Informatics Creativity: A Role For Abductive Reasoning?

Despite the fact that creative problem solving is desirable at all levels, it is constantly in short supply. Technology developers always wish they, and their peers, could devise novel solutions to the problems at hand. Yet too often, common brainstorming efforts generate only fog or drizzle. Part of the problem may be that while some technology trainers advocate the search for creative solutions, the learning side of the equation is little changed by a “be creative” lecture component. This inquiry attempts to equip computer educators, and practitioners, with a broadened approach to fostering creativity, targeted at expanding discovery within Informatics areas. Particularly, it is suggested that promoting abductive reasoning might help computer professionals understand the benefits of wider investigations within an expanded range of topics. Although discussed additionally later, a short overview of the abductive reasoning approach is that “abduction, or inference to the best explanation, is a form of inference that goes from data describing something to a hypothesis that best explains or accounts for the data.”⁸

Few professionals would suggest there is no need for new, creative solutions to the problems they encounter. An example of the recognition of such needs is highlighted in industry. Consider that each year Procter and Gamble spends nearly $2 billion looking for innovative ideas. This R&D budget spans 150 science areas including biotechnology, imaging, and robotics. Aside from pure research, P&G pursues “aggressive mining of the scientific literature,” as well as employs upwards of 70 “technology entrepreneurs” around the world who analyze local markets to see what has been created elsewhere that might be acquired, modified, or simply spark new ideas internally.⁷

Sparking new computing ideas is the intent of this discourse. Various overlapping and synergistic issues come to mind when considering what has been called “discovery informatics.”¹ Concepts suggested therein are extended here to consider the possible value of abductive reasoning in Informatics. Specifically, how might abduction help generate creative approaches to problems as Informatics reaches into the fabric of so many disciplines. Understanding this technique might assist Informatics workers to progress from being data rich and discovery poor, to a state of information wealth, as professionals benefit from applying new ideas.

In 1990, abductive-related efforts in computing had already been occurring for nearly 20 years.¹⁰ And yet, after another 15 years has passed, the awareness of the potential of abductive reasoning in computing fields remains relatively spotty. The current growth of Informatics programs may offer a chance to change that profile. Since Informatics programs tend to be more cross-disciplinary, it might achieve the goal of that 1990 Automated Abduction symposium—to help a wider range of researchers to recognize that “they might benefit from work on abduction by people in other areas.” Overall, it may turn out that this diffusion cycle will be reminiscent of how slowly object-oriented programming came to be understood, let alone embraced.

We set out to consider the potential of abductive reasoning to promote creativity within the interdisciplinary field of Informatics. As will be presented in the following sections, we believe in the applicability of this reasoning style as a tool for expanding the power of Informatics as it seeks to solve complex problems in a wide range of disciplines. Abductive reasoning is well suited for facing problems that are vague, or even unrecognized. The goal? Turning massive amounts of data lead into tiny, although much more valuable, quantities of information gold.

Back to Top

Creative Thought Overview

Since the beginnings of recorded history, new ideas have propelled peoples, although not necessarily the individual who had the original creative idea, to success. Many early civilizations constructed grand buildings, but it was the Romans who are credited with inventing the arch. Leonardo da Vinci imagined things that could not be constructed in his time, such as the helicopter and the airplane. The Chinese are remembered for creating gunpowder. Genghis Kahn might be the developer of what was later called Blitzkrieg during World War II. Isaac Newton named the force we call gravity. Thomas Edison worked tirelessly to develop the light bulb. And Charles Sanders Peirce is recognized for identifying a line of creative thinking called abductive reasoning. Unfortunately, like da Vinci, the values of Peirce’s contributions are taking many years to be widely applied.

When discussing computing in 1945, Vannevar Bush³ noted that “the human mind does not work” the same way as computers when processing data. He said the mind “operates by association. With one item in its grasp, it snaps instantly to the next that is suggested by the association of thoughts, in accordance with some intricate Web of trails carried by the cells of the brain.” Consider this view as the next subsection examines using abductive reasoning, or reasoned guessing, when looking for novel solutions.

Back to Top

Peirce’s Contributions

Peirce? Abductive reasoning? Yes, both are familiar today, albeit within relatively small circles. Peirce (1839–1914) was an American philosopher who is known for his contributions in numerous areas, such as logic and semiotics (the study of signs). Abductive reasoning perhaps explains how creative ideas often pop to mind. Unlike more traditional deductive and inductive approaches, abduction jumps from a Result to a possible Rule that would, if true, make a Case reasonable. Each of these three logic paths is shown in Table 1.

Another way to consider the abductive approach is that “abduction is a kind of theory-forming or interpretive inference.” A summary view of the process is detailed as steps in Table 2.⁸

As awkward as abductive reasoning might seem initially, in the following section highlights abduction’s potential in a number of areas. Beyond that, abduction is at work in general thinking since “Peirce’s definition of abduction allows us to get a handle on such processes as ‘following a hunch,’ ‘making an educated guess,’ or ‘trusting one’s intuitions.'”¹¹ While these approaches are not the way for university students to learn initial skills, they are highly valued, even if underlying, skills in problem solving, debugging, and creativity in general. Sometimes called heuristics, or rules of thumb, these non-specific skills appear in job advertisements under the term “experience” and are assumed to be acquired during some years of employment.

Back to Top

Current Uses of Abduction

Within computing fields, abductive reasoning seems to have been primarily applied as a database tool with AI undertones, rather than having been promoted for fostering creativity on a broader human level. Outside of computing, one source⁸ notes that abduction is already at work in “medical diagnostics, scientific theory formulation, accident investigation, language understanding and jury deliberation.” As an example of the worldwide nature of interest in abduction in computing, consider these paper titles and nationalities from a 2005 Workshop on the Integration of Abduction and Induction in Artificial Intelligence: Experiences and Directions for Abduction and Induction using Constraint Handling Rules (Denmark); What Bioinformatics Applications Demand of Abduction and Induction (UK); and All’s Well that Ends Well—Planning with Global Abduction (Japan).

Within computing, “Abduction has taken on fundamental importance in Artificial Intelligence and related disciplines. Abductive reasoning is used to generate explanations for observed symptoms and manifestations.”⁶ Furthermore, the viewpoint that “abduction appears to be a powerful concept underlying common sense reasoning” supports the idea of informing students, on a broader basis, of its existence and value. They cite others who suggest that “there is a wide consensus that humans typically use abduction in the diagnosis process.” Unfortunately, in analyzing the complexity of logic-based abduction, they conclude that “there is no hope for complete and efficient algorithms that solve these problems.” Rather, the assertion was made that “a more profitable idea may be to give up on solution completeness and accept approximate solutions, that is, explanations that do not explain all manifestations, but only relevant parts of the present manifestations.” It also was stated that “abduction is an important form of nonmonotonic reasoning allowing one to find explanations for certain symptoms or manifestations.” The richness of this last sentence is revealed via a Web term review (en.wikipedia.org/wiki/Non-monotonic_logic; italics added):

A non-monotonic logic is a formal logic whose consequence relation is not monotonic. Most formal logics have a monotonic consequence relation, meaning that adding a formula to a theory never produces a reduction of its set of consequences. Intuitively, monotonicity indicates that learning a new piece of knowledge cannot reduce the set of what is known. A monotonic logic cannot handle various reasoning tasks such as reasoning by default (facts may be known only because of lack of evidence of the contrary), abductive reasoning (facts are only deduced as most likely explanations), reasoning about knowledge (the ignorance of a fact must be retracted when the fact becomes known), and belief revision (new knowledge may contradict old beliefs).

In spite of its complexity, as mentioned earlier, abductive reasoning is already at work in numerous computing areas. Take for instance agent-oriented programming. In one study, an analysis of how to create a multi-agent system contemplated how to “implement a form of coordination for multi-agent computations extended with abduction. In this framework, the computation is carried out by several agents executing in parallel for solving goals. To make such an environment work, “Each agent has its own (possibly incomplete) knowledge base, and uses abduction as a way of hypothetical reasoning.” Using such a concept, “each time one agent tries to assume a new hypothesis h, it must reach an agreement with the other agents, to verify that h is consistent with the knowledge of the other agents and with the assumptions” formed until that point. Under the model being considered, the “system can be partitioned among several bunches of arguing agents. Only agents belonging to the same bunch cooperate in performing an abductive computation.”⁴

Another study also values the ability of abductive reasoning to explain “observations by using background knowledge (such as the design) and relating possible causes to observed effects; thus, abduction would be beneficial in understanding agent behavior.”² Specifically, abduction would be “initiated as a query for causes of” unexpected or surprising observations. “The surprising observation generates doubt in the background knowledge. When abduction is triggered by a surprise, the abductive process attempts to explain the surprising observation and calm the state of doubt.”

An illustration of the possible route taken during creative problem solving is shown in Figure 1. It starts with a problem focus in one place, has the problem solver looking around, and often ends up with a novel idea somewhere else—possibly sparked by the juxtaposition of a “stray” thought against the task to be solved.

Back to Top

Abduction in Informatics?

With data collection burgeoning throughout computing fields, effectively using what is accumulated will take automation beyond today’s typical statistical analysis. Information is buried so deeply that “some data nuggets never hint at their worth as predictors or indicators when considered in isolation.”¹ As a result, “as we harvest more data faster, the challenge of making sense of it all becomes ever more pressing.” The use of data-driven discovery, using subconscious-level searching for relationships between unrelated data, may provide answers.

Relying on what some might consider unscientific methods, has considerable support. In a 1983 presentation to members of the National Academy of Science,¹² noted that “the fact that a person sometimes does creative things does not mean that he understands the creative process.” It was then suggested that “much of what goes on when we are thinking is inaccessible to our conscious awareness. We use such terms as judgment, intuition, and creative insight to name and label those phenomena that occur without awareness.”

Such search environments as KartOO.com (Figure 2) and Viewzi.com (Figure 3) embrace approaches that graphically group and display results. They allow searchers to view suggested relationships and clusters that are purported to exist amongst the search matches.

Bananaslug.com adds a random element to a user search to broaden the results returned from Google. For example, entering the three search terms “computer abductive reasoning,” followed by a click on the button labeled “Great Ideas,” resulted in the random addition of “citizen” to the Google search criteria. The modified hits included the following link to a related article (from the first screen of results).

Back to Top

Creativity That Goes Deep

Design consultancies value and encourage abductive reasoning alongside deductive… business design, global competitiveness, and corporate citizenship. (http://www.businessweek.com/innovate/content/aug2005/di20050803_8…-62k).

Searching Google directly for the original terms results in many more page of scrolling than most Web searchers will bother with—to happen across the same item. Another search tool that makes use of randomness is Stumble-Upon.com. Both programs end up with searchers finding results that are outside of the normal mix. Sometimes this might facilitate making an abductive connection to a new idea.

Supporting collaborative inquiry, while working alone, social book marking approaches, as well as Amazon. com’s “Customers who bought this item also bought” approach, likely both promote new ideas through juxtaposition of what the searcher was interested in with what other similar inquirers found of note.

Is there a need for abductive reasoning in Informatics? The evolving discipline certainly covers a wide range of areas as various programs are being implemented worldwide. Consider this sweeping look at the field from an investigation entitled “Information in Formation: A Peircean Approach⁵:”

What is today called informatics or information science has recently evolved beyond the realm of computer science and information technology, at least at some institutions, to become a confluence of studies in artificial intelligence, cognitive science, formal logic, and other related activities that study how natural or artificial systems represent, process, transform, and communicate information, taking into account epistemic, sociological, computational, graphic, and even economical dimensions.

In answer to the issue of whether abductive approaches would prove useful in helping computing users meet their needs, it is likely that Peirce would have felt that abductive processes were already at work in many realms today, whether those relying on it are aware of its potential or not. It would appear that a more worthwhile question to pursue would be how to better promote the use of abductive reasoning.

Back to Top

Conclusion

From the onset of this effort, expectations were not to solve anything, nor achieve statistical significance. Rather, it was hoped that this review would introduce faculty and researchers to the concept of using abductive techniques within Informatics. The value of this is that it could assist future workers to develop creative approaches to ever-changing problems.

The ever-expanding use and usefulness of the Web for research, has changed the techniques used to search for new or supporting ideas. The Web is used as “a source for learning and exploratory discovery.”⁹ This organized searching by trial-and-error is just the situation that will be facing today’s developers tomorrow.

While it was found that abductive techniques are currently applied in a number of computing areas, little was found to suggest the concept is being taught to Informatics or other computing majors as a method of fostering creative thinking. We hope to help change that by promoting the conscious inclusion of abduction in computing curriculums. This would expand the number of workers and researchers considering its role and value with respect to how Informatics could solve problems that are vague or unknown. Certainly the field needs new tools to progress beyond being data rich and discovery poor.

Perhaps looking for new uses of abductive approaches will help find creative solutions. The basic task before informatics educators, then, is nicely summed up in an Adobe advertising poster for Creative Suite 2 (2006): “Think. Think. Think. Create. Getting the idea is the hard part.” Perhaps abductive reasoning can help us discover what we are not even looking for.

Back to Top

Back to Top

Back to Top

Back to Top

Back to Top

Back to top