Introduction

In prolonged work tasks subjects are required to engage in simple, repetitive activities, such as letter cancelation, detecting differences in simple shapes, adding three digits, and so on. Performance is recorded as a series of response (or reaction) times. Subjects are instructed to work as quickly and as accurately as possible. The items should be answered in a self-paced continuous manner, in which the subject cannot afford to take intermediate rest pauses between responses. In this way one is able to study "... the fluctuations which always occur in any person's continuous output of mental work, even when this is so devised as to remain of approximately constant difficulty." (Spearman, 1927, p.320). Prolonged work tasks were especially used to measure concentration ability, which was already the case in the very beginning of this century (see e.a. Binet, 1900).

Pieters and Van der Ven (1982) introduced the explanatory concept of distraction to account for the response time fluctuations occurring in the test. They assumed that the manifest response time should be considered as being composed of a relatively constant (over items or groupings of items) real total work time, interrupted by a series of random distraction times. This led to the formulation of the so-called Poisson-Erlang model which is based on the following three assumptions: (i) each separate distraction time has an exponential distribution; (ii) the number of distractions has a Poisson distribution; and (iii) the real total working time is constant over responses. The Poisson-Erlang model is able to account for the short-term variation in the response times, but not for any long-term trend effect, although, in actual time series, this effect usually does occur. In many cases in the beginning of the test, there is an increase in the reaction times, which in the long run, starts fluctuating at some stationary level. In the past few years several alternative models have been developed in order to account for the long-term trend and the possible interdependency of the response times. These models are all based on the assumption that distractions are periods of recovery from accumulated inhibition. It is generally assumed that during work intervals inhibition increases and during distraction intervals inhibition decreases. The most recent model proposed thusfar is known as the Poisson-Inhibition model (Smit and Van der Ven, 1995).

All inhibition models, including the PE model, only apply to tasks which are administered in a continuous, homogeneous and over-learned form and which do not involve any type of random search process. The term "continuous" refers to the fact that the person controls his own speed: the subject's response to each part (task unit) releases the next one in the sequence. No intermediate rest pauses are allowed. The consecutive reaction times may, therefore, be dependent. The term "homogeneous" refers to the property of low and equal difficulty. Equal difficulty is required to make sure, that, the purely technical assumption, that the real total working time Ais constant across task units, is warranted. If the task is "hereogeneous" in the sense that individual task units (reaction times correspond to task units) could have unequal difficulties, then A would be dependent on the task unit and different task units could have different A's. Low difficulty is desirable to make it easier for the subject to get used to the task. The fact is that the task should be "over-learned" in order to prevent systematic changes in A. If the subject is performing the task for the first time and is not yet familiar with it, then it may well be the case that the real-processing time A in the beginning of the task will be larger and also more variable. This should be avoided in order to warrant constancy of A. If the task is not over-learned in advance, then sufficient practice trials must be given in order to guarantee the constancy of A. One should also avoid tasks in which any type of random search process is involved. A typical example is the sub-test Form Matching which belongs to the General Aptitude Test Battery, abbreviated GATB (U.S. Department of Labor, 1962). This test consists of two groups of simple two dimensional figures. The figures in the second group are the same as those in the first group but arranged in a different order. The subject must find each figure in the second group in the order in which they are displayed in the first group. The random configuration of the figures leads to random search times. Since the search time is a part of the processing time, A too would be random.