But while several inventors may be working on the same reverse salient, there is no guarantee that they view the problem in the same way--or even identify the salient as the same problem. In order to understand the way each inventor views a problem such as the transmission of multiple messages, we will use the framework outlined in previous chapters. To refresh the reader's memory, we will use Edison's kinetoscope as an example.
(1) Mental models:
In developing his kinetoscope, or motion picture camera, Edison's goal was to do "for the eye what the phonograph does for the ear, which is the recording and reproduction of things in motion, and in such a form as to be both cheap, practical and convenient [by] photographing continuously a series of pictures occurring at intervals... in a continuous spiral on a cylinder or plate in the same manner as sound is recorded on a phonograph." (Josephson, 1959). Indeed, he intended to put his kinetoscope cylinder on the same shaft with a phonograph cylinder, in order to coordinate sound and pictures.
Now in what sense is this rough idea of Edison's a mental model? He could run alternatives for this system in his 'mind's eye', imagining how it might work. To select among alternatives and turn imagination into reality, Edison needed to rely on:
(2) Mechanical Representations:
Robert Fulton, inventor of the steamboat, argued that "the mechanic should sit down among levers, screws, wedges, wheels, etc. like a poet among the letters of the alphabet, considering them as the exhibition of his thoughts; in which a new arrangement transmits a new idea to the world" (Gorman, 1992, p. 47). If we substitute inventor for mechanic, and include the possibility that an inventor can transform the standard 'levers, screws, and wedges' into devices suited to her class of problems, then Fulton's quote contains the idea of a mechanical representation--a familiar component that an inventor can use repeatedly to create new designs. As Reese Jenkins noted, "Any creative technologist possesses a mental set of stock solutions from which he draws in addressing problems" (Jenkins, 1984, p. 153).
The drum cylinder Edison used in both the phonograph and the kinetoscope serves as an example: it became for Edison a standard solution to the problem of creating a smooth, continuous rotation. To interrupt this rotation to allow pictures to be shown in his kinetoscope, he used another mechanical representation: a double-acting pawl he had developed for use in stock tickers (Carlson, 1990).
The difference between a mental model and a mechanical representation is that the former is incomplete and represents a possible path to a solution, whereas the latter is embodied in a completed device that represents a solution to part of a problem. Why refer to this sort of a device as a representation? Because not all aspects of the solution embodied in the device are recoverable simply by studying the device--one also must know how the device is represented by the user (Norman, 1983).
Inventors can plug these mechanical representations into their mental models, thereby using a lower-level representation that is embodied in a device to fill in a higher level model. Several years ago I attended an extraordinary conference on invention organized by Robert Weber and David Perkins (Weber & Perkins, 1992), and I am indebted to them for drawing my attention to the idea of slots.
For Weber and Perkins, the fundamental representation is a frame:
An entity with slots in which particular values, relations, procedures, or even other frames reside; as such, the frame is a framework or skeletal structure with places in which to put things. The slot is a generalization of the idea of a variable. The frame then represents an object, event, or concept in which the slots are the defining characteristics; the values of the slots are the instantiations of the variables, attributes, relations or procedures. (Weber & Perkins, 1989)
The idea of a frame was developed as a way of translating representations into a computational form. A simple example of a frame is a tax form, which has slots like 'name' and 'number of dependents', each of which can contain different values, e.g., 'Sue Smith' and '2'. A more complex example would be a frame for dog. This frame would inherit some characteristics from the overall frame for mammal, including certain slots and values. For example, we might include a slot for 'means of propagation'. Under mammal, that means would be 'live birth'. As an instance of mammal, dog would inherit that slot and value. Dog would also have some unique slots. We might include a slot for 'type of breed', which would include values like spaniel or retriever; within each type of breed, there might be slots for particular breeds like Springer Spaniel or Golden Retriever. This example illustrates the flexibility of the notion of a frame; it gives us lots of room for creating different slots.
A mental model can be viewed as a kind of frame, but one that is more visual and kinesthetic than other more prepositional types of frames. Dog, for example, is more than an abstract concept; most of us have a visual mental model of a dog, perhaps based on our favorite dog. Depending on our individual mental models, certain breeds may look less 'dog-like' to some of us (Rosch, 1973). (A friend of mine was walking his deerhound one day, and a stranger asked if it was a goat or a llama).
Frames can be nested within frames. Similarly, mental models can be nested within mental models. An inventor can have a mental model of an overall system, and also a mental model of how a part of a system might work.
Like frames, one could imagine mental models being divided into slots. For example, Edison's mental model for a kinetoscope included a slot for a mechanism by which the pictures would be advanced continuously. Edison filled this slot with the drum cylinder from the phonograph.
Substituting different mechanical representations can lead to a transformation in the overall mental model. Edison's capable assistant William Dickson substituted a different mechanical representation for the drum cylinder; he used a tachyscope, a rapidly rotating wheel on which pictures could be mounted. Dickson managed to project these moving images on a screen and coordinate them with a phonograph so when Edison returned from a trip in March of 1890, he was greeted by a moving image of Dickson, raising his hat and saying, "Good morning, Mr. Edison, glad to see you back, I hope you are satisfied with the kinetophonograph." (Carlson and Gorman, 1990, p. 107).
Instead of embracing this solution, Edison ordered Dickson to abandon it. Dickson's adoption of a new mechanical representation had forced a change in the overall mental model guiding the development of the system. It was no longer a 'phonograph for the eyes', intended--like the early phonographs--for use in an individual viewing booth.
The Edison/Dickson example illustrates one of the ways in which mental models can become evident--when an inventor resists an alternate design. This kind of 'mental inertia' corresponds to the kind of 'confirmation bias' found in the early experimental simulations of scientific reasoning. In the last chapter, we noted that more recent research has focused on the advantages of confirmation. As Tweney & Chitwood point out, "In contrast to the usual focus on confirmation bias as a reflection of the limits of human cognition, the evidence suggests that a confirmation heuristic is one of the highly functional means by which knowledge is made possible" (Tweney & Chitwood, 1995, p. 235).
(3) Heuristics:
David Perkins (Perkins, 1997) uses the metaphor of wilderness to describe the search for a solution to an invention problem. In Perkins' view, the inventor's goal should be to transform a Klondike space, in which the inventor knows there is gold somewhere in the wilderness but is not sure where, into a homing space, in which the inventor knows the goal is near. Heuristics are rules of thumb for making this transformation. Just as the gold prospector might use some rough rules for deciding where to pan, so the inventor can apply heuristics like 'see if nature has solved a similar problem and if so, imitate nature'. The first inventions were modifications of natural objects (Basalla, 1988); current inventions are more likely to be based on an analogy to nature. Velcro, for example, was based on such an analogy; George de Mestral used a microsope to study the way in which burrs attached to his clothing, and noted the collection of miniature hooks and eyes. It took about ten years to translate this mental model into a product.
The problem with the wilderness metaphor is that it suggests all the inventor's gold is 'out there', waiting to be discovered. In fact, inventors are in the business of creating new kinds of substances and convincing the rest of us that they are precious (Ward, Finke, & Smith, 1995).
However, if not taken too literally, the metaphor is helpful. There are general heuristics that can be used across a wide range of problems, in order to create homing spaces--like looking for an analogy in nature, and following it. There are also domain-specific heuristics that are useful in homing spaces within well-defined domains. Sociologist of science Harry Collins provides several good examples, including "In crystal growing always start the melt cooling from well above the putative melting point", and "The tolerance of TEA-laser electrodes are sufficiently large to make it unlikely that the exact shape of the electrodes is the cause of failure" (Collins, 1990, p. 108) . Hans Krebs learned a set of heuristics like tissue-slicing from his first mentor on laboratory methods, Otto Warburg (Holmes, 1991, p. 295).
These domain-specific heuristics can be transformed by a particular scientist or engineer into individual heuristics. This is particularly likely to happen with inventors and discoverers, whose work frequently takes them beyond the bounds of existing techniques. At one critical point in his researches on ornithine, Hans Krebs had to modify the tissue-slicing heuristic he had learned from Otto Warburg by developing a new heuristic for determining the best medium in which to bathe the slices while they were being tested. Basically, he used a general heuristic--when in doubt, look at nature's solution to a similar problem--and decided to "imitate as closely as possible the actual physiological situation in which tissues normally exist" by duplicating the composition of plasma as closely as possible (Holmes, 1991). Krebs used a weak or general heuristic to develop his own strong heuristic. Alexander Graham Bell used the same weak heuristic, which he called "follow the analogy of nature," to develop a powerful mental model for creating a device that could transmit speech.
Heuristics, like mental models, often become apparent when a kind of resistance is encountered--a resistance that forces the problem-solver to articulate and defend her approach. For example, an expert who solves a certain class of problems automatically, without thinking, may have to struggle to describe her heuristics when queried by a novice. This fact has led Suchman and others to argue that heuristics, like plans, may be post-hoc rationalizations invented by problem-solvers to explain what they do (Suchman, 1987).
I had a calculus professor in college who, in the days before hand-held calculators and desktop computers, would put a long integral on the board and solve it in seconds. He had worked in industry and had developed a powerful set of domain-specific heuristics which he had trouble explaining. I had to get another math professor to work with him and explain his tacit knowledge to me. This example illustrates that heuristics do not always have to be post-hoc rationalizations.
Taken together, heuristics, mental models and mechanical representations allow us to study and compare the cognitive styles of inventors and discoverers, by which I mean the manner in which each individual practitioner finds, transforms and solves problems.
To investigate general heuristics and representations, we need multiple case studies of inventors and discoverers working in different domains. To investigate domain-specific heuristics and the sorts of mental models and mechanical representations that are shared by experts, we need to add multiple case-studies within an area of expertise. Inventors and discoverers, however, are continually stretching beyond recognized domains, reconfiguring the landscape of expertise; to investigate this process of problem transformation, we need to add comparisons of different inventors and discoverers working on what in hindsight came to be regarded as the same problem.
Hence, we will spend a good portion of the rest of the chapter on a fine-grained comparison of two men, each of whom claimed to have been inventor of the telephone. In addition to the three categories above, we will need also to talk about themes, goals and plans (Schank & Abelson, 1977). Themes correspond to the very general goals people adopt, e.g., make tons of money or be creative, and also the roles they adopt to achieve them, e.g., entrepreneur or painter. Unlike generals and entrepreneurs, inventors rarely talk about goals and plans--they simply design, and often one must infer their intentions from their design processes.
This page was last edited: Wednesday, July 14, 1999