NeuroCOLT

Neural Networks and Computational Learning Theory

About NeuroCOLT

Papers Archive

Books

info@neurocolt.org

NeuroCOLT Technical Report NC-TR-02-135

2002-135
Why Students don't ask Questions
Peter Auer

ABSTRACT

We analyse the number of questions which a rational student will ask in a learning situation, depending on his or her objective. For this we propose a formal learning model and two possible objectives of the student. In this learning model material is presented to the student in small bits, the student may ask arbitrary yes-no questions about the material, and occasionally the teacher will query the student to access his understanding of the material.

In the first scenario that we will consider, questions asked by the
student are regarded as faults. The objective of the student is to
minimize the number of questions that he asks plus the number of incorrect answers that he gives to the teacher's queries. The result of our analysis is that in such a scenario a student will ask rather few questions.

In the second scenario the student's objective is to minimize the
number of incorrectly answered queries and to be efficient with her learning time. We will argue that there are cases in which a student who asks no questions will need much more time to learn than a student who asks questions.

Download Postscript