Creativity and idea generation

By Gordon Rugg

So what is creativity, and how can you generate more and better ideas?

There’s pretty general agreement that:

  • Creativity is a Good Thing
  • Thinking outside the box is a Good Thing
  • Thinking laterally is a Good Thing

That’s a good start.

However, when you start asking about how creativity works, or just how you’re supposed to think outside the box, or think laterally, an element of vagueness starts to roll in, like a dense bank of fog off the Atlantic at the start of a horror movie…

You start hearing stories of people and organisations that thought successfully and laterally outside the box, in a way that solved their problems with designing better elevators. You encounter puzzles involving people and items being found in improbable situations, such as stabbed to death with no weapon visible, in the middle of a field of unsullied snow. It’s all very edifying and interesting, but it doesn’t get to grips with what creativity really is, or how to do anything systematic about creating new ideas.

This article gives a brief overview of a systematic framework for making sense of creativity, and for choosing appropriate methods for generating new ideas.

The framework in this article is based on a distinction between three levels of knowledge, namely explicit, semi-tacit and tacit.

We use this distinction a lot in our work; it’s invaluable for making sense of what would otherwise appear to be a tangled mess. Each of these three levels can be divided into sub-types, as described in a previous article, but I’ll stick with three levels for simplicity and clarity.

Mapping creativity methods onto knowledge types: Explicit knowledge

Explicit knowledge is what most people think of as ordinary, straightforward knowledge; it’s things like your name, or the number of days in June, or the capital city of France.

Explicit knowledge often involves serial processing, i.e. step by step reasoning with “facts”.

If you now start looking for idea generation techniques that work with explicit knowledge and serial processing, you soon discover methods such as systematic constraint relaxation. This is a method which is serial (as in the “systematic” part of the name) and explicit (as in the “constraint relaxation” part of the name).

In systematic constraint relaxation, you start by producing a written list of constraints that limit you when tackling a problem. For instance, if you’re an architect designing a skyscraper, then the constraints for lifts (“elevators” in the USA) might include:

  • Must hold at least seven people
  • Must hold at most 25 people
  • Must be on the inside of the building
  • Must be fully enclosed

The next step is to throw away each constraint in turn, and see which new ideas emerge.

For instance, if you throw away the constraint of must hold at least seven people, then you might have the idea of one-person lifts. These would have some strong advantages, such as increased speed (because the lift wouldn’t stop at any intermediate floors between starting point and destination) and increased security (because the user wouldn’t need to worry about potential criminals getting into the lift with them). They would also have disadvantages, but that isn’t the point; the key point is that you only need to generate one really good idea to make the session worth doing, so you concentrate on generating the ideas first, and then thinking through the details later.

Here’s a real life example of what happens when you throw away one of those constraints. Just in case you’re wondering, yes, the yellow blob is an elevator, and yes, it’s on the outside of the building, with a very long, very visible drop beneath it. It’s very popular with some users, but not with others…

1024px-Niagara_falls_fallsview_04.07.2012_15-25-48(Picture credits at the end of this article)

You can use different ways of deciding which constraints to throw away – for instance, you might only throw one away at a time, or you might throw away more than one at a time, or you might rank them in terms of current cost of the constraint, or some other criterion.

You can also combine this approach with other methods, such as scenarios to get a richer understanding of problems with the current constraints, or Multi-Criterion Decision-Making to assess the ideas that you generate.

This is a useful method for dealing with explicit knowledge and serial processing. However, that only applies to some problems. More often, you’re dealing with semi-tacit or tacit knowledge, and with parallel processing and pattern matching.

Mapping creativity methods onto knowledge types: Semi-tacit knowledge

Semi-tacit knowledge is knowledge that can be accessed using some techniques, but not all techniques. (Interviews, focus groups and questionnaires are usually between bad and useless for dealing with semi-tacit knowledge; questionnaires in particular can be worse than useless if they produce impressive-looking statistics that are actively misleading.)

In the case of lift design, for instance, a lot of women don’t like using lifts late at night because of the risk of being assaulted while inside, with no way of escaping. However, they might not be keen on saying so in as many words, particularly if they’re being interviewed by a clueless male engineer.

Another classic problem is that people will recognise something as being highly relevant when they see it, but won’t think of it otherwise.

These examples demonstrate a couple of sub-types of semi-tacit knowledge; there are numerous others.

One effective way of taking semi-tacit knowledge into account when generating ideas is to use Idea Writing. We’ve written about it in a previous article, here. In brief, it involves writing a group of people sitting round a table. They start by each writing ideas at the top of sheets of blank paper, one idea per sheet; they pile the sheets in the middle of the table, and then pull out a sheet at random, and write a constructive comment on it, then return it to the pile. The illustration below shows a fictitious example.


This method involves everyone working at the same time (so you don’t get people having to wait for their turn) and working in silence (so the group doesn’t get dominated by aggressive or long-winded individuals). It’s also fairly anonymous, if the group aren’t familiar with each other’s handwriting. This combination of features helps you to access sensitive topics that might otherwise not be mentioned (a common form of semi-tacit knowledge).

The format also means that people can recognise interesting ideas in the previous comments, which will be likely to spark off further ideas in their own mind, tapping into another form of semi-tacit knowledge.

A practical advantage of this method is that it’s self-documenting; you don’t need a scribe to record what people say, and you can simply photocopy or scan the sheets at the end of the session, then circulate them to everyone who wants them.

Mapping creativity methods onto knowledge types: Tacit knowledge

Tacit knowledge is defined in different ways by different disciplines. We use the term in a strict sense, to describe real knowledge or skill that the individual can’t validly introspect into. When you deal with experts, you often see them demonstrating impressive skills, but you often find that they can’t put those skills into words. You also often find that when the experts do try describing their skills in words, their descriptions don’t actually correspond very well with reality.

Tacit knowledge often involves “gut feel” and “intuition” and other concepts that are usually treated as semi-mystical black boxes.

In this article we’ll focus on one aspect of tacit knowledge that can be fairly easily explained without using mystical terms. It’s deep structure pattern recognition.

Deep structure is a concept that crops up in a wide range of places, from linguistics to film theory and bridge design. It’s about the underlying regularities that you see when you strip away the surface detail. In story telling, for example, there’s a common deep structure ending in which the protagonist defeats the villain. The surface features may vary; for instance, this deep structure may take the form of Ripley kills the alien queen, or Arnie shoots the bad guy, but it’s the same deep structure underneath the different surface forms.

Pattern recognition is what it sounds like; you recognise a pattern. This is something that humans do extremely well. It’s only when you try writing pattern recognition software for computers that you realise just how ubiquitous pattern matching is in everyday human life, and just how different it is from step by step sequential thinking.

There are various forms of pattern matching; for instance, one form is recognising the face of someone that you know.

In this article I’ll focus on pattern matching in which you see a deep structure that matches what you’re looking for.

The classic example is an engineer looking at a reconstructed dinosaur, such as the one below, and seeing that the deep structure of the dinosaur’s body structure is the same as the deep structure of a cantilever bridge, such as the one below. This is a genuine mechanical deep structure match, not just a coincidental resemblance, and it’s given a lot of useful insights to palaeontologists about how dinosaur anatomy worked.

AmpelosaurusDB(Picture credits at the end of this article)

Pierre_Pflimlin_UC_AdjAndCrop(Picture credits at the end of this article)

You can apply this principle to creativity by simply showing people arbitrary images, and letting the people look for deep structure pattern matches within those images, for instance via a PowerPoint slide show. Most of the images won’t spark any recognition, but some probably will, either immediately, or via stirring subconscious associations that will bubble to the surface much later.

There’s a temptation when using this approach to choose images that you think ought to be relevant, but that temptation should be avoided. The key point of this approach is that you’re using it because the obvious approaches (such as looking for solutions in places that ought to be relevant) haven’t worked, so you need to look somewhere else.

Because the pattern recognition process is non-verbal, the result is often an “aha” experience that occurs instantly, with no conscious reasoning involved, making the process hard to put into words, which is why it is often viewed in semi-mystical terms.

A key advantage of this method is that it deliberately bypasses the need for words, and lets the participants focus on the pattern recognition that is at the heart of the method.

Closing thoughts

This article described a swift overview of a way of approaching creativity and idea generation in a way that is systematic, and that is based on the underlying mechanisms involved in creativity and idea generation.

We’ll return to these themes in later articles; we’ll also return to deep structure and surface structure.

Attributions, notes and links

“AmpelosaurusDB” by ДиБгд at Russian Wikipedia – Transferred from ru.wikipedia to Commons.. Licensed under Public Domain via Commons –

“Pierre Pflimlin UC AdjAndCrop” by Image Adjustment by User:Leonard G. – See below. Licensed under Public Domain via Wikimedia Commons –

“Niagara falls fallsview 04.07.2012 15-25-48” by Dirk Ingo Franke – Own work. Licensed under CC BY-SA 3.0 via Wikimedia Commons –

Notes and links:

You’re welcome to use Hyde & Rugg copyleft images for any non-commercial purpose, including lectures, provided that you state that they’re copyleft Hyde & Rugg.

Overviews of the articles on this blog:


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s