Sleeping and waking thought: effects of external stimulation.

This article describes some of the variables that distinguish waking and sleeping (REM) thought. Mentation reports from the waking state, as described here, tend to have more topic shifts than those from the REM state, which often have a single-theme storylike quality. It is assumed that heightened response thresholds to sensory stimuli, in conjunction with the state of high cortical activation typical of REM sleep, account for the storylike quality of REM imagery. In this experiment, an intermittent auditory stimulus was the model for environmental influences on waking mentation. It was hypothesized that the removal of this intermittent auditory stimulation, simulating in waking subjects the increased sensory thresholds of REM sleep, would decrease the number of topic shifts in spontaneous thought. It was expected that this reduction in number of topics would approach levels achieved in REM sleep. Thirty subjects participated in individual sessions in which they lay in a sound-attenuated, lightproof room with eyes closed. They were asked for mentation reports as follows: after lying awake with external stimulation (W), after lying awake without external stimulation (WO), and after being wakened from REM sleep. Transcribed mentation reports were scored on seven content rating scales, including total recall count, a count of all words in which the subject was describing his/her experience during the previous interval, and number of thought units (TU) per report, a count of the distinct, thematically homogeneous thought sequences. Hotelling t-squared tests were performed with the different states as the independent variables and the scores on the cognitive scales as the dependent variables. The major factor distinguishing mentation reports of waking subjects and subjects wakened from REM sleep was the TU count, with waking subjects changing topics more frequently. Removal of the intermittent auditory stimulus reduced the number of topic shifts in waking subjects, with a significance approaching the 95% confidence limit.