Philosophy of Deep Learning

Visual Metaphor for Deep Learning

The company logo is the Sierpinski Triangle, an example of a shape known as a fractal. A fractal is a geometric construction that is self-similar at different scales. What this means it that it a fractal shape will look almost, or even exactly, the same no matter what size it is viewed at. The logo is the third iteration of the fractal. Start with an equilateral triangle and fill it with black (the first iteration). Mark the mid-point of each side and join the mid-points. Cut out the triangle in the middle (the second iteration). Repeat for the three remaining black triangles (the third iteration). This process can be repeated infinitely and if you zoom in to any depth you will always find an exact replica of the Sierpinski Triangle. This is deep, very deep, and is a visual metaphor for the educational philosophy of Deep Learning.

What is deep learning?

Deep learning is to be contrasted to surface learning. Surface learning is an approach to learning that has the intention of being able to reproduce content for the purposes of assessment. It involves a passive, uncritical approach to content. Deep learning has the intention of understanding for oneself and results in active critical interest the discipline as well as an ability to tackle unfamiliar problems. Whilst a visual metaphor for surface learning might be a straight line (memorising “one damned thing after another”). A deep approach leads to the development of conceptual understanding as well as the development of metacognition. Conceptual understanding can be viewed as the ability to abstract generalised ideas from the object of study. Metacognition (knowing how we know) can be viewed as the opposite process of understanding how these abstractions are related to the real world as well as being able to think that such abstractions are just one way of thinking about the world. This two-way travel up and down the abstraction chain is captured in the logo. The overall triangular shape reflects the ideal of a ‘theory of everything’ whereas the multiple triangles reflect the reality that all we might attain is a multiplicity of perspectives on the object of study.

Although the philosophy of deep learning is based on Noel Entwistle and colleagues’ studies of effective learning in the university sector, Rick Instrell has found that this is an effective approach to school learning. The approach was used in the design of the SQA Media Studies qualifications in the 1990s where there was effectively no specified content other than the application of high-level abstractions i.e. the ‘key aspects’ of Categories, Language, Narrative, Representation, Institution, Audience and Technology. Students are expected to analyse professional media content or their own productions by integrating understandings gained by use of a range of the key aspects.

See: Entwistle, Noel (2009) Teaching for Understanding at University: Deep Approaches and Distinctive Ways of Thinking. Houndmills: Palgrave Macmillan.

 

Header Go to CPD page Go to Home page Go to Consultancy page Go to Resources page Go to Adult courses page Go to Contact page Go to Contact page About the logo
Go to Contact page Go to Home Page Go to Contact page Go to Adult Courses page Go to Resources page Go to Consultancy page Go to CPD page Go to Home Page