Kevin Dunbar

• Research on how scientists think and reason and how scientists train graduate students and post-docs. Over the past decade we have investigated the ways that scientists think “live” at lab meetings and how students in the labs are trained to conduct science. We have also gone back into our own lab and conducted controlled experiments on phenomena that we have identified in naturalistic settings. This methodology I have labeled InVivo/InVitro Cognition.

What are the findings or theories from your area of expertise that we could apply to higher education?

Analogy is a key process used to both educate scientists and to help scientists make discoveries. The classic findings in psychology are that people have great difficulty at using analogy; either in remembering analogs or using analogs to solve problems. Experimenters have shown that people tend to rely on superficial similarities between components of an analogy, rather than the underlying structural relations.   In my research, we have found that analogy both based on superficial features and on deep structural features is frequently used.  What is most important is the goal that the reasoner has. Superficial similarities are good at fixing minor problems, more structural analogies are good for formulating hypotheses, and really far analogies are good for explaining things to others. I have also found that analogy is a key to training scientists and that it is key to helping scientists make discoveries. Education must tailor the analogies to fit the particular goal. In addition, there are inherent dangers in making analogies, such as only seeing the similarities and not seeing the differences that can lead people astray. These problems with analogy also need to be confronted.

Distributed cognition.  What happens to thinking and problem solving in the group process?  What type of interactions lead people to make new discoveries and learn new things?  The composition of the group is a key factor.  Groups derived of people from the same background do not solve problems as easily as groups with people from different backgrounds with similar goals.  One of the problems is that people from the same background all draw from the same base of knowledge, those with different backgrounds are able to pull in different types of knowledge.

If you want to use groups to learn, it is good to have groups with different backgrounds.  The key, however, is that the groups must have a common goal.  If group members have different goals, the group will not learn, and the results can be disastrous.  I have studied groups with common goals and with different goals and found that the reasoning and problem solving strategies used by the groups differs widely. While collaborative learning is a hot topic in education, it is important to structure the groups in ways that will facilitate learning. Overlapping goals and knowledge is the key factor.

Training Methods.  We have also been looking at how people plan experiments and have found that the classic training in experimental design (using control groups and experimental groups) is often too basic for the work that current scientists are doing.  When we are training scientists, we are often doing so in very simple ways. Simple research designs are not consistent with what scientists actually do.  For the most part, scientists are currently trained in an apprenticeship method.  Students are expected to learn experimentation by being in the lab, as if through osmosis. This sort of training for future scientists is very much a hit or miss process.   It can result in a loss of some very talented students that are just not given enough instruction or motivation to continue to become scientists.  Thus, we are faced with the problem of preventing talented students from dropping out. This is a very large problem all across the educational spectrum.  

Unexpected findings in research. .  Unexpected findings account for about 50-80% of the findings.  There is little explicit training  on how to do this and scientist are very vague and quasi religious about what they really do talking of the goddess Fortuna and flashes of insight. We have found that rather than being the victims of chance, scientists have well developed strategies for dealing with the unexpected. Professors tend to blame the method; post-docs and grad students tend to think that they have found something new.  The types of analogies used and distributed reasoning strategies that are used vary as a function of how many unexpected findings are obtained. Furthermore, there gender differences in the way s that unexpected findings are treated: Men are more likely to assume that they know what was wrong.  Women are more likely to repeat the experiment with different hypothesis and to systematically eliminate problems.  Dealing with the unexpected should be a key component of our science education.

What are the (most important) unsolved problems? What should be included in an agenda for research?

Looking at how distributed reasoning works in groups.  If you want a group to work the group has to have not only common goals, but also must have knowledge about the problem.  Much of the work that has been done demonstrating that groups do not help in problem solving and reasoning has often given the groups arbitrary problems to solve that they do not care about.  To understand how groups work in thinking and reasoning processes, research must look at groups that are involved in solving problems that the individuals in the group care about solving.

There is also a problem with our culture of the individual.  A great amount of science research is done in groups and in lab settings.  An interesting occurrence is that someone working in the group will come across an unexpected finding that they cannot explain.  They will then bring this finding to the group and will begin the process of discovering some meaning to the finding.  If you then go back several months later and ask the person about the discovery they will say that they did it all on their own, even when you have the transcripts that demonstrate that the group was actively involved in changing the way that the person thought about the problem.  This notion of the “lone researcher” is definitely something that we need to move away from.  One way to do this is implement group work earlier in the educational experience.

What prototypes can you point us toward where principles from the science of learning are already being applied?

I don’t think that there has been a lot of application of the science of learning principles.  Analogies have been used quite a bit, but have shown mixed results in their level of effectiveness.  With analogies it really depends on how they are used.  There is also the MIT media-lab approach that uses computer programs to help guide learning, however unless you give people underlying structure this is not going to be very useful.

Some other examples of prototypes include case-based reasoning approaches and the work by John Anderson.

What are the major problems with or barriers to redesigning higher education? Do you have any ideas for overcoming them?

It is not simple and it is not straightforward.  There are no easy answers and there is no magic wand that you can wave that will instantly make people better learners.  People have unrealistic expectations about what can be done.  Applying new techniques will generally only bring about modest increases in performance. 

We have things such as drop out rates, especially among women in the sciences. The limited numbers of women in sciences is a great waste of talent and is an example of a problem that cannot be solved by simply changing the cognitive approach to education.  There are many things that intertwine to cause this problem, issues related to the culture and to various social issues.  It is important that we include motivation and social/cultural issues in our research.  Merely making changes in the cognitive area of education will not necessarily change things like drop out rates.  We need to look at the cultural and social issues as well.  We need both situational research and educational research.  The divide between these two areas is a major barrier that we need to overcome.

What do we need to do so that one outcome of the retreat is to effect change (in ways that we want)?

Break people into small groups with a common goal.  Want to avoid having the groups argue that their specific solutions are right.  What would be helpful is to get people to come up a list of various options a number of different strategies that could be used in small groups and then getting these lists out to the larger groups.