DesignWIKI

Fil Salustri's Design Site

COLLECTIONS

2015.01.09 11:13
research:analogical_reasoning

Analogical Reasoning

Drawing analogies is a fundamental design method, and (I think) an “explanation” for many instances of creativity.

• ATTENTION TODO This page desperately needs freshening up.
• TODO Do something with the cmaps of analogies, and include the dyson vacuum case study in that.

What is Analogy?

Google has a number of very interesting definitions.

Preliminary working definition: Finding abstract similarities between multiple concretely different things.

Assorted Notes

Mistaken process

From [Cas07]: “The relevance of metaphors to problem-solving is pertinent to three fundamental steps (Gentner, Bowdle, Wolff, & Boronat, 2001). The first step consists of extracting a variety of unfamiliar concepts from remote domains, where possible relationships with the problem at hand are not always evident. The second step involves establishing a mapping of deep or high- level relationships between the metaphorical concept and the problem. Correspondences are identified by means of abstractions and generalizations. Relationships of secondary importance are discarded, and only structural correspondences between the metaphorical source and the problem are set up. The last step deals with transferring and applying structural correspondences associated with the metaphorical source to the problem at hand, which at the end generally leads to a novel solution.”

• Step 1 cannot really occur successfully without extracting the right unfamiliar concepts. I think it's far better to model the problem domain, then abstract it, then look for other “unfamiliar concepts” that share the abstract structure. In other words, invert steps 1 and 2.
• Also, the dependence on structural correspondences is wrong. Real analogy is based on function, not structure. Even in the case of Velcro, the structure of fur & burr was transferred only because it provided the necessary function.

Accidental analogies

A design that is, only in hindsight, noted to bear an analogical similarity to some other design. If analogy-making involves the development of an abstract model common to two or more situations that is the basis for implementation in new situations, then if an abstract model is already (possibly unconsciously) known, it can lead to an analogy but without consciously doing the otherwise requisite reasoning.

Connection to creativity

[CS14] describes how analogy works in brainstorming in engineering design. Digest that paper carefully.

A problem with Gentner's work

Gentner's approach assumes both sides are represented in the same way. But this isn't so; one person represents different things in different ways. This seems to be a shortcoming of Gentner's work; it assumes there is a canonical/correct model of an analogy that is shared by all agents. So - do we need to research analogical reasoning by multiple agents?

Metaphors

Metaphors are more abstract/broader than analogies. The “easy button” is a metaphor?

From J. Vincent to biomimetics mailing list 4 June 2013

W.r.t. the mindset of de Mestral when he thought of the concept of Velcro, he had a very well-defined reason for thinking of it. It appears that a few nights previously he had been having trouble with his wife's zip…. It's the mental jump that makes the analogy which I find interesting.

Oh yes… and Kalman has obviously never heard of the one-legged milking stool. Very effective, doesn't have a problem with uneven floors. Slight problem - you have to strap it on to your bum. But maybe even that's an advantage in cold weather.

Non-biomimetic cases by J. Vincent

(My thought: But are they still analogies at all? Or post-hoc rationalizations? Or just PR?)

Roof of the Crystal Palace: The corrugated roof was invented in 1810 or earlier by John Claudius Loudon, an inventive horticulturalist, some 40 years before the Crystal Palace was designed and (as far as I can tell) before people in the UK had come across the floating leaves of the lily, Amazonica.. The corrugated roof bears no relation to the leaves of lily, but the half-round arch which tops the Crystal Palace (not present in the original drawings) is very reminiscent of the leaf in its design. There may be a connection there. A result of lax reportage by the Press?

Eiffel Tower: This was the first structure to be designed according to wind loadings. Its hierarchical strutted structure is probably a result of limited access to the site. The Tower is nothing to do with the structure of bones, tulip stems, or anything else biological.

Sydney Opera House: Nothing whatsoever to do with shells. It's a shell structure, but that's a technical description. Nothing in the original accounts of its design or structure says anything about a biomimetic origin.

Polar Bear light guides: The bear's hair does not function as a light guide (shown experimentally) although light guides arranged in the same way can have useful properties.

Eastgate Centre, Harare: Doesn't work like a termite mound (technically as a stack - chimney - which can draw air through the system) because termite mounds don't work like that! The building was designed before people understood how the nest's gas exchange system really works (it seems to be more like our lungs, semi-tidal and not mixing very much). And people ignore that insects can cope with a wide range of CO2 in the air surrounding them.

Research Ideas

My hypothesis

The fundamental mode of creativity in design is reasoning by analogy (consciously or otherwise).

My approach

Essentially science-based - an inductive argument: If (1) all creative acts can be explained provisionally by analogy and (2) no better explanation presents itself, then the hypothesis (above) holds.

NOTE: Since it is impossible to be sure if a given case was due to analogy, I just consider the notion of provisional explanation, not “knowing.”

Implications

If the hypothesis holds, then there could be many implications. For instance, one could focus on analogical reasoning to “teach creativity”, and it could impact on design tools, AI applications, etc.

Analogies as Thesaurus Entries

If analogy is really about finding the abstract similarities in the concretely different, then an analogy is like an entry in a thesaurus. The similarities noted in thesauri are of varying kinds and degrees. They are conceptual similarities expressed in words (symbols).

If the symbols represent designs (or design concepts, or design ideas, partial or complete), then one may see a [Design thesaurus] as an encyclopedia of design analogies.

To Check

• Swarm behaviour and its use in various technologies is another case.
• There is a strong connection to bio-inspired design (BID).
• There is a connection to pattern recognition.
• There is a question around the values we place on the characteristics of analogies. Different characteristics can lead to different analogies. How do we prioritize those characteristics (give them value)? How does this connect to subjective vs objective values?
• Mak & Shu, 2004, analogical reasoning papers
• Gentner, 1989, taxonomy of analogies
• Case: James Crocker / hubble repair: Based his fix for the hubble telescope while taking a shower, based on the shower nozzle. Look it up; add it to the library.
• Google scholar and plain Google searches for “analogical reasoning” cognitive style.