You go out to your favorite noodle shop. They’ve got four types of ramen, but you always order the shio. Should you stick with it or try a different one?
The ramen-lover’s quandary illustrates, on a personal (and whimsical) level, a trade-off that occurs across systems from biological to organizational and social: exploiting current tendencies, understandings, or viabilities comes at the expense of exploring novel variations, approaches, or possibilities.
In 1975, John Holland called this situation the “mutual interference of search and exploitation.”
In “Dual thinking for scientists,” a paper in the Ecology and Society special-feature-in-progress on “Reconciling Art and Science for Sustainability,” Marten Scheffer and ten coauthors discuss the exploration/exploitation-like duality in human cognition and consider its implications for scientific progress.
To readers of Daniel Kahneman, these two cognitive modes are familiar as system-I and system-II. The former diverges, associates, and intuits; the latter reasons and rationalizes.
“Here we will defend the perhaps provocative view,” Scheffer and coauthors write, “that the way science and its institutions are organized reflects [an] overestimation of system-II thinking in producing scientific progress.”
Provocative indeed. How then to engage this science-as-usual regime? — so as to engender a shift toward a science that more effectively stimulates and incorporates system-I thinking?
“Perhaps we should use some of the education techniques from arts to boost adventurous exploration and ‘learning at the edge of chaos,’” they conclude.
I couldn’t agree more.
(Written with an eye toward next week’s Transformations 2105 conference.)
- Root-Bernstein, R. and M. Root-Bernstein. 1999. Sparks of genius: the thirteen thinking tools of the world’s most creative people. Mariner Books, New York, NY, USA.
- Holland, J. H. 1975. Adaptation in natural and artificial systems: an introduction with application to biology, control and artificial intelligence. University of Michigan Press, Ann Arbor, MI, USA.
- Kahneman, D. 2011. Thinking, fast and slow. Farrar, Straus and Giroux, New York, NY, USA.
- March, J. G. 1991. Exploration and exploitation in organizational learning. Organization Science 2(1):71-87.
- Page, S. E. 2015 (accessed). Understanding complexity. The Teaching Company, Chantilly, VA, USA.
- Scheffer, M., J. Bascompte, T. K. Bjordam, S. R. Carpenter, L. B. Clarke, C. Folke, P. Marquet, N. Mazzeo, M. Meerhoff, O. Sala, and F. R. Westley. 2015. Dual thinking for scientists. Ecology and Society 20(2):3.
This question of dualism has absolutely nothing to do with the reality of choice.
Is it a taco or not?
Hey John, the lede about the ramen is indeed whimsical. Metaphorical, if you will. In fact, the reference link to Page is a h/t, as he describes a similar metaphor.
Points of note are that (1) this exploration-exploitation pattern plays out in multiple system domains; and (2) on a personal level, one’s participation in, affiliation with, identification with, “exploitation of” a particular system (food system, energy system, etc.) is subject to reinforcement.
Hey Howard, in the spirit of intuitive system-1 connections, I thought I’d share this talk by Kenneth Stanley entitled “Why Greatness Can’t be Planned”. https://www.youtube.com/watch?v=dXQPL9GooyI
I call it an intuitive connection because I didn’t read the source article you link to, so it is possible that the connection is pretty tangential :-p
The talk is interesting because it describes how Stanley, an AI compsci researcher, found that explicitly rewarding innovation (rather than just rewarding progress toward an objective goal — the dominant approach) led to far better outcomes when creating learning agents.
My takeaway was that this provided some compelling evidence that putting disciplined, goal oriented thinking/action/planning and spontaneous, opportunistic exploration (as you perhaps might call artistic) into dynamic harmony is likely a better approach than overemphasizing either system-I or system-II thinking.
This reinforces the necessity for an explicitly integrative approach between system-I and system-II cognitive exploration in our scientific inquiries and supported by our institutions.
Thanks for the pointer to Kenneth Stanley’s talk. Fascinating and highly relevant, I would say.
Thanks also for reviving a comment thread on a 4-year-old post. I like this kind of asynchronicity, and I’ll take this opportunity to revisit and explore.
My principal analogy in this post was between the explore~exploit analysis of behavior and the dual-process analysis of cognition. I think that clarifying analogies and distinctions may raise questions and/or afford insights. (It has for me.)
In the 1930s, geneticist Sewall Wright described the fitness landscape, a visual analogy for conceptualizing explore~exploit in terms of valleys and peaks.
“[T]here must be some trial and error mechanism on a grand scale by which the species may explore the region surrounding the small portion of the field which it occupies. To evolve the species must not be under strict control of natural selection. Is there such a trial and error mechanism?” (p.359)
In 1975, John Holland described his development of genetic algorithms, emphasizing explore~exploit trade-offs as among the “obstacles to adaptation.”
A classic paper applying these concepts to human systems was James March’s 1991 “Exploration and Exploitation in Organizational Learning.” March emphasized both integrative strategies and their trade-offs: “Both exploration and exploitation are essential for organizations, but they compete for scarce resources.”
March characterized the explore~exploit duality as follows:
“Exploration includes things captured by terms such as search, variation, risk taking, experimentation, play, flexibility, discovery, innovation. Exploitation includes such things as refinement, choice, production, efficiency, selection, implementation, execution.” (p.71)
Having watched Kenneth Stanley’s video but not (yet) read his book, I’d place his work with Picbreeder in the Wright-Holland-March lineage.
Let’s switch now to the other line of research. Daniel Kahneman and many others have analyzed cognition in terms of dual processes: system-I thinking is fast, intuitive, associative, while system-II thinking is slow, deliberate, evaluative.
Kahneman’s (2011) narrative was largely about cognitive bias research that can be explained in terms of dual-process theory. A supporting narrative of his was about a rationale for the evolutionary development of dual cognitive processes. The rationale: fast system-I processing reduces cognitive load and thus serves the energetic efficiency of the organism. “The evidence is persuasive: activities that impose high demands on System 2 require self control, and the exertion of self-control is depleting and unpleasant.” (p.42)
What caught my attention about the article by Marten Scheffer and colleagues was the shift in narrative. Instead of the biases engendered by associative, system-I thinking, they emphasized opportunities for creativity. They flipped the script on Kahneman to a dual-process story that sounds more like the explore~exploit one.
For what it’s worth, the explicit analogy between dual-process and explore~exploit is mine, not Scheffer et al.’s. They neither mention “exploit” nor cite Holland.
That said, I’m not the first to make this association. Consider, for example, Kurt Matzler et al.’s 2014 paper: “The Role of Intuition and Deliberation for Exploration and Exploitation Success.”
Wrapping up this comment, which is already longer than the original post, here are a few analogies/distinctions I’ve noted.
1. Marten Scheffer and colleagues advocated for system-I thinking. Kenneth Stanley advocated for objective-free exploration. Analogous in intent, I’d say. And in practice?
2. For the organism, system-I processing is described as evolutionarily efficient (per Kahneman). Conversely — or seemingly so — for the organization, exploitation strategies are described as strategically efficient (per March). Hmm.
3. On trade-offs, I agree with you Travis and would expand on your point. For thinking about explore~exploit in social systems, one of my preferred visual analogies is the landscape of basins (not peaks). This version — https://www.solvingforpattern.org/wp-content/uploads/2019/09/regime-shifting.png — adapted from Frances Westley and colleagues (2011), locates four strategies on the landscape: weaken the undesirable, dominant regime; question and explore alternative possibilities; strengthen (i.e., exploit) a preferred, niche regime; and enable others, individually and collectively, in shifting affiliations to the preferred regime.
Matzler, K., B. Uzelac, and F. Bauer. 2014. The role of intuition and deliberation for exploration and exploitation success. Creativity and Innovation Management 23(3):252-263. https://onlinelibrary.wiley.com/doi/abs/10.1111/caim.12065
Stanley, K. 2015. Why greatness cannot be planned: the myth of the objective. TTI/Vanguard. https://www.youtube.com/watch?v=dXQPL9GooyI
Westley, F., P. Olsson, C. Folke, T. Homer-Dixon, H. Vredenburg, D. Loorbach, J. Thompson, M. Nilsson, E. Lambin, J. Sendzimir, B. Banerjee, V. Galaz, and S. van der Leeuw. 2011. Tipping toward sustainability: emerging pathways of transformation. Ambio 40(7):762-780. http://dx.doi.org/10.1007/s13280-011-0186-9
Wright, S. 1932. The roles of mutation, inbreeding, crossbreeding, and selection in evolution. Proceedings of the sixth international congress of genetics 1(8):356-366.
I happened upon this blog, through combing the resources section of SFI’s Complexity Explorer site. (That’s the Santa Fe Institute, btw.)
Thanks for making this point. It’s a small comfort: at least there are people who still contemplate the nature of the large divide between the humanities and the natural sciences. But one can get stuck in articulating and re-articulating this difference between the creatives and so-called bean counters, or the realists and the so-called dreamers. Isn’t black-or-white, dualistic thinking, and this classification example itself, also problematic?
Taxonomists will have no idea what I’m getting at… 😉
Thanks for happening by. Good points.
I like how Daniel Kahneman calls his System 1 and 2 distinction a “fiction.”
I also like to think of the distinctions we draw in terms of complementaries: the 2-within-1 and the 1-within-2.