What are the limits of feedback? - Chapter 11

What does claim 6 entail?

Claim 6: To get people to learn, give them feedback on the consequences of their actions.

You can’t learn without feedback. Faster feedback is better than slower feedback. More feedback is better than less feedback. In countless experiments experiments, with animals as well as with humans, feedback is essential for helping people become more skilled. Is this always true? Before we think feedback is always perfect, we should know what its limitations are,

When is feedback not sufficient?

Feedback about consequences helps us judge if we are making progress.

It can motivate us to do better. However, outcome feedback doesn’t help us figure out how to do better.

For example, in archery, if a shot continually hits to the left it means, I have to aim more to the right, but does that mean changing my aim point, my stance, my shoulder alignment, or the way I lean my head over the end of the arrow? Perhaps the wind is affecting the shots.

What can be concluded? Outcome feedback doesn’t help us diagnose what we need to improve. To improve, we need process feedback about the way we are acting. I need feedback on the way I am aiming and releasing the arrows.

Researchers have found that outcome feedback (knowledge of results) doesn’t improve our performance as much as process feedback, which helps us understand how to correct flaws. But process feedback is harder to come by.

If we get process feedback, we still have to make sense of it. To learn that a certain set of actions will lead to good outcomes, we have to figure out the cause-effect relationships between the actions we took and the later events.

Learning is more than just making connections. Intelligent cybernetic systems can strengthen relationships that are followed by positive feedback, and ignore (or weaken) connections followed by negative feedback. If we don’t get the positive feedback that we expected, we try to figure out what went wrong. When we get negative feedback, we try to understand what we did wrong. Cybernetic systems don’t do any diagnosis.

It can be impossible to understand feedback if we distort it

We all simplify the world. We chop events into artificial stages. We pretend that simultaneous events are sequential, that dynamic events are static, and that nonlinear processes are linear. We deal separately with factors that are interacting with one another. It is hard enough to understand accurate feedback.

It can be impossible to draw the right lessons from feedback that we have distorted. Even if feedback is clear, we may distort how we remember it. For example, typical advice to students taking Multiple-choice tests: “Your first impulse is correct”. Many students interpret this advice to mean that they shouldn’t change an answer if they go back over the question and suspect that another answer is correct. They are much too reluctant to change responses.

Also trained analysts may run into trouble with the way they remember feedback.

Feedback will not help if people can’t detect it

This is another reason why giving people feedback isn’t sufficient. They have to notice it and attend to it. How can we learn from our mistakes if we haven’t been watching the right things?

Example: novices vs experts in car simulations. The novices were much less likely than experienced drivers to direct their attention to potential hazards. They just didn’t know where to look. Not only were they less able to respond to a hazard, they were slower to spot it. That would make it harder to figure out later what caused an accident or even to be aware of near misses. Feedback will not have an effect if drivers aren’t looking in the right place to see the critical cues.

So far these were examples of difficulties we have in learning from feedback in complex situations. Another difficulty is in giving feedback, particularly about subtle cues and tacit knowledge.

Why is it hard to provide feedback?

It is particularly hard to give feedback on tacit knowledge. If someone is trying to coach you, and wants to built up your expertise, how can they give you feedback about the accuracy of your mental models, or about your ability to notice subtle cues, or about your ability to sense what is typical and what is unusual? It is very difficult to notice these kinds of knowledge.

Aside from the difficulty of giving feedback about tacit knowledge, instructors who give us rapid and accurate feedback can actually interfere with our learning.

Schmidt-Wulf’s research involved motor skills rather than cognitive skills. They found that continuous , concurrent feedback does increase our learning curve while we are being trained, but that it reduces any performance gains once the training is over. The reason is that we never have to learn how to get and interpret our own feedback as long as the instructor is doing all that work. The performance goes nicely up in the training environment, but then we are at a loss once we move into our work environment.

We would be better of struggling to get our own feedback than having it spoon-fed to us by the instructors. We have to learn for ourselves how things work and how to evaluate.

The very idea of giving feedback suggests a passive learner. Learners are better off seeking their own feedback and asking for advice when they need it. Giving feedback is probably the most important thing an instructor can do, and even good instructors are often challenged when it comes time to give feedback.

Why is feedback not always necessary?

Feedback applies more to some kinds of learning than to others. It applies to learning about the actions we take. The feedback shows of our actions were successful (outcome feedback) and perhaps what we were doing wrong (process feedback)

We need feedback to learn skills and procedures and routines, not to learn facts or gain insights. Feedback isn’t sufficient for learning, it is often hard to understand in complex situations. People may not even detect the feedback cues.

Conclusion: Claim 6 (to get people to learn, give them feedback on the consequences of their actions ) holds only when the conditions are very straightforward. In complex situations, learners must interpret the feedback. This is learn to connect their actions to consequences, learn to sort out the relevant causal threads and the coincidental ones. Sensemaking is at the heart of learning cognitive skills.

We aren’t just acquiring new knowledge. We are changing the way we see things and think about them.It is known that tacit knowledge will probably be de-emphasized. It is too hard to give feedback about that. We probably focus only on the actions people take, not the reasons for their mistakes or the ways they are understanding the tasks.

Replacement

In complex situations, the hardest party of using feedback is understanding it and sorting out the tangle of causes and effects. We have to make sense of feedback in order to learn from it.

Sensemaking is seeing what led up to the events that are happening and anticipating how our actions are likely to influence future events.

For cognitive skills, we want feedback to change the way we think, not just add more facts into our memory.

The replacement for claim 6 is: ‘we can’t just give feedback; we have to find ways to make it understandable.” Providing useful feedback in complex situations can be very difficult.

Sensemaking is important even in activities that seem as straightforward as giving and getting feedback. The next chapter examines what is involved in making sense of events.

Claim 6: To get people to learn, give them feedback on the consequences of their actions.

You can’t learn without feedback. Faster feedback is better than slower feedback. More feedback is better than less feedback. In countless experiments experiments, with animals as well as with humans, feedback is essential for helping people become more skilled. Is this always true? Before we think feedback is always perfect, we should know what its limitations are,