gies when specific adjunct information was available, and that more able readers were more likely to find the specific difficulty information distracting, confirms a result reported by Brown ( 1980). In her experiments, externally imposed metacognitive strategies were not effective in improving task performance relative to strategies spontaneously adopted by individuals. The results of the present experiment suggest that more able readers are aware of the effectiveness of their natural reading strategies and therefore are less likely to abandon their natural strategies, even in the face of circumstances that encourage them to do so.
Why providing adjunct importance information had no effect on task performance is difficult to determine. An explanation that is suggested by Anderson ( 1980) is that, with approximately half the text highlighted and with at least some questions tapping information from nonhighlighted areas, the highlighting may not have been very informative about what to read. Perhaps the only conclusion to draw from the adjunct importance information condition in Experiment 2 is that the effects of adjunct information appear to depend heavily on the nature of the adjunct information.
The results of both experiments suggest that executive control skills such as effective allocation of reading time and efficient planning of task strategy are related to comprehension performance, even for mature skilled readers. Our immediate goal for future research is to specify the nature of this relation, a goal we are beginning to pursue by attempting the integration of an executive control mechanism and a model of performance components of reading. Other goals for future research include determining which executive processes are specific to reading and which are not, and determining more precisely sources of individual differences in executive performance. For example, in the second experiment, more able readers revised their task strategy during ongoing performance more than did less able readers. The actual source of this difference could be (a) more accurate performance monitoring, (b) availability of a wider range of well-practiced reading strategies, (c) differential ability to implement alternative strategies, or (d) some combination of the aforementioned.
The development of cognitive models of reading that better represent the intricacies of the reading of skilled readers in naturalistic situations depends on our willingness and ability to incorporate -- into our already well-specified models of performance components of reading -- equally well-specified executive controlling mechanisms.
The work described in this chapter was supported by Contract N0001483K0013 from the Office of Naval Research and the Army Re