This is a review post on the article “Why Minimal Guidance During Instruction Does Not Work: An Analysis of the Failure of Constructivist, Discovery, Problem-Based, Experiential, and Inquiry-Based Teaching BY Paul A. Kirschner , John Sweller & Richard E. Clark
Following are 6 major points of contention I would like to make as part of my critical review of this article.
1. A Polarized argument based on Cognitive Architecture
Here I must mention a concept often talked about by Charlie Munger; “MAN WITH A HAMMER SYNDROME” i.e. – To the man with a hammer, every problem looks like a nail. This means people are biased to use the tools they possess to solve problems, regardless of whether such tools are appropriate for the problem at hand.
This could be the problem here too.
The authors start the argument by establishing the existence of a fundamental cognitive architecture, which includes Working memory and Long term memory. They use this conceptual framework to make the point that Constructivism is bad for cultivating such fundamental aspects of learning as long term memory and working memory.
But, In my understanding, Constructivism is not an ideology that is set against or has any inherent disposition against fundamental Cognitive-Architecture. In-fact since it is integrative in approach, it can even be adapted to suit the better cultivation of this fundamental Cognitive-architecture, If that may seem fit in a particular context.
Further, In the real world, things are not black and white. There can be multiple ways of combining things together. A teacher or school may subscribe to Direct teaching methods to deal with more traditional aspects of learning. They can also have elements of constructivism
For example, they can teach and learn theory using the Direct approach and try things practically using Constructivist approaches. So, it is totally possible to integrate appropriate elements from each approach to suit situational needs.
2. Reference to expertise literature.
As part of their effort to establish the importance of working and long-term memory, the authors try to use expertise literature, i.e. referring to studies done by De Groot, Chase, and Herbert Simon on chess expertise. The problem with this approach is that expertise is local to the chess game, we can all agree on that.
Similarly, we can also agree that Car-driving-expertise is local to the act of driving a car in a specific environment. Just like that, Educational teaching and learning are supposed to have local expertise, e.g. test-taking, or memorizing, which is optimized for the local evaluative cultures.
But I guess that is not the claim of this article. Its application in education is sometimes about General-Intelligence and Expertise.
If you know anything about expertise literature, its foundation is based on an understanding or studying of real expertise and debunking claims of generalized expertise, and false expertise, or concepts like that for E.g. Dunning Kruger effect.
The reductionism is evident when the article has to rely on Controlled studies, as evidenced in the conclusion paragraph; i.e. :
“In so far as there is any evidence from controlled studies, it almost uniformly supports direct, strong instructional guidance rather than constructivist-based minimal guidance during the instruction of novice to intermediate learners.”
Updated: A tweet from Nassim Taleb about the 2-dimensional correlation.
I appreciate the contribution and importance of Cognitive science, particularly the Cognitive Load Theory, but I think it is playing a disproportionately large role using its reductionist lab (e.g. paper tests, MCQ tests) methods, which mostly lacks external validity and often fails in replication studies.
My argument is that real-world learning is often complex, embedded, and embodied.
Going against my previous criticism of using expertise literature; even if we consider studies on such expertise, a recent meta-study found that deliberate practice explained only 26% of the variance in performance for games,21% music, 18%sports, 4% education, and less than 1% professions. This means that even in domains of extremely structured knowledge like chess and music, the correlation with deliberate practice to performance is low. Consider then the correlation with instruction(which is one or two layers outside of real learning), which often does not even take account of individual dispositional factors like- pre-existent skills, socio-economic background, disabilities, etc.
Further(a relevant tangent), in society, we have social systems, organizations, institutions which form part of our distributed cognition, which solves complex problems for us every single day 24/7. We have decision support systems like in air traffic, we have intelligent apps, we have brain-friendly GUI’s, we have robots, we have google to find information, etc.
The point I am trying to make is–We are evolutionarily superior because of our cumulative culture and tool-building capacity, not essentially because of our test-taking ability, which is the de-facto tool used by cognitive science.
I am all for tests for learning and assessment, I appreciate its value, but I will always argue that it should be limited. It should not overstep its scope to do things like branding(often entire racial groups) people as Low IQ, etc( I am not saying this particular article does this).
Also, we must always remember that, in evolution, committing errors is more valuable than giving a correct answer like a machine. If a task is valuable and repeatable(predictable) human society will always find a way to solve it with ease(law of least effort). The cognitive version of education proposed by this article is all about single mind cognition with no consideration to cumulative culture and tool building.
Note: Same authors(of this article) published their paper on “Collaborative Cognitive Load Theory.” which I will review in another post.
4. Correlation & Causation
The next logical argument from the other side will be to show the correlation of success with the test-taking ability(or achievement in general).
My rebut will be to show that it is the social system and structure which decides individual success, not an arbitrary cognitive test.
When you require higher points in Cognitive Tests to access institutions, success in that test can show up as a correlation to success. But this success should not be confused with causation. The test doesn’t cause any success by itself. The social-institutions used the test to give more opportunities for people to learn, grow, and brand. It is this ignored variable that caused the ultimate success.
For example, to get admission to a great Masters-program, you need good grades and often outstanding performance in tests. You have to have a good Masters’s degree to get a good research job, which provides you the opportunity to learn embedded tools like SPSS, Stata, R, etc. These tools are necessary for Ph.D. admission.
Thus the real game-changer is access to institutions and their embedded tools. So, access to the tools and skill-producing actions depends on access to institutions that work with that tools. People who work in places like Google, NASA or CIA, will most likely pick up useful skills and tools which help them accumulate more skills and material prosperity in the future.
Related and relevant tangents on how institutional evaluative culture impacts life
Jerome Bruner(quote on meritocracy)
Relative Age effect + Mathew Effect +Pygmalion effect +
5. Nature Of Information
How much can, and should we remember in a dynamic and information-abundant-world is a very important question we should ask. How much fundamentally transient information should we remember? Who gets to decide. Are we deliberately creating an educational system in which blind-men( with reductionism) leading blind men(young children and their parents.).
If you are teaching a particular information today, how will you decide the value of that information 10 years from now? How will you value the value of information in comparison to another?
Secondly, Information also has dimensions. How will you decide on what dimensionality of information you should teach)? For example, you can make students memorize the periodic table, but in a real practical situation, as in a lab, the dimension of retention may not be valid. They may instead use heuristics, search-based thinking, analogical thinking, recombination, or generate ideas using accidents as a result of making mistakes, etc.
Another possible issue with this approach is that of the schema consolidation and shutting down of novelty receptors. This could lead to habituation, lack of creative and innovative thinking, etc.
To help students learn effectively in a dynamic world, it is important to understand the science of motivation.
It should be noted that motivation has an important role in establishing learning as a permanent action, and not just as a part of a structured curriculum.
The Direct Instructional methods may stress external rewards more in the classroom, and by doing so it may diminish the intrinsic motivation of students,
The goal of a sensible and adaptive educational system should be to cultivate Intrinsic motivation that enables students to work independently and enthusiastically even after the school years are over, Or when nobody is there to push them to learn things.
It is something we may lack in the case of the Direct approach advocated by this paper.
One thought on “Constructivism vs Direct Instruction”