Describing and explaining the adsorption behaviour of oxides

Abstract

The title of this chapter includes both “describing” and “explaining”. The distinction is not always clear, yet it can be important. It is possible to describe a set of observations in a completely non-explanatory manner. An example is the use of a quadratic function to describe a simple curve. In most cases such descriptions are non-mechanistic. They can be of value for summarizing a large set of observations by a few numbers. They are also of value for interpolation but of limited value for extrapolation. However, if we want to understand the processes involved, and thus explain the observations, we need a different approach. If this understanding is formalized into sets of relations, and these relations are precisely expressed as equations, we may call it a mathematical model. Thus a mathematical model is no more than a precisely expressed hypothesis. The confusion arises when we wish to test that hypothesis. This involves testing whether it describes the observed behaviour. Thus in both cases the test is the same. What then are the characteristics of an explanatory or mechanistic model? One is that it contains elements of an underlying mechanism. However mechanisms seem to be like Russian dolls — there is always another one inside. What seems mechanistic to one may seem superficial and descriptive to another. Thus judging on this criteria is partly subjective. Another characteristic of a mechanistic description is that it is comprehensive. There are several ways of describing an elephant’s trunk — but a comprehensive model is also capable of describing the rest of the elephant. The criteria of comprehensiveness is part of the normal scientific process of testing hypotheses/ models; if a model is comprehensive, we should be able to make predictions and to test them.