Tap. Tap. Tap-tap. Tap.
You know the song.
At least, you do.
Now imagine that someone sitting across from you has to guess the song using only the rhythm you tap on the table.
For you, the melody is obvious. You can hear it in your head. You know where the pauses belong. You know what comes next.
The listener hears only the taps.
That simple difference between what one person knows and what another person receives has stayed with me.
Because I increasingly wonder whether teaching sometimes works in exactly the same way.
The song in the teacher's head
The idea comes from Elizabeth Newton's doctoral research at Stanford University. In one part of her tapping study, participants tapped familiar tunes for listeners and estimated that listeners would identify the songs about 50% of the time. In reality, listeners correctly identified only 3 of 120 tunes — 2.5% (Newton, 1990).
The tappers were not simply bad at tapping.
The more interesting problem was that they were poor at imagining what the tapping sounded like to someone who did not already know the tune.
This connects closely with what Camerer, Loewenstein, and Weber (1989) called the curse of knowledge: once we possess information, it becomes difficult to set that knowledge aside and accurately judge the perspective of someone who does not have it.
The more I teach, the more I recognise this problem in myself.
I can hear the melody
Take something relatively simple such as goniometry.
If I am measuring elbow range of motion, I know roughly where the lateral epicondyle should be.
I know what I mean when I say the lateral midline of the humerus.
I know why the acromion matters.
I know where I expect the radial styloid to be.
So, if I tell a student:
"Place the axis over the lateral epicondyle, align the stationary arm with the lateral midline of the humerus towards the acromion, and align the moving arm with the radius towards the radial styloid."
it sounds reasonably clear to me.
But what does the student actually hear?
Perhaps something closer to:
Lateral epicondyle. Humerus. Midline. Acromion. Radius. Styloid.
Tap. Tap. Tap. Tap.
The words are there.
The melody may not be.
"But I already taught them this"
I have caught myself thinking this before.
"We already covered this."
Or:
"They learnt this in anatomy."
Or, perhaps more dangerously:
"They should know this by now."
But the tapping experiment makes me question what those statements actually mean.
Perhaps the important question is not whether I taught it.
Perhaps the question is whether the learner ever built the same mental picture that I now take for granted.
A concept appearing in a lecture does not necessarily mean it has become usable knowledge.
A student may know the name of an anatomical landmark but still struggle to locate it on a person.
They may remember the steps of a procedure but not understand why those steps matter.
They may even pass an assessment without connecting those individual pieces into something they can use in practice.
Maybe I have been assuming that because I can hear the song, they can hear it too.
Expertise hides the steps
One reason this happens may be that expertise compresses thinking.
Things that once required deliberate effort eventually become automatic.
When I was first learning to identify an anatomical landmark, I might have consciously worked through several questions:
- Where am I?
- What structure should be here?
- What can I palpate?
- What movement might help me confirm it?
- What should sit proximal or distal to it?
Over time, many of those steps disappear from conscious thought.
Eventually, you put your hand down and think:
There it is.
That is useful clinically.
But it can create problems when teaching novices.
A student asks:
"How do you know that is the landmark?"
And the experienced answer can easily become:
"You can feel it."
True.
But not especially useful.
The student cannot yet feel what I feel.
There may be years of anatomy, palpation, repetition, clinical exposure and pattern recognition hidden inside that simple statement.
Research on expertise has described a similar problem. Hinds (1999), for example, found that greater expertise could make people worse at predicting how difficult a task would be for novices. In education, Nathan and Petrosino (2003) described an expert blind spot, where stronger subject-matter expertise could shape teachers' assumptions about what learners should already understand.
Teaching, then, sometimes requires us to do something quite unnatural:
unpack what expertise has compressed.
Maybe I need to explain less
When students do not understand something, my instinct is often to explain it again.
Another example.
Another diagram.
Another slide.
And sometimes that is exactly what is needed.
But sometimes I wonder whether I am simply tapping the same song slightly louder.
Tap. Tap. Tap-tap.
"Can you hear it now?"
I am becoming more interested in asking students to show me what they are hearing instead.
Why did you place the axis there?
What landmark are you using?
How did you decide this was the correct position?
What are you trying to align with?
Talk me through your reasoning.
Those questions expose something another explanation may not.
They reveal the student's mental model.
The same mistake can mean different things
Two students can make exactly the same visible error for completely different reasons.
One may misunderstand the anatomy.
Another may know the anatomy but misunderstand the procedure.
Another may understand both but struggle to translate a two-dimensional diagram into a real body.
Another may simply be uncertain.
The visible mistake can be identical.
The thinking underneath it may be completely different.
If I correct only the final action, I may never discover which problem I am actually trying to solve.
This is one reason I increasingly think that correct answers are not always enough.
Sometimes the more interesting question is:
Why did you do that?
Why simulation matters to me
This is also one reason I have become increasingly interested in simulation and interactive learning.
In a lecture, understanding can remain surprisingly invisible.
Students can nod.
They can take notes.
They can answer a multiple-choice question correctly.
And I can leave the classroom thinking:
Good. They understood it.
But place the learner in front of a practical task and the picture changes.
Put the goniometer down.
Choose the landmark.
Position the stethoscope.
Measure chest expansion.
Count respiratory rate.
Make the decision.
Now I can see what the learner actually does with the information.
This is one reason I have been developing small interactive clinical-skills activities.
The interesting part, for me, is becoming less about whether the computer can simply say:
Correct
or
Incorrect
and more about what happens next.
From "incorrect" to "why?"
Imagine a student places a virtual stethoscope over the scapula.
The simulation already knows the placement is incorrect.
It could flash red and say:
Incorrect. Try again.
But imagine instead that it asks:
"What structure are you currently positioned over, and why might that affect auscultation?"
That creates a very different learning interaction.
The mistake becomes the beginning of a conversation rather than the end of an attempt.
And perhaps, more importantly, it gives the student an opportunity to reconstruct some of the melody for themselves.
Where AI might help
This is also where my thinking about AI in education has started to shift.
Much of the discussion around generative AI still focuses on producing things:
Write this.
Summarise that.
Generate ten questions.
Explain this topic.
Those applications can be useful.
But I am increasingly more interested in AI as a responsive tutor.
Imagine a student completing an interactive goniometry activity.
They position the axis incorrectly.
The simulation itself already knows what happened. It knows the stage, the location, the size of the error, the number of attempts and perhaps what mistakes the learner made previously.
The AI does not need to decide whether the learner is correct.
Instead, its role could be to decide how much support to provide.
On the first attempt:
"Which anatomical landmark should act as the fulcrum for this measurement?"
After another unsuccessful attempt:
"Think about the bony landmark on the lateral aspect of the elbow."
And later, if the learner is still struggling:
"The axis should be positioned over the lateral epicondyle."
The support becomes progressively clearer.
The simulation determines performance.
The AI responds to that performance with scaffolding.
That, to me, is a much more interesting use of AI than simply asking it to provide the answer.
Perhaps the AI's role is to decide how much of the melody to reveal.
How much is obvious only to me?
The tapping experiment keeps bringing me back to one uncomfortable question:
How much of what I consider obvious in my teaching is only obvious because I already know it?
When I look at a diagram, I know what I am supposed to notice.
When I read a clinical case, certain details immediately stand out.
When I watch movement, I already know which relationships matter.
Students do not necessarily have that filter yet.
And there may be moments when I mistake their inability to hear the song for a lack of effort.
Sometimes that may be true.
But sometimes I may simply have forgotten what the tapping sounds like without the melody.
That is something I want to become more conscious of in my own teaching.
Becoming the listener again
I do not think we can completely escape the curse of knowledge.
Nor should expertise itself be treated as a problem.
Our knowledge and experience are precisely what allow us to teach.
But perhaps good teaching requires us to occasionally move to the other side of the table.
To become the listener again.
To ask:
Would this explanation still make sense if I did not already know the answer?
To watch students attempt something before correcting them.
To ask why.
To let them explain.
To create activities in which their decisions become visible.
And to pay attention to where they hesitate.
Because sometimes that hesitation may tell us more about learning than another correct examination answer ever could.
The teacher taps confidently.
The song is perfectly clear.
The student looks confused.
And the teacher thinks:
How can you not hear this?
Maybe the better question is:
What melody am I hearing in my head that I have not yet helped the student to hear?
References
Camerer, C., Loewenstein, G., & Weber, M. (1989). The curse of knowledge in economic settings: An experimental analysis. Journal of Political Economy, 97(5), 1232–1254. https://doi.org/10.1086/261651
Hinds, P. J. (1999). The curse of expertise: The effects of expertise and debiasing methods on prediction of novice performance. Journal of Experimental Psychology: Applied, 5(2), 205–221. https://doi.org/10.1037/1076-898X.5.2.205
Nathan, M. J., & Petrosino, A. (2003). Expert blind spot among preservice teachers. American Educational Research Journal, 40(4), 905–928. https://doi.org/10.3102/00028312040004905
Newton, E. L. (1990). The rocky road from actions to intentions [Doctoral dissertation, Stanford University]. Stanford University.