Feedback Loops in Learning

The feedback loop is required for all learning. You can encode responses to stimulus in DNA or programming, but learning is the ability to selectively change thought or response based upon feedback. The internet wouldn't exist without feedback loops, and neither would education. There is a multitude of other factors involved in human learning, but feedback loops are one of the most basic, and most easily resolved factors.

I’d like to describe how the feedback loop can be optimized in human education with software. However, first we must briefly explore the feedback loop in its simplest, most efficient form.

The two key parts are highlighted in red. Both of these must exist in the feedback loop for learning to occur. Without one, it degrades into either random response, or simple stimulus/response. When no feedback is collected, an entity may act based on pre-programmed behaviors to a changing stimulus, but it will not be intelligently defined action. When the entity does not select a response to the stimulus, it is doomed to either perform the same response for eternity, or give no response at all.

Now we come to optimization factors. The example below is a common school feedback loop.

1. Student gets homework assignment
2. Student answers all homework questions (Response Selection)
3. Student turns in homework assignment
4. Time passes
5. Teacher grades all questions
6. Time passes
7. Teacher returns assignment
8. Student reviews graded assignment (Feedback Collection)

The above homework assignment creates a feedback map like this:

It takes 35 steps to complete 1 feedback improvement loop. This type of feedback loop is good for a final examination of learned knowledge, but not for practice homework. So how do we fix this?

Loop Steps

A measure of efficacy for a feedback loop is the number of steps involved before a loop is closed. Most feedback loops are more complex than the basic 4-step example. Like our homework example, often there are multiple process loops interwoven into a single feedback loop. Reducing steps in a loop can decrease complexity of response selection, reduce variables in response, reduce time, thus increasing efficacy.

Loop Speed

Fewer steps should also decrease loop time, which is another important factor. In this homework example, the student may respond incorrectly several times before getting feedback, reinforcing incorrect concepts. While waiting for grades, the student will make the same mistakes as in the assignment, crystalizing incorrect learning. The faster a feedback cycle is complete, the more informed subsequent responses will be.

Loop Efficacy

The last measure of performance for a feedback loop is effectiveness. Was the proper response chosen? Was a concept only partially learned? Was the response detrimental? Did the response change based on new data? Can it be applied in different contexts, or combined in novel ways and still used properly? These are all ways to view the efficacy of response selection after a feedback loop. It is up to the validation manager (or teacher) to ask these types of questions when evaluating studies.

The Solution?

Using software, we could greatly improve the feedback loop speed as shown below:

Since we can't all have teachers in our home, but we can have software, programs and apps can unlock capability that was previously impossible. Duolingo is one of my favorite learning apps because it does exactly this. It reduces the feedback loop of an assignment to its simplest form, improving accuracy with each question answered. Proper feedback is given for both correct and incorrect responses. If all education could be performed in this way with shorter and more engaging feedback loops, we would have higher participation, and higher efficacy in learning.

---

---

What other measures of performance for a learning feedback loop can you think of? Please feel free to share if you come up with any!