How to encourage deep learning of your students and discourage their habits of shallow or superficial learning?
Shifting paradigm from informative learning to transformative learning is crucial to encourage deep learning. I believe that student engagement is an essential aspect of meaningful learning. Hence, implementing different means of better engaging pedagogical approaches such as active learning, learning communities, service learning, cooperative education, inquiry, and problem-based learning, and team projects are central.
My first association on this question is connected with the whole society's picture. Namely, our students are looking - on daily basis - that many ''important'' persons doesn't have deep understanding of many concepts that they are dealing with. It's very clear message to them - you don't need to learn too much, but to be skilled in some other areas. Maria Montessori told that child is always adopted to his/her time and environment.
I think deep learning can be encourage by giving the students a lot of practical exercises ( learning by doing). When students are actively involved in the teaching/ learning process they learn better. This can be done by using indirect teaching methods and strategies. Students can also be encourage to use metacognitive strategies. Metacognitive strategies help learners to control their own learning.
Building on the seminal work of F. Marton and R. Säljö (“On qualitative differences in learning: 1– Outcome and Process,” British Journal of Educational Psychology 46, pp. 4-11 (1976)), The BYU centre for teaching and learning have produced this:
https://ctl.byu.edu/tip/teaching-deep-learning
There are some really useful strategies in here to add/ammend to your teaching to promote deep learning.
Superficial or shallow learning is not a weakness of the brain that needs to be turned off or treated. Nor is superficial learning attributable to laziness and lack of interest on the part of students - quite the opposite: superficial learning is the result of a high supply of stimuli, which at the same time is a challenge to the selection function of perception with limited learning capacity of the psycho-physical system of each individual. Superficial learning is the form of coping with life that corresponds to life in the modern age. The academic no longer learns profoundly from books - otherwise he would have to read all the books - but instead reads a Wikipedia article. And at best his knowledge consists of such superficial information, if practice is added to information processing and application. This is a fairly normal process - a process of adaptation of the organism to the flood of environmental stimuli.
For professional purposes - deep shallow learning for the selected topic/domain. However, I'd suggest a reasonable combination of both, because a broad general education is also necessary.
Deep learning should be encouraged for students. Working students on individual projects and by objectives can help more deep learning. When students are involved in learning process, they improve their learning more and more. We should to make them responsible and work in competition.
Students should be provided with an example for everything taught to them. Additionally they should be given a demo or real life application for the same. It can start with a simple project on the subject taught as well as an interesting assignment.
Based on the cases I have experienced, engineering students are very smart and can, in many cases, get the information in a quick and easy manner. The bad point is, they become disconnected by certain issues such as: repetitions, too much theoretic issues, long lectures. After knowing the illness, I believe the following points might give a cure:
- let them work in simple projects progressively developed.
- Integrate them in the lectures/workshops, that would let them be more curious, motivated and eager to learn.
- Show them videos on how to do things, how people around thinks and create (3D works and good presentations always work on that too).
- Finally, let them follow a group of book references with much more info, details (1-2 could be enough).
My regards! Ahmed
Hi! Deep learning comes primarily from deep interest and personal true engagement in something, especially from constructing some artefact (physical or abstract) and using it. Shallow learning comes from trying to find the shortest way to pass a test - which is also the fault of the examiner if the test clearly rewards this shallow learning.
The temptation to learn complex facts only superficially is particularly great today because the offer of superficialities is immense. Even a test can be easily passed with superficial learning.
There is a challenge for students not to acquire superficial but thorough knowledge: When you lead a learning group and need to explain to other students what theories are important and what research results are available for a particular area.
Student centered learning that depends on small group teaching, criticsl thinking, reflection and formative assessments may help to improvr learning
Shifting paradigm from informative learning to transformative learning is crucial to encourage deep learning. I believe that student engagement is an essential aspect of meaningful learning. Hence, implementing different means of better engaging pedagogical approaches such as active learning, learning communities, service learning, cooperative education, inquiry, and problem-based learning, and team projects are central.
I think it is not about encouraging or discouraging but to teach in what situation deep learning should be employed and when shallow networks are the solution.
John H. Navarro-Devia :
It depends on some parameters:
- Size of dataset is important (typically for big data, deep learning is practiced. It also depends on the number of features or the dimension of your dataset. There is no rule, but depending on the size of dataset, if your data is high-dimensional, you may try deep learning or dimensional reduction methods (PCA, FDA, etc.) and the apply a shallow ANN.
- What is your data? Are you dealing with images, text, voice, or just numbers? If your inputs are images, the maybe a CNN is your best choice.
- feature selection methodology is also important. Are you planning for fully automatic feature selection or manual feature selection is OK as well? To answer that, you may consider the uncertainty level you are facing in your application. Does manual frature selection is satisfactory enough in most cases or not? Deep learning is an option for automatic feature selection, where you do not intend to do sophisticated signal/image processing by yourself.
All in all, if you are dealing with big data or image/text and plan for automatic feature selection, deep learning is highly recommended. However, you may still (depending on the size and complexity of your dataset) practice dimensional reduction method combined with a shallow ANN. The most important issue is the robustness of the model and avoiding issues like overfitting. We should bear in mind that choosing more complex models might cause the system to have more parameters that should be tuned during the training process. So, the question is do I have enough data for such a tuning. If our model has 100 unknown parameters and the size of our dataset is also 100, we might have a challenging task to do!
- Badges
- Science topic