Kaizen

Bad News from MIT for the PDCA Cycle?

Avatar photo By Jon Miller Updated on May 19th, 2017

According to research reported from the Massachusetts Institute of Technology (MIT) Picower Institute for Learning and Memory we don’t learn from our mistakes. Oh-ho. If this is true the whole scientific basis for kaizen is in trouble. Professor of neuroscience Earl K. Miller (no known relation) is quoted, “We have shown that brain cells keep track of whether recent behaviors were successful or not” regarding a series of experiments involving monkeys. It appears that monkey brains process information more sharply after a success, but not after a failure. The conclusion presented from these experiments is that brain processing improves after successes, but not failures, and therefore we don’t learn from our mistakes. There is a bit of a leap there. On balance most of us fail far more often than we succeed, so we could say that in total we learn more from mistakes than successes. Also, unlike chimps who have been wired up in a Massachusetts lab, we can reflect on our mistakes rather than apishly repeat our successful or failed behaviors. Of course this requires that we choose to do so. If our brains are wired to gain no performance boost from failures, while for whatever reason such as the chemical rush of success or the stored information on success parameters in neurons, we need to deliberately take advantage of our failures.

We must reflect on the mistakes, as well as the successes, and turn the PDCA cycle back to plan. This is where lean thinking is once against counter-intuitive and perhaps against our very biological nature.


  1. Emil

    August 30, 2009 - 2:41 am
    Reply

    Kaizen is not in danger. This research emphasize that we must have the mindset that learning from failure => success for us and out tribe. Example: if you get rewarded and maybe promoted for finding a problem then the failure leads to success. The research however indicates that we are bad at learning from our own failures. Maybe we should always work in pairs

  2. Paul Bayer

    August 30, 2009 - 4:11 am
    Reply

    Hi Jon,
    as Deming said:”Without questions there is no learning”. We can learn from mistakes only if we have a questioning attitude. So the most important thing are questions.
    My second point is, that in experimenting we should try different things. So we learn more easily from our failures since we see what works and what doesn’t work. This is the power of designed experiments.
    Cheers,
    Paul

  3. John Hunter

    August 30, 2009 - 9:46 am
    Reply

    Actually I would say this is another reason the PDSA cycle is so important. Without it we are much more likely to fail to learn from mistakes. By following it as it should be (see the Improvement Guide for an excellent book on the topic) you will learn not only from success but failure. And you will actually wait until success is shown before adopting broadly just because you think it should work. Related: http://management.curiouscatblog.net/2007/03/01/write-it-down/ http://curiouscat.com/management/theoryofknowledge.cfm

  4. jamie flinchbaugh

    August 30, 2009 - 11:23 am
    Reply

    Interesting choice Jon. The build on your point, we must choose to do so. And the best way to choose, is to have a PROCESS that forces us into the opportunity to learn. Learning is both a process and eventually a habit. It doesn’t happen, at least not effectively, without an underlying reflection and learning process. That’s why we love the process of experimentation, PDCA, and After Action Reviews.
    Jamie

  5. Anonymous

    September 5, 2009 - 12:30 pm
    Reply

    When did MIT start admitting chimpanzees?

  6. Richard Perrin

    September 10, 2009 - 3:55 pm
    Reply

    Jon – my thoughts on this are that learning from our mistakes does not equate to doing something about it. Robert Sutton of Stanford gave this phenomenon a perceptive review in his book ‘The Knowing-Doing Gap’. Toyota embraces the idea of ‘fail often, fail early’ to eventually produce the most robust designs, expressed in their process of Set Based Concurrent Engineering.
    The idea of embracing failure as an integral part of any deep learning is usually a culture changing challenge for most organizations. There are many business cultures that do not tolerate mistakes – failure is an orphan. It takes counter-intuitive leadership to embrace the failures and take action.

Have something to say?

Leave your comment and let's talk!

Start your Lean & Six Sigma training today.