A Holistic Approach to Deep Learning and Expertise

Key Insight

Effective teaching extends beyond merely imparting facts that students often forget (up to 90 percent within weeks) to shaping their perception of the world and helping them absorb the underlying rules of a discipline. This involves an 'apprenticeship' model where the teacher demonstrates a way of thinking through problems, encouraging student participation and imitation for unconscious learning. Integral to this approach is the deliberate creation of 'emotional experiences' through mistakes and the effort to overcome them, which helps engrain knowledge more deeply. Students are also guided to interrogate their unconscious opinions, recognizing that understanding often involves discovering ideas already present within.

This approach emphasizes hard work, not just natural curiosity, including assigning unwelcome homework and frequent tests. Retrieval practice, such as testing, strengthens neural networks related to knowledge. The ultimate goal is to cultivate autodidacts who become 'addicted' to the pleasure of discovery—the 'jolt of pleasure' when effort and suffering lead to understanding. This learning journey involves multiple stages: initial knowledge acquisition through self-directed reading from diverse sources, prioritizing engagement over immediate brilliance, as 'the effect of this first phase of learning seemed to be to get the learner involved, captivated, hooked, and to get the learner to need and want more information and expertise.'

The second stage focuses on automatizing knowledge through repetition, transforming conscious understanding into unconscious competence, akin to learning to drive a car. This frees the conscious mind for new challenges, as 'Civilization advances by extending the number of operations which we can perform without thinking about them.' Learning is a non-linear process involving 'reach and reciprocity'—venturing out from core knowledge to new information and then integrating it back. Expertise is not just accumulating facts but internalizing relationships, forming internal connections, and creating 'networked chunks' of information, allowing experts to see 'formations' and 'stories' rather than individual pieces, as exemplified by chess grandmasters remembering game boards, but not random piece arrangements, vastly better than average players.