«Material anchors for conceptual blends Edwin Hutchins Department of Cognitive Science, University of California, San Diego, 9500 Gilman Drive, La ...»
Journal of Pragmatics 37 (2005) 1555–1577
Material anchors for conceptual blends
Department of Cognitive Science, University of California, San Diego,
9500 Gilman Drive, La Jolla, CA 92093-0515, USA
Received 1 March 2003; received in revised form 19 September 2003; accepted 28 June 2004
Reasoning processes require stable representations of constraints. There are two principal ways to
achieve stability in conceptual models. First, the conceptual models that anthropologists call cultural models achieve representational stability via a combination of intrapersonal and interpersonal processes. Second, the association of conceptual structure with material structure can stabilize conceptual representations. This is an old and pervasive cognitive strategy. Conceptual blending theory provides a useful framework for considering the joint contributions and mutual constraints of mental and material structure. Projecting material structure into a blended space can stabilize the conceptual blend. I call an input space from which material structure is projected into a blend a ‘material anchor’ for the blend. The term material anchor is meant to emphasize the stabilizing role of the material structure. In this article, I will present and discuss a number of examples of materially anchored blends, which depend to different degrees on material structure. Materially anchored blends vary on a number of complexly related dimensions, including the extent to which the blend relies on the presence of material structure in the perceptual ﬁeld, the complexity of the material structure, and whether the material structure was designed to support the blend or is used opportunistically.
# 2005 Elsevier B.V. All rights reserved.
Keywords: Blending; Cognitive artifacts; Distributed cognition; Integration; Material anchors
1. Introduction In this article I discuss a general and ancient human cognitive phenomenon, the association of conceptual structure with material structure. Examples range from the linguistic construction of ﬁctive motion, to the cultural construction of queuing or standing E-mail address: email@example.com.
0378-2166/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.pragma.2004.06.008 1556 E. Hutchins / Journal of Pragmatics 37 (2005) 1555–1577 in line; from children’s games to essential tools of life-or-death endeavors; from the conceptualization of simple everyday logical distinctions to the tools of science and mathematics. All of these seemingly disparate phenomena are instances of a particular way of associating conceptual structure with materialstructure.
The ability to combine conceptual structure with material structure is a key cognitive strategy. The basic abilities are very old. In fact, non-human primates may be capable of some forms of conceptual and material combination. What are the processes that can do this? What varieties of combining are possible? What are the cognitive consequences of making such associations?
I have found the framework of conceptual blending (Fauconnier and Turner, 2002) to be especially useful for thinking about the relations between conceptual and material structure. It allows us to take a uniﬁed view of phenomena that would have otherwise been considered quite disparate and frames the relations among conceptual and material structures in ways that make it possible to answer some of the questions posed above.
A few words about what this approach is not. Material structure can have many kinds of relations with conceptual structure. One of the most common is the relationship of representation. All of language depends on this sort of relationship so this one comes easily to mind for linguists. Spoken or written structure in the world represents conceptual entities. This immediately opens the door to the ﬁeld of semiotics and the huge literature on representations. But before jumping directly to signs and symbols it is useful to examine some more basic processes by which meaning is attached to or projected onto material structure. The relation we are looking for here is not quite the same as the ‘representation’ function described in the preceding phrase. I will return to this topic at the end of the article.
In order to see this other sort of relation, it is necessary to introduce the basics of conceptual blending. I will leave it to other authors in this volume to present the details of that theory.
Fauconnier (1997) describes the essence of the process as follows:
A conceptual blend operates in two input mental spaces to yield a third space, the blend. Partial structure from the input spaces is projected into the blended space, which has emergent structure of its own. The emergent structure arises in three
Composition: Taken together, the projections from the inputs make new relations available that did not exist in the separate inputs. Completion: Knowledge of background frames, cognitive and cultural models, allows the composite structure projected into the blend from the inputs to be viewed as part of a larger self-contained structure in the blend. The pattern in the blend triggered by the inherited structures is ‘completed’ into the larger, emergent structure. Elaboration: The structure in the blend can then be elaborated. This is called ‘running the blend.’ It consists of cognitive work performed within the blend, according to its own emergent logic.
(Fauconnier, 1997: 150–151) The cases collected in this article show how material structure provides an input space that is blended with other inputs. The principles of composition, completion, and elaboration will ﬁgure prominently in these examples. Conceptual blending theory makes E. Hutchins / Journal of Pragmatics 37 (2005) 1555–1577 1557 Fig. 1. A conventional conceptual blend (A), and a conceptual blend with a material anchor (B).
it clear that the material input space need not be a representation of the concepts in the input space with which it is blended. Fig. 1 illustrates these relations in schematic form. Fig. 1A is a depiction of a conventional conceptual blend with two conceptual input spaces and a blended space. Fig. 1B introduces a new notation.1 The square around input space 2 indicates that material structure is contributed by that input space to the blended space.
2. The problem of conceptual stability
Thinking processes sometimes involve complex manipulations of conceptual structure.
Conceptual structure must be represented in a way that allows some parts of the representation to be manipulated, while other parts remain stable. The complexity of the manipulations of structure can be increased if the stability of the representations can be increased. The stability of the representations is a necessary feature of the reasoning process, but it is often taken for granted. The need for representational stability becomes more visible in circumstances where the necessary stability is not present.
3. Culturally based reasoning
In the 1960s and 1970s a movement arose in cognitive psychology investigating the biases and limitations of human reasoning processes (see Kahneman et al., 1982). For example, the famous four-card task, originally developed by Wason (1966), demonstrated that otherwise intelligent subjects (university undergraduates) had surprising amount of difﬁculty reasoning about logical relations. This study triggered a large number of followup studies of reasoning and inference. Of these, one clearly showed the role of a cultural heritage of mental models in the abilities of individuals to reason correctly. D’Andrade (1989) considered a number of logical isomorphs consisting of a premise in the form of an This notation was ﬁrst used by Robert F. Williams in his thesis proposal to the Department of Cognitive Science, University of California, San Diego, in June, 2003.
1558 E. Hutchins / Journal of Pragmatics 37 (2005) 1555–1577 ‘if-then’ statement followed by an observation and a set of potential conclusions. For example, a subject is shown a premise, ‘‘If x is true then y is true’’ and then an observation, ‘‘We know that y is not true’’. The subject is then asked what can be inferred about x from the observation that y is not true. The choices are: ‘‘[It is the case that x is true.’’ (invalid conclusion), ‘‘[We cannot know if x is true or not.’’ (invalid conclusion), and ‘‘[It is the case that x is not true.’’ (valid conclusion). Only a small minority (about 15%) of subjects choose the correct conclusion.
This example illustrates an
premise. From D’Andrade’s perspective, what is important about the premise in this problem is that, for most subjects, it is not culturally coherent. That is, unless x and y are associated with particular known concepts, our culture has nothing in particular to say about the relationship between x and y. Contrast this with a
culturally coherent isomorph:
Given the premise, ‘‘If this is a garnet, then it is a semi-precious stone’’ and the observation, ‘‘This is not a semi-precious stone’’, most subjects correctly reject the conclusions ‘‘[This is a garnet.’’ (invalid conclusion) and ‘‘[We can’t know if this is a garnet or not.’’ (invalid conclusion), preferring the correct conclusion ‘‘[This is not a garnet.’’ Formally the two problems are identical. A formal reasoning system that can solve one correctly should be able to solve the other correctly.
In a series of experiments, D’Andrade showed that university students can draw correct logical inferences when such problems are expressed in terms of culturally coherent mental schemas. D’Andrade hypothesized that the ability to reason correctly with culturally coherent materials is due to the stability of the representation of the premises. Performing the modus tollens inference (If x then y; not x, therefore not y), as illustrated in the two examples above requires the subject to make two shifts in perspective. First from the ‘x’ case to the ‘y’ case, and then from the ‘y’ case to the ‘not y’ case.2 It appears that premises that lack cultural coherence cannot be held stably in memory while the transformations required to make the inferences are performed.
D’Andrade’s study indicates that one way to establish the stability of representation to facilitate reasoning is to express the problems in terms of familiar cultural models. Because they are shared, cultural models tend to be supported and reinforced by the behavior and thinking of others. Cultural models are also systematic in the sense that they exist in a complex nexus of models that mutually constrain one another. Most cultural models are closely related to many other models. This inter-linking contributes to the conservatism of cultural beliefs over time and to the stability of cultural models as resources for individual and group reasoning processes.
4. Material anchors for conceptual blends
There is another way to increase the stability of conceptual representations. Cultural models are not only ideas that reside inside minds, they are often also embodied in material Other theories of reasoning posit other operations, but in all cases, stable representation of the relation given in the premise is required and modus tollens inference requires more complex manipulations than modus ponens does.