A previous study by Purcell (1987) found that subjects exposed to training designed to increase attending to dream processes exhibited increased overall maximum dream self-reflectiveness (SR) levels, including an increased frequency of lucidity. The training group was compared to an Attention Control and a Baseline condition. The purpose of the present study was to investigate the effects of the training and the increases in dream SR on the pattern of changes in dream SR levels throughout the course of the dream report. In the present study the dreams from the Purcell study were broken down into textual information units (lUs) and each lU was assigned a separate SR score. The temporal sequence of the lUs was maintained and the resulting sequence of SR scores were analyzed in Chi squared analyses. A number of measures were investigated including the length of the dream reports, the length of the lU, the maximum SR score for all lUs, the average SR score for the dream report, the relative frequency of the SR levels, and patterns of changes in SR scores between sequential lUs. Conventional tests of significance were not used in light of the large sample size, which resulted in statistically significant chi squared values for empirically unimportant group distributional differences.

Results indicated that the training and the successful induction of lucidity had little effect on most measures of SR cores in the dream reports. What was striking in the group distributions was the similarities of the group patterns of SR in dream reports, particularly at the first five SR levels. Lucidity and SR enhancement training are concluded to have little effect on patterns of SR within dream reports. The limited effect of the training on SR is discussed in opposition to claims that lucidity may disrupt dream processes.