive. Once this assumption has been made, the others follow naturally. For example, the claims that imagination needs to be innovative and that imagination has the tendency to be predictable can both be seen as a consequence of the assumption that imagination is productive.

These assumptions about imagination are readily justifiable since they stem from general methodological requirements. Cognitive science relies on empirical data and it needs certain testing procedures in order to theorize about imagination. Research involving empirical data is only possible if there are some empirically testable phenomena "out there." However, one may remark that insistence on empirical data in theorizing about imagination should not be attributed exclusively to cognitive science but to any other philosophical or theoretical approach concerned with imagination. Consequently, one may express doubt about any serious sense in which imagination could be taken as non-productive. In order to illustrate at least some senses in which imagination can be seen as non-productive, let me elaborate the two cases: imagination in its "adverbial" usage and imagination in the "constructivist" sense.

(a) Some Alternative Views on Imagination

One usage in which imagining is not seen as a purposive action with tangible results could be defined in this way: one could understand imagining not as a verb but as an adverb. Instead of talking about an activity of "imagining" or "imagination" per se one would see actions, events, designs, and writings as produced, induced, or brought forth "imaginatively." For example, one would be able to say that a solution to a mathematical problem has been worked out imaginatively, or that an ice skater has performed some of the necessary moves imaginatively.

A generation ago, some philosophers might have argued that the adverbial use of "imaginatively" is the core or central use. "Ordinary language" philosophers, influenced largely by Wittgenstein's later writings, refused to acknowledge the existence of any intrinsic qualities, states, or dispositions in human cognition. Influenced also by behaviourist approach to human psychology, ordinary language philosophers tended to characterise human cognition by the effects it produces in human behaviour. Consequently, the activity of "imagining" would, properly speaking, not exist. There would only be certain actions or events brought forth in an "imaginative" manner.

Take, for example, Ryle's discussion of intelligence in The Concept of Mind. In this book, Ryle criticizes the belief that there is "intelligence" behind one's "intelligent behaviour." According to Ryle, there are neither special prescriptions for intelligence nor is there a special place for intelligence in one's mind. This is how he characterizes a person who reasons logically:

He reasons with a correct method but without considering the prescriptions of a methodology. The rules that he observes have become his way of thinking, when he is taking care; they are not external rubrics with which he has to square his thoughts (Ryle 1949, 48).

Ryle proposes that to say that a person argues intelligently is to adjudge that person by certain criteria (such as being cogent, clear, and so on). Silent argumentation falls into the same category, although it is not subjected to the criticism of audience and readers. Furthermore, argumentation is always unique: the argumentative setting, the particular issues are never the same.

Imagination, understood in this sense, would be much more difficult to describe and analyze than the notion of structured imagination. Additionally, it would be very difficult to see what the products of such notion of imagination would be. What is the problem with attempting to square Ryle's and cognitive science approaches to human intelligence and creativity? A misunderstanding or some sort of a category mistake?

Instead of pointing out the shortcomings of "ordinary language" understanding of imagination, it is much more important, I believe, to point to a change in methodology. This change is implicit in most non-productive interpretations of imagination. Such interpretations require that the methodological approach becomes normative much more that theoretic and descriptive. In other words, imagination would have to be oriented towards an implicit or explicit norm, or the basic criterion, for ascribing the term "imaginatively" to actions and events instead of relying on products of the process of imagination.

Another essentially non-purposive and non-productive approach to imagination can be found in thought of Nelson Goodman and Jerome Bruner. Bruner follows Michael Halliday in dividing the functions of language into two superordinate classes — pragmatic and mathetic. The pragmatic class contains functions such as the instrumental, regulatory, interactional, and personal. The mathetic class contains the heuristic, imaginative and informative functions. The mathetic class is an interesting one for the purposes of this chapter. The heuristic function of language is to gain information and correction from others. The informative function of language presupposes intersubjectivity; that somebody has knowledge I do not possess and vice versa, and that there are ways of communicating this kind of knowledge. The imaginative function of language is defined by Bruner as "...the means whereby we create possible worlds and go beyond the immediately referential" (Bruner 1986, 125). Bruner's understanding of imagination and its link to possible worlds has been largely inspired by Nelson Goodman's understanding of possible worlds.

Goodman argues that what we call the "world" is a product of a mind whose symbolic procedures construct it. In the case of perception, for example, it has often been claimed in philosophy that the world of appearance, the very world we live in, is "created" by the mind. Imaginative creation of possible worlds would be, therefore, just one of the means of constructing reality, although an extremely important one.

In this second case we see that the notion of imagination appears integrated into a much broader world view, one which relies on numerous metaphysical assumptions. The role of imagination in this world view is undoubtedly significant. However, this kind of imagination lacks precise definition and cannot be empirically tested. If it became abstracted from its metaphysical background, this understanding of imagination would lose most of its meaning. Additionally, since the focus of this chapter is variation and some other related concepts in cognitive science, it would be difficult, at the same time, to keep this focus in mind and to retain Goodman's "large metaphysical picture."

(b) Structured Imagination Empirically Tested

The authors of Creative Cognition began their examination of structured imagination largely influenced by the research in structured imagination conducted by cognitive psychologists. This research revealed that attributes of real-world categories, instead of co-occurring randomly, tend to occur in distinct clusters. "For example, feathers and wings occur much more often together than do fur and wings" (Finke, Ward, Smith 1993, 118). The traditional tests indicated, additionally, that correlated attributes are so inherent to human cognition that even infants as young as ten months of age appear sensitive to them.

Thomas Ward conducted a number of experiments in 1991, results of which were presented in a paper under the title "Structured Imagination: The Role of Conceptual Structure in Exemplar Generation" to the meeting of the Psychonomic Society in San Francisco. Ward's research was largely integrated into the chapter "Structured Imagination" in Creative Cognition. What follows is a short description of the first of Ward's several studies:

In the first study, Ward gave fifty college students the task of imagining a planet that exists somewhere in the galaxy but is similar to earth in size, terrain, and climate. They were then asked to imagine and draw an animal that lived on that planet. Because the planet was described as similar to earth, the properties of the exemplars served as a baseline against which to compare exemplars from subsequent studies in which that planet was described as being very different from earth. In addition to providing the initial drawing of their imagined animal, subjects responded to questions about the diet, habitat, sensory organs, and appendages of the creature. The questions were designed to provide information about nonvisible properties of the creatures and to help clarify ambiguities in interpreting the visible properties of the drawings (Finke, Ward, Smith 1993, 119).

The results of the study indicated that a large majority of subjects drew creatures with highly predictable features, such as bilaterally symmetric individuals with at least one major sense organ and at least one major type of appendage. In other words, the way in which the tested subjects used their imagination indicated a strong influence of familiar and characteristic features implicitly correlated to certain categories:

Rather than being idiosyncratic and unpredictable, the use of imagination to generate new exemplars of a category appears to be highly structured by the characteristic attributes of known category members (Finke, Ward, Smith 1993, 120).

Furthermore, Ward showed that subjects could be expected to structure their mental variations in imagining, both within and between categories. For example, subjects would typically vary shape across species but not within species. In simple words, a fish, imagined or drawn on paper, would appear to have an entirely different shape from a bird. However, once a fish-like creature has been chosen, the extent of variations within the chosen species becomes greatly limited. Finally, if the subjects were asked to generate a second member of the same species, it is most likely that they would vary just its size.

Categorization models are important not only in recognizing categories but also in generating new members of a given category. The person appears to evaluate an entity against some stored representation of the category in deciding whether the given entity is a member of some known category.

However, categorization may depend on the initial set of requirements or, in other words, the initial restraint parameters. To illustrate this, consider what I have termed interspecies and intraspecies variation. One might begin the process of mental variation with a broad, abstract category. For example, one might begin with the more general category of an animal, "mammal," which is associated with a number of salient properties and generate a fitting concrete exemplar by varying different kinds of animals against this category. One may produce a dog, a cat, a cheetah, etc. by this kind of variation. Finally, one might produce a novel kind of animal by combining some known properties of mammals. This kind of categorization might be termed categorization by inter-species variation. Second, one might begin with a concrete example of an animal and mentally vary some of its parts. For example, one might produce different breeds of a species "dog" varying some of its parts: ears, tail, hair length, etc. Combining the known breeds, one may produce a novel breed of dog. This kind of categorization might be termed categorization by intra-species variation.

Schemes are even more complex knowledge structures than categories. They have also been called "scripts," "frames," and "concepts." Categories, in contrast, consist of single kinds of objects that share certain taxonomic links. People recognize that more abstract categories might be divided in various ways, following different taxonomic features. Thus we speak of time-measuring devices, among which are sundials, hour-glasses, clocks, watches, and stop watches. Notice that there may be very little actual similarity between the objects that fall under the abstract category of time-measuring devices. Moreover, hour-glasses and watches, for example, may alternatively be defined as each falling under separate and distinct categories, e.g. the former under glass art objects and the latter under mechanical devices.

Schemes are even more complex entities, and somewhat more arbitrary structures then abstract categories. They tend to straddle taxonomic boundaries. This is how Creative Cognition contrasts schematic and categorical thinking:

In contrast, schemes specify the relations among several discrete categories, and these often cut across standard taxonomic boundaries. A schema for a living room might contain not only chairs, televisions, paintings, and other discrete categories but also the typical relations among them (Finke, Ward, Smith 1993, 134).

Cognitive science makes use of schemes in the following manner: plausible inferences and problem solutions are generated by means of schemes, or clusters stored in memory. These clusters can be accessed as large units which contain information about the likely properties of the environment.

Schemes, in general, have an important impact on encoding new information, retrieving old information, and inferences based on that information. Schemes also guide and limit imagination. For example, a characteristic hat and pipe still constitute a "Sherlock Holmes" schema. It is interesting to note that neither of the two objects, in the form in which they exist in common perception, have been introduced by Sir Arthur Conan Doyle. They are the product of the imagination of subsequent Sherlock Holmes illustrators. However, generations of readers of Sherlock Holmes have had their imagination shaped by such a schema. Schemes are stubborn; it usually happens that people form exception rules for cases that do not fit existing schemes rather than change the more general schema.

Ernst Mach wrote about the inherent need for "economy of thought" in human cognition which is particularly evident in scientific thought experiments. This general regulatory principle requires that people employ the maximum number of typical, routine problem solving techniques even in ordinary thinking processes. However, people are often faced with atypical events and objects. "In everyday life we are often faced with inputs that do not readily fit any prestored category representation" (Holland, Holyoak, Nisbett, and Thagard 1986, 13). In cases like these, people introspect the novel situation and solve the problems by the simultaneous activation and integration of multiple schemes.

However, the role of schemes in cognitive processes is by no means exhausted by problem solving. Schemes may, for example, be employed to serve a humorous function. Cartoonist Gary Larson has created funny situations in his "Far Side" series by juxtaposing mutually exclusive, or extremely distant schemes.

(d) Structured Imagination and Variation

Let me now turn to what I believe to be one of the most important feature of structured imagination, namely, the activity of mental variation. Most of the experiments concerning structured imagination in Creative Cognition employ the following strategy: the subjects were asked to extend a property, for example "intelligence," of a creature living on earth to some, slightly altered, imaginary instances of that creature living somewhere in the universe. It appears that the subjects were supposed, although tacitly, to produce the creatures which would be different from humans to some degree. In other words, the subjects were asked to imagine variants of the known and familiar category "intelligent creature living on earth." Thus, the process of mental variation kept most of the typical properties of "intelligent creature" or "human being" constant, while placing this imaginary creature, in the process of variation, in a slightly altered environment: "living on a planet similar to earth." There was a degree of freedom, of course, left to decide how and where to implement variation in imagination. First, the concrete features of this imaginary planet were not specified — the only thing specified was its apparent similarity to earth. Second, the typical features of the earthly human creatures to be kept constant in imagination were not specified. One could keep certain typical human features constant and vary some others. For example, one could choose to attribute to this imagined creature two human-like legs and to give it several antennas instead of arms. However, the subjects of those experiments seem not to have exploited this freedom of variation in structured imagination to a great degree. They have largely chosen to keep most typical features associated with "intelligent creatures" constant by default processes of cognition, as it were.

Cognitive science, in contrast to some other styles of philosophy, is practical and theory-oriented. Its understanding of imagination involves the notion of intervening in the process of mental variation. This intervention serves the role of rendering the products of mental variation interpretable and useful to the process of theory formation. Consider, for example, varying some aspects of the notion "car." A cognitive scientist would almost certainly intervene in the process of mental variation with two basic motives in mind.

First, a cognitive scientist would strive to better structure the process of mental variation, limiting it by some additional constraints. Second, a cognitive scientist would primarily set a process of mental variation in a manner such that it results in a tangible product, e.g. in drawing.

If a cognitive scientists had to deal with subjects imagining anything about "car" their procedure would be quite different from the phenomenological one. The subjects would likely be asked, for example, to imagine a planet similar to earth where human-like intelligent creatures "see" things by using ultrasound. Their vision does not depend on light. What would their cars look like? Subjects would probably keep, in their imagination, constant most of the known properties of cars, such as engine, wheels, passenger and driver seats, etc. However, they would probably, in their imagination, remove glass windshields and windows and replace them, for example, with some sort of ultrasound "ears" or radars placed on the front and the back of that car. To make this product of the imagination more intuitive, one could perhaps call it "batcar." Finally, the final product of this process of mental variation would be drawn on paper. Consequently, the results of such an experiment may be used as material for the statistical analysis and interpretation.

In this short discussion I have indicated, I believe, close dependence between "structured imagination" and "mental variation." It appears that structured imagination in cognitive science makes use of mental variation as a vehicle, or as a primitive cognitive procedure. In the next section, I will attempt to show that the process of "conceptual combinations," as presented in cognitive science, makes use of mental variation in a manner similar to what structured imagination does.

(3) Mental variation and Conceptual Combinations

(a) Introduction

Inquiry into combining of concepts in cognitive science has been prompted by the attempts to characterize the structure of conceptual knowledge in human behaviour. The problem to be solved has largely consisted in giving an account of the employment of what were considered relatively static pockets of information (definitions, schemes, frames) in greatly varying contexts. These packets of information, however, appear not only capable of fitting the diverse contexts but the concepts within the packets of information have the capacity of being combined in various manners. This latter phenomenon just adds to complexity and dynamic nature of the human conceptual knowledge which emerges in the concrete process of cognizing.

In order to illustrate the approach which cognitive psychology adopts in order to inquire into conceptual combinations I will examine some issues presented in the paper "Context and Structure in Conceptual Combination" by Douglas L. Medin and Edward J. Shoben, though I will not go into the particular theoretical background which this paper seems to be addressing. I will focus on the extent to which inquiry into conceptual combinations by cognitive psychology makes use of the basic and primitive cognitive operation, that of mental variation.

In cognitive psychology, conceptual combinations are the best demonstration of the basic need for concept flexibility:

One of the basic properties of the categories is that they can be combined to form more restricted concepts. Thus, we can restrict birds to songbirds, water birds, predatory birds, and migratory birds. In addition to these combinations, we can construct and understand novel combinations such as gregarious birds or noisy birds (Medin and Shoben 1988, 160).

Cognitive psychologists, striving to give an empirical account of human conceptual knowledge using the computational models of explanation, have to grant that it is extremely unlikely that people have a stored representation for every possible conceptual combination. This is not just because there would be too many possible combinations but also because the pairing of terms involves some sort of augmentation of their initial "semantical mass." For example, the pairing of adjectives and nouns does not involve mere conjunction of attributes of two constituents. In Medin and Shoben's words:

For example, although most friendly people are warm and outgoing, it seems silly to describe a computer in this way. Moreover, in conceptual combinations such as pet fish, the constituents often have competing values on some dimensions; most pets have fur as their body covering, yet pet fish have scales (Medin and Shoben 1988, 160).

However, in order to understand how people structure and remodel categories and schemes, one seems compelled to examine conceptual combinations. One particular case combining the elements of language and cognition, that Medin and Shoben chose to examine in their article, is the combination of adjectives and nouns.

Research in cognitive psychology has already indicated that concepts are generally organized on the basis of their typicality which, in turn, is based on common similarity standards. For example, pictures of typical examples (e.g. robins) are identified more rapidly when preceded by a category label (e.g. bird). When it comes to combination of nouns and adjectives, one of the models adopted in order to account for conceptual combinations in cognitive psychology was the modification model. This is a short description of the modification model:

The modification model postulates a pair of simple mechanisms by which the meaning of a conjoined category can be derived from the individual meanings of the adjective and noun constituents. The basic idea is that the adjective directs the knowledge restructuring in a straightforward manner by restricting the range of acceptable values and by increasing the importance of the corresponding dimension. If one considers the color of an apple, for example, one notes that most apples are red, but that some are green or yellow and that a few are brown. According to the modification model, for the conjoined category red apple, the acceptable values are restricted to red and the dimension of color is given more weight or importance than it has in the simple concept of apple. The typicality of potential instances of the red apple will be a function of their similarity to the newly created prototypic red apple (Medin and Shoben 1988, 161).

However, Medin and Shoben argue that models that attempt to explain combined categories by adding or changing a single feature have largely been unsuccessful. It appears that attributes are interdependent and that correlations of the attributes are a matter of common knowledge and largely dependent on the context. It seems that one of the most difficult hurdles to any "single feature modification model" of explanation of conceptual combinations is solving the problem of centrality. Let me present this problem in the next section of this chapter.

(b) Centrality and Conceptual Combinations

Medin and Shoben approached the problem of centrality in conceptual combinations in the following way: they generated concept properties that would always or nearly always be true of two different concepts but perhaps more important in one concept than another. Here is their own account:

For example, we felt that the shape of a basketball was central to the meaning of the concept. If a basketball is not round, then it cannot be a basketball because it cannot be dribbled, passed, or shot in the manner that is characteristic of the game. On the other hand, we believed that shape was relatively less important for a cantaloupe. Although square cantaloupes never occur in nature, by our intuitions there is nothing essential about the shape of a cantaloupe (Medin and Shoben 1988, 175-176).

The shape of a basketball appears to be central to the meaning of this concept. One may claim that its weight is central to the notion of a basketball to a somewhat lesser degree. For example, one can imagine a statue of a basketball player holding an iron or brass basketball in his hand. This kind of a ball would be too heavy in order to be actually played with. However, one can still grant that this objection does not affect its status of being a basketball. If one were to continue such queries in one's mind, it would be possible to establish that the colour of the ball probably figures as one of the least central of all the potential properties of the basketball. All those different "perspectives" of mental variation may serve the purpose of illustrating the possibility of disclosing the inherent degrees of centrality of properties throughout various concept combinations.

Consequently, it soon becomes clear that centrality is almost exclusively tested in the mind, by the process of mental variation. In this particular case, examined by Medin and Shoben, of combining adjectives and nouns it appears that there is some kind of a structure among attributes that guides reasoning by means of adjective-noun pairs:

Consider again flying whales versus flying ostriches. For flying whale, one must make changes to a large number of attributes. One must add some kind of wings, slim down the body considerably, and find some means of propulsion to get our redesigned whale into the air. In contrast, a flying ostrich requires fewer and less drastic changes. Shortening the legs and neck, for example, might very well enable this hypothetical ostrich to fly (Medin and Shoben 1988, 179).

In this example, one can see that the extent to which mental variation ("making changes to attributes" in Medin and Shoben's words) establishes the level of typicality embedded in certain concepts and discloses the regularities inherent in the way certain concepts "hang together."

(4) General Summary

As I attempted to show in this paper, inquiries into concept combinations in cognitive science revealed and thematised the importance of context and background knowledge in human cognition. The research in structured imagination disclosed the importance of known properties and background information in imagining novel entities.

However, the main goal of my paper was to provide plausibility to the claim that in testing, analysing, and disclosing the regularities both in the context combinations and the process of structured imagination cognitive psychology employed a procedure I defined as mental variation.