Anticipation, Windshield Wiper, Part 2

Follow-up to Anticipation, Windshield Wiper, Part 2


6. Is an exit point for the circular process available?

This question should clarify whether the received information justifies proceeding from the internal model to the response selection. A response cannot be selected without enough information. This is a real problem in this example. Continuing in one direction or switching to the exact opposite direction are appropriate options. The missing patterns and causality prohibit proceeding to the response selection step. This way, you get caught in this step when your partner changes the direction. The directional change is the first moment when the necessary information appears, but this is way too late.


Exit point for circular process



7. Is there a stop criterion for the response selection?

This question is needed when more than one movement is possible to solve a given situation. It is not a problem to have more than one solution. Not knowing how to choose is a problem. It is neither wrong nor right to practice more than just one solution to a specific problem. But, you have to have clear criteria about how to make a decision. This is very important for automating movements. There are indications that this decision-making hinders the learning process of automated movements (et al. Schack). Our example fulfills this requirement, since there is only one solution for each direction.


8. Is the internal model valid during response execution?

The internal model develops scenarios for the next few seconds. Especially the projected future events are the basis for the response selection, the start of the response and the response execution. If these scenarios don’t fit, the executed movement impinges on other possible alternatives (see Scissor, Stone, Paper). This question targets this specific aspect.

This is not fulfilled in our example, since the internal model cannot model the upcoming events. If you follow your partner from right to left and the directional change happens, your internal model is still in the “going from right-to-left” mode.


Example Scissor, Stone, Paper

9. Is the time for response execution sufficient?

Every movement takes time to be executed. Thus, looking at the execution time is the last aspect on this list. If the execution time for the chosen movement is too long, the ongoing events render the response useless. The response execution time in our example is relatively short and needs no further examination. The other aspects are more pressing.


As you can see, you can go through this list and find the problematic parts. This analysis allows you to work out several very important aspects. Especially the late availability of any information concerning the upcoming directional change prevents successful anticipation. The crucial point of this model and its´ checklist is to clearly identify the real problems so that countermeasures can be taken. You can practice movements for years without signs of improvements if problems are not identified and addressed in a helpful manner. It is standard practice to blame insufficient or ineffective training. For example, standard training improves pattern recognition. It doesn’t solve the inherent problems, which emerge as a result of the process structure itself.

So, you fundamentally know why you fail. I’m quite sure that you are eager for the solutions to these problems. You either find something yourself or you apply work techniques. But these techniques can be complicated. Thus, please be content with the current level. I have to address some other things first… like a lot more examples!

The post was published 25. July 2014 related to the category Miscellaneous and tagged with .