Site icon ICE Blog

Learning Curve Basis of CBME: The Nonlinearity of Learning

Admin note: This blog is Part 3 in a series! Scroll to the end of this post to view a list of the other posts.


By: Martin Pusic (@mpusic) and Kathy Boutis (@ImageSim)

Learning is Non-Linear

There’s a reason it’s called a learning CURVE! In this post we’re going to focus on the notion that learning is generally non-linear.  This runs counter to our usual perception that equal amounts of effort are repaid with an equal rise in the learning. In Figure 1 we demonstrate that over some stretch learning does indeed look linear. However, look more closely and you see that there two inflection points, one where learning accelerates and a second where it decelerates. Let’s see what’s going on there. 

Figure 1:  Prototypical Learning Curve showing its non-linearity

A prototypical learning curve demonstrating the non-linearity of learning, including two inflection points where learning first accelerates and then decelerates. The apparent linear phase in the middle of the learning curve suggests a more uniform process than is actually seen. 

First Inflection Point – Acceleration

The first inflection point marks the end of the latent phase (See Post #2 here) after a period of organization, all the elements are in place for measurable improvement. The learning conditions are optimal with each new bit of learning reinforcing the previous one, and more. With each repetition the mental model gets exponentially stronger. Consider the radiology intern who sees their first two cases of ankle trauma with one being due to Diagnosis A and the other to Diagnosis B. What they learn is: what diagnosis A looks like; what diagnosis B looks like; how Dx A is different from Dx B; how each are (wildly) different from their preconceptions. All in two cases.  In the early stages a learner is like an enzyme awash in substrate, merrily consuming it in service of filling in a mental model that has lots of space available. 

Second Inflection Point – Deceleration 

But the good learning times cannot last forever. As the learner progresses up their learning curve, at a certain point the learning rate (the amount of learning per unit time or per repetition) begins to decelerate.  This is usually due to good things: the learner’s mental model is more solid now with fewer gaps and fewer misconceptions to correct. They now have a good understanding of common cases so that each time they see another case of Diagnosis A what they can incrementally learn is less, especially when compared with how much learning they extracted from that first case of Diagnosis A. The enzyme is having a harder time finding substrate.  Learning is now more effortful, a hunt for rare cases or variants, the final pieces in the mental model jigsaw puzzle.

The Asymptote

Beyond the second inflection point is the region of expert learning: the asymptote. This is where learning butts up against the partially or completely unknowable in a domain. Cases where a fully trained master clinician has well-founded uncertainty…but can learn through the habits like feedback-seeking (e.g., being assiduous about follow-up; consulting colleagues). The key here is to NOT think of this phase as a plateau, a word we actively resist (Watts1). Instead, an asymptote is a sloped line that keeps increasing out to infinity, representing a lifelong learning journey towards realizing the full potential of a particular learning system/environment. This is an idealized representation, of course, with many confounders that might be more important to the final result; however, no matter how much of the asymptote is real and how much is metaphor, the notion of striving towards improvement in the region where improvement is most incremental, most difficult, can engender a response process that is consistent with the highest possible levels of clinical care. This is the basis of Anders Ericsson’s tireless advocacy for the highest levels of expertise, ideas that apply widely (Ericsson2).


If educators and learners embrace the non-linearity of the learning curve, there are several ways they can customize their approach based on where the learner is in their journey. Early on, at the first inflection point, educators can just get out of the way and reap the benefits of having set up the learning conditions well. The second inflection on the other hand requires coaching. Here we need to help our learners break past the “false dawn” where the slowing of learning is misinterpreted as competency*. Instead, in many respects, the development of true expertise has only just begun.  The senior pediatric resident who has seen 40 asthmatic patients probably has yet to see their first case of myocarditis presenting like asthma. The key is to celebrate the final asymptotic phase of the learning curve as being what true expertise looks like: an ongoing “feral vigilance” for improvement (Croskerry3).

*See our blog post on Competency Standards and the Learning curve

Previous Blog Posts in this Series:
Part 1: Overture Click here to read
Part 2: The early phases of learningClick here to read

Future Blog Posts in this Series
Part 4: Standard setting and the learning curve Now available Click here to read!
Part 5: Inter-Individual variability of learning curves (coming Dec 15)
Part 6: Summary (coming Dec 22)

About the authors:
Martin Pusic, MD PhD is Associate Professor of Pediatrics and Emergency Medicine at Harvard Medical School, Senior Associate Faculty at Boston Children’s Hospital and Scholar-In-Residence at the Brigham Education Institute. 
Kathy Boutis, MD FRCPC MSc is Staff Emergency Physician, Senior Associate Scientist, Research Institute at The Hospital for Sick Children and Professor of Pediatrics at the University of Toronto.


1. Watts C. Escaping the Plateau: How to Keep Climbing When Your Language Learning Goes Flat | by Engramo Team | Engramo English Blog | Medium accessed 11 Nov 2022. 

2. Ericsson KA. Acquisition and maintenance of medical expertise: a perspective from the expert-performance approach with deliberate practice. Academic Medicine. 2015;90(11):1471-86.

3. Croskerry P. ED cognition: any decision by anyone at any time. Canadian Journal of Emergency Medicine. 2014;16(1):13-9. 

The views and opinions expressed in this post are those of the author(s) and do not necessarily reflect the official policy or position of The Royal College of Physicians and Surgeons of Canada. For more details on our site disclaimers, please see our ‘About’ page

Exit mobile version