Does Poincare´ still rule creativity studies? Comments on two recent (2010) books on creative processes…”Where good ideas come from”, by Steven Johnson (Allan Lane) and “Sudden Genius?” by Andrew Robinson (Oxford).

For the experimentally oriented psychologist it is striking that both these books rely heavily on biographical and personal accounts as their evidence base for understanding creativity. Experiments rarely get mentioned.

Poincare´ s (1908) theorising, which drew on his own delayed recalls of solving mathematics problems decades previously, is given considerable attention in both books.

Andrew Robinson, in “Sudden Genius?”, while stating that unconscious processing is surely involved in creative productions, as Poincare´ argued, does criticise Poincare´’s particular theory, with its stress on incubation, as not well established and casts doubt on the replicability of incubation. (Incubation is the beneficial effect of consciously putting the problem aside for a while; either you do better on return or have a spontaneous and sudden insight, when not actively thinking about the task - anecdotally, often in bed, on a bus or beach). However, this is an area in which Robinson’s coverage of experimental work really is too brief and ignores the many studies meta-analysed by Sio & Ormerod (2009, Psychological Bulletin) that support incubation, especially in divergent thinking tasks (where many possible answers are to be produced).

In Robinson’s final chapter, cognitive theories, such as those of Simonton are deemed to be of little explanatory value. Yet Simonton’s theory, based on random variation and selective retention of ideas, is well specified and fits a lot of objective data based on historical records of scientific and artistic production. It neatly handles the relationships between quantity and quality of production and the different peak ages of productivity and creativity in different fields. So Simonton’s theory may well be pointing in the right direction even if the underlying mechanisms are not given in detail (a bit like Darwin’s version of evolution on which it is modelled).

Robinson’s discussion of the links between measured intelligence and creativity gets into a questionable interpretation of the Flynn effect in an attempt to boost the famous physicist Richard Feynman’s measured IQ (in c. 1930) from 125 to something higher. It would be neater for the hypothesis that genius always requires very high IQ if Feynman had done a bit better on the day of his childhood test. However, the Flynn effect, viz., that raw scores on IQ type tests have increased steadily over decades would actually downgrade Feynman by current norms rather than upgrade him; however Robinson tries to argue the opposite(to save notion that genius requires very high IQ). This seems to be a clear error. Another error of which I am pretty sure ist Robinson’s ascription of Jackson Pollock’s death to suicide. Pollock’s death was due to a drunk driving accident, as far as I know, rather than deliberate suicide.

Is it really the case that 100 years after Poincare´, we are no further forward in understanding the processes underlying creativity? The only solid generalisation that Robinson identifies is the 10 year rule, well known from studies of expertise. The main thrust of “Sudden Genius?” is that major creative work is not “sudden” but is spread over a long period, often around 10 years and that sudden eureka moments play little role in real creativity.

Steven Johnson in his “Where good ideas come from” also stresses a notion related to the 10 year rule, i.e, the idea of slow hunches, in which creative ideas take years to be fully developed and tested. So, for example, it seems that Darwin’s idea of evolution by natural selection took a year or two to be developed in its basic form and a further period of 20 years or so, to be further elaborated and tested to Darwin’s satisfaction.

Johnson focuses very much on the cultural context in which the innovator operates and argues that the availability of many sources of information is vital to allowing the combination of ideas from different domains. He argues that new ideas are essentially built out of parts of existing ideas to make new combinations. To use Darwin again, the idea of evolution combines ideas of competition for resources among living things and selective breeding (minus the deliberate breeder). As for a cognitive mechanism, Johnson recruits Poincare´ once more with the proposal that unconscious mental work during incubation involves trial and error combining and recombining of ideas. In terms of the social context Johnson argues that cities, coffee houses, open plan offices, lab meetings, conferences, a wide circle of acquaintances with different backgrounds and now the World Wide Web facilitate idea combining and merging across many fields.

Steven Johnson does attempt to define “idea” which, all too often, is left unanalysed in studies of creativity. He writes (p45) that –

“A good idea is a network. A specific constellation of neurons- thousands of them-fire in sync with each other for the first time in your brain, and an idea pops into your consciousness. A new idea is a network of cells exploring the adjacent possible of connections they can make in your mind.”

However, this definition seems to be muddling the neural implementation (cells firing in sync) with the cognitive construct (new idea) and with phenomenology/experience (“pops into in your consciousness”). The new jargon term in the quotation, borrowed from theoretical biologist, Stuart Kauffman, the “adjacent possible” is the set of next states that can be reached in one change to the current state of knowledge. In problem space terms, it is where you can go in one step from the current state.

Johnson’s attempt at a definition does bring it home that a good definition in cognitive/information processing terms of “idea” would indeed be a good idea!
Surprisingly we do seem to be lacking a good definition. A future blog will focus on this question of defining “idea” and “new idea”.

Ken Gilhooly Feb 21, 2011.
--------------------------------------------------------------------------------------------------------

Incubation, Internalism and Distributed Cognition

Incubation, Internalism and Distributed Cognition: an e-Discussion

Following the Kingston 2010 Meeting on Distributed Cognition and Problem Solving, there was brief email discussion about DCog and Incubation effects. (Incubation is when a problem is solved readily after time away from the problem.) An exchange between Fred Vallee-Tourangeau and me is shown below.



Fred Vallee-Tourangeau:

I liked the incubation gauntlet( i.e., challenge issued at DCog meeting for a DCog account of Incubation), but if right from the start the theoretical perspective doesn’t allow for representations to be distributed over internal and external resources, then a distributed cognition perspective will not be able to rise to the challenge. For my money, incubation ‘works’ for stochastic/aleatory reasons – leaving a problem aside, the representation decays. Confronting the problem again, the representation is put together again, constructing and drawing upon a slightly different interpretation of the problem (the semantic activation spreading slightly differently), recruiting slightly different operators maybe, and out of that chancy configuration may surface a more productive problem representation with possibly more traction to get you out of the impasse. But this assembly process, from a distributed cognition perspective, could also recruit external resources, and in fact, you may double the possibilities by re-assembling and re-coupling internal and external parts to form a problem representation (or perhaps you double the noise!)

Perhaps you have heard me talk about the Watson episode in the Double Helix (or you read it yourself) about how he put the base pairs together and discovered how the helix was held together. Here’s the passage:

“When I got to our office….I quickly cleared away the papers from the desk top so that I would have a large, flat surface on which to form pairs of bases…Though I initially went back to my like-like prejudices, I saw all too well that they led nowhere….I began shifting the bases in and out of various other pairing possibilities. Suddenly I became aware that a [A-T] pair held together by two hydrogen bonds was identical in shape to a pair [G-C] pair held together by at least two hydrogen bonds. All the hydrogen bonds seems to form naturally; no fudging was required to make the two types of base pairs identical in shape. Upon his arrival Francis did not get more than halfway through the door before I let loose that the answer to everything was in our hands”. (Watson, 1968, pp.123-125).

The final understanding gels in Watson’s mind, but the external scaffolding necessary for him to achieve insight was both strategic and I suspect possibly opportunistic (is it possible to rule that random movements might have led to felicitous configurations?).

You probably know Peter Cheng, and perhaps the work he did with Herb Simon in the early/mid 90’s on the importance of multiple representations (diagrammatic/mathematical) in scientific discovery. I saw him in DC at CogSci 08. We talked a bit about some of my work with Wason’s rule discovery task using multiple representations (people are more likely to discover the rule and formulate more pertinent hypotheses when they have access to both a graphical and algebraic representation. In light of his writings, I thought he’d be particularly fond of the DCog perspective. But he is staunch internalist. He does not deny that information format and annotations can help, but he argues that all the interesting psychology takes place in the head, not as a result of the interaction with reified projections and artefacts, cognitive or otherwise.

Ken Gilhooly:

Incubation effects as results of “beneficial” forgetting that brings about changes in problem representations is a theory of some standing and indeed was put forward by Simon among others. This does seem to be a thoroughly internalist explanation.

A role for environmental cues has been incorporated into some approaches, but the cues have to be processed internally and lead to changes in the internal representation.

The Watson case of insight, while manipulating a concrete model, is interesting as an example of insight following overt trial and error accomplishment of a solution, but not really a case of an incubation effect?

From the internalist point of view, concrete material/models are helpful because they facilitate exploration that would otherwise overload working memory.

This seems to be pretty intra-organismic.

I am inclined to a rather fence-sitting view that some important cognitive processes (e.g., incubation) are essentially in-the-head and others are distributed across heads and across the environment (e.g., group problem solving with external memory aids)?

Attended Symposium at Kingston University, London, on Problem Solving and Distributed Cognition, July 15-16, 2010.

Although colleagues at UH (Stephen Cowley, Sue Anthony, Nuala Ryder and Evie Fioratou) have tried to explain this approach to me off and on over the last 6 years or so, I had never quite got it.

Would attending this meeting rouse me from my dogmatic slumbers? I had been strongly influenced in my formative years by the basically internalist view of the pioneers of cognitive science (AI and Cognitive Psychology) such as Newell & Simon , and Miller, Galanter & Pribram. In those heady days of the early 60s, they were the revolutionaries taking on the dinosaurs of Behaviourism by re-establishing mental processes as valid topics of study. Were they now to hand over the study of cognition to new insurgents who adopted a somewhat bewildering range of slogans for externalism v internbalism, distributed v localised, embedded (situated) v de-contextualised and embodied v. disembodied abstract symbolism?

A wide range of topics were addressed at the meeting including, how people coped with infuriatingly poor voice recognition based information systems, tool use, hints from other solvers' eye movements, aggregating votes, solving puzzles, applications of power laws to insight shifts, philosophical analyses among other things. Few delegates could probably follow all of these papers and I struggled to find common threads and to understand the more philosophical presentations.

Overall, I felt the most valuable results of the Situated etc Movement was to draw attention to aspects of cognition in the wild that had been set aside by the lab based artificial puzzle oriented classic approach. However, I was still left feeling that the old paradigm could cope with these points and data favouring embodied cognition with a bit of stretching. I didn't sense much support among the delegates for the wilder shores of the "extended mind" hypothesis...this would say that the total informational resources of the internet become literally part of my LTM when I have Google access, or that a notepad becomes part of my working memory if I use it during mental arithmetic.

I don't think my own research into , say, incubation effects, will be affected by the Situated Movement. I did ask, but no one came up with an Externalist/Distributed/Situated approach to incubation.

If this works my draft Kingston talk can be seen at

http://docs.google.com/present/edit?id=0Aao3SCUWGonAZGZic3ZwMzVfMGR6N2c5MnA5&hl=en

Talk for Distributed Cognition & Problem Solving meeting, July 15-16, 2010, Kingston U.

Abstract for Kingston meeting on Distributed Cognition and Problem Solving, July 15-16,2010.

Creative problem solving, incubation and the internalist/externalist debate.

Ken Gilhooly, School of Psychology, University of Hertfordshire, Hatfield AL10 9AB, England, UK.
k.j.gilhooly@herts.ac.uk


Traditional cognitive psychology approaches to problem solving aim to explain how people solve problems in terms of transformations to internal representations of problems. Typically problem states are seen as represented by symbol structures in a working memory system which are subject to goal driven transformations by rules. These transformations yield a search through a problem space from starting state to goal state. This approach, which one might call “good old fashioned cognitive psychology” (GOFCOP?), derives very strongly from “good old fashioned artificial intelligence”, sometimes known as GOFAI. Although Herbert Simon, a founding figure of both GOFAI and GOFCOP, stressed the intrinsic roles of both the internal organisation of problem solving systems and the structure of the environment in which the system found itself, and how these components interact (eg through the ant on the beach case), the traditional focus has been on the internal side of the story. This leads to a caricature of the traditional approach as completely internalist. I suppose it is probably true, in my case, anyway, that I think the interesting goings on are internal, but it cannot be ignored that the internal processes must be very affected by the external situation.

Representational enrichment: the way tasks are presented can strongly affect the nature of the representations that are formed internally. For example, in a study of divergent thinking we found that more novel uses for familiar objects were produced when participants were presented with the actual object as against a photograph or the verbal label of the object. These 3 conditions can be expected to produce more detailed representations of the object for internal manipulation which would highlight different cues and different uses. The actual object condition is likely to be especially useful when the object itself can be manipulated to yield unusual perspectives. Actual manipulation opens up more prospects for serendipity as Fioratou has noted in studies of real v. paper versions of the cheap necklace problem. Ours account draws on a previous “internalist” analysis based on think aloud records of people generating uses from the verbal labels only. It may be noted that imagery representations are often generated as a basis for divergent production of uses.

Working memory support: In a study of age effects in planning and solving in Tower tasks in older and younger participants, we found marked age differences in purely mental planning measures (such as think ahead depth, planning errors) but no age differences in actual solving performance. It seems that the older participants reduced WM capacity severely curtailed mental planning (lookahead depth) but actually moving the discs gave an external support for WM which thus became less important.

Finally, a challenge for non-internalism: can incubation effects (benefits from setting problem aside and returning after a while ) be other than based on internal processes (beneficial forgetting, attentional shift or unconscious work)?

Hello, getting started

Hello

Planning to post thoughts, observations, ideas here to do with psychology of thinking, especially problem solving and creativity.

More to follow...

Ken