the anecdotal moment of epiphany happened a couple of years ago and I forgot to write it down but the latest videos on object behaviors reminded me of that one.
During a tennis training the coach showed a simple trick to have better chances at returning a high trajectory ball (flying very high above the net): point your free hand’s finger at the ball while it’s approaching from above.
Like “magic”, the trick worked. I feel, this anecdote can have value for thinking about reference frames and prediction, and maybe connecting various scales (while the ball is descending, the view angle doesn’t change much from the viewpoint of the returning player.
Considering, the prediction of when/where to hit the ball with the racket was improved, the virtual line between the eyes and the ball, passing through the fingertip must ground the reference frames much better than predicting the trajectory simply from the view-angle changes. The visual changes are small but the positioning of the hands is relatively precise.
There are many reference frames involved, but the trick seems to bring them nicely together.
There’s probably lots going on (many behaviors involved?). What seems explainable that the ball moving along the virtual line is an easy behavior, and the “child object” – the ball is moving along it.
I’m sorry for not being able to try to describe this more formally / theory-related, but does that sound interesting enough?
Hi @DLed, sounds interesting to me! If I may try to rephrase what you are saying:
Learning the parabolic trajectory of ball in 3D space is “complex” behavior and therefore hard to predict the (x, y, z) location.
If the reference frame is not 3D space but instead the trajectory itself (let’s say the upside down parabola or arc), then we only need to predict where we are in the parabola or arc, essentially reducing down to predicting some 1-dimensional “x” location along the trajectory.
I think it is definitely a more useful or compact representation of object behavior. There may be other implications too - we are still in the midst of hashing out Object Behaviors so I’m not sure yet where it may fit in But thanks for sharing the idea! It’ll be in my head as we discuss more Object Behaviors.
Hi @hlee , glad you found it interesting! After reflecting on your formulation, another aspect comes to mind: apart from the little input that is available to recognize the complex trajectory, the acceleration is another aspect of that ball behavior. Perhaps, to add to the example’s complexity: the trajectories can be different due e.g. spin: The Physics of Tennis: Tennis Ball Trajectories — The Role of Aerodynamic Drag and Lift in Tennis Shots"
but in the worst case, from the standpoint of the receiving player, the ball only goes up and then down for both eyes - without much input as to the position on the parabola. The player won’t see the nice parabola from the side.
Another idea: the eyes have a vertical axis but could it be that its resolution is worse than the stereoscopical - horizontal axis? The finger would add a better resolution for the vertical movement due to anchoring it to the body position (the arm has fixed hinges) - but the eyes might have more complex movement (saccades, head movement) relative to the ball, adding to confusion.
My (amateur) guess would be that having a fixed reference point in the visual field (finger point at arms length being a well trained reference point) may assist in cancelling optical acceleration as well as priming serve-adjacent mechanics.
There may be other implications too - we are still in the midst of hashing out Object Behaviors so I’m not sure yet where it may fit in 
But thanks for sharing the idea! It’ll be in my head as we discuss more Object Behaviors.