Illustrating the cognitive consequences of object-oriented systems development

Abstract

The purported “naturalness” of object-oriented (OO) approaches refers to the assertion that an alignment exists among the innate cognitive activities of system designers and the characteristics of OO system development approaches. That is, designers can think about the objects that make up the system in the same way they think about objects in the world. This fit of cognitive activities and OO characteristics appeals to many system developers. Contrary to the claim of naturalness, many students and practitioners report difficulties learning and using OO techniques. To understand why OO approaches seem to be difficult to learn and use, we illustrate the consistency of common characteristics of OO techniques and the cognitive activities of systems designers. A group of graduate students identified 148 beliefs, facts, and ideas pertaining to OO complexity. Researchers classified each item in terms of the abstraction, communication, problem-orientation, and shared behavior characteristics common to OO approaches. Then classified each item in terms of the problem understanding, problem decomposition, and design solution cognitive activities. The consistency of the two classifications confirms that claims are supported by usage. However, it seems that methodology and project management are part of the task of developing information systems (ISs) that are not adequately addressed by OO approaches.

Introduction

Claims about the naturalness and usefulness of object-oriented (OO) techniques are driving their adoption by information systems (ISs) organizations (Booch, 1994; Coad and Yourdon, 1991a, Coad Yourdon, 1991b). Researchers suggest that the abstraction, communication, problem-orientation, and shared behavior characteristics of OO methods can facilitate the creation of a computational representation that matches closely the problem domain (Rosson and Alpert, 1990; Rosson and Gold, 1989). Naturalness refers to the consistency of these common OO characteristics with designers' innate problem solving activities.

Research on the cognitive activities of experienced software designers shows that these activities include problem understanding, problem decomposition, and design solution (Guindon, 1990; Rosson et al., 1988). Based on the identified activities, a rationale for the cognitive consequences of OO design has been proposed (Rosson and Alpert, 1990). The common characteristics of OO methods (abstraction, communication, problem-orientation, and shared behavior) were suggested to complement the understanding, decomposing, and solving cognitive activities of software designers.

Despite claims of naturalness, both students and experienced developers report difficulties in learning and using OO techniques. Adopters are cautioned that the scope of organizational change required by OO, the cognitive load of OO environments, and complexity of the techniques are factors contributing to these difficulties (Fichman and Kemerer, 1993; Rosson and Carroll, 1990; Vessey and Conger, 1994). Similarly, system users also report difficulties understanding OO analyses (Moynihan, 1996).

It seems contradictory that a “natural” approach is difficult to learn. Presumably, designers already have the innate cognitive activities and know how to use them, so learning OO approaches should be a simple matter of recognizing the need for and applying familiar patterns. So, where does the problem of learning originate, in the claims of naturalness or in the usage of OO techniques. This leads to the following question:

Why are so-called natural OO approaches difficult to learn?

Understanding what is difficult about learning OO techniques requires investigating the perceptions of designers that recently learned OO techniques. Evaluating these perceptions in terms of the OO characteristics and cognitive activities allows observation of the consistency of OO characteristics and cognitive activities.

It is this observation that is of interest when defining the “naturalness” of OO systems. The expression of OO characteristics in terms of cognitive activities reveals the match (or mismatch) that may indicate reasons for difficulties in learning and/or imply techniques for controlling these difficulties.