Abstract
According to Borbély's (1982) Two-Process Model of Sleep-Wake regulation sleepiness/alertness is regulated by the joint action of two mutually independent processes (Achermann & Borbély, 1994). However, the assumption of the circadian and the homeostatic process operating independently has been contested in different studies (Dijk, Duffy & Czeisler, 1992, Jewett & Kronauer, 1999). Usually sleep researchers manipulate processes S and C in such way that both variables become fully separable, where after a mathematical function describing their joint action is derived from neurobehavioral measurements. If a nonlinear relation between both processes is observed in the data, then it is assumed to be a true manifestation of their interaction on a substantial level. However, unless the quantitative structure of the latent variable is known, there are no direct means to verify if the measurements are indeed a true reflection of the latent variable in question. In fact, the observed nonlinearity in the data could simply be the result of ceiling and floor effects in the metrics used. This concern was raised by Achermann (1999) and so far it remained largely unsolved. The research carried out in this dissertation aimed at resolving this issue.
Unfortunately, the issue of linearity is not limited to controlling for ceiling and floor effects. A measurement instrument yielding interval data (e.g. vigilance tasks based on reaction times, see Chapter I, section 1.3.3.) may still, erroneously, represent the latent variable and consequently impede on the interpretation of the results. Therefore, the question we have to answer is whether or not neurobehavioral measurements of sleepiness/alertness are linearly related to the latent variable they are designed to measure.
In order to achieve this, we adopted a strong methodological and theoretical framework which roots are deeply embedded in psychophysics and measurement theory. By means of Functional Measurement (FM: Anderson, 1981, 1982) we investigated how homeostatic and circadian components combine when judgments of sleepiness are made based on symbolic information on prior sleep and time of day. Within sleep research this approach may appear unlikely. However, as FM allows for the validation of the response function, it consequently offers means for inferring on the true relation between processes S and C , which is incidentally why this line of research is pursued in the first place.
Throughout this dissertation, we opted for a naive-pragmatic and Cartesian approach. Put differently, we decomposed the problem into smaller parts and started our research right at the source: the metric itself. First, we selected a measurement instrument with adequate psychometric qualities (Chapter II) and designed a tool to conduct FM experiments (Chapter III). Then the validity of the method was tested and the linearity of the response instrument was established in the context of the Two-Process Model (Chapter IV and V). Eventually, we assessed the linearity of the instruments in a slightly different context to avoid circularity in our reasoning (Chapter VI). Furthermore, by replicating our experiments, we insured the reliability of the judgment task (Chapter IV and V). Finally, we conducted an experiment where the orthogonality of our original designs was improved in order to infer about the true interaction of homeostatic and circadian components in sleepiness judgments (Chapter VI). In what follows we will summarize some of our findings in order to draw conclusions that transcend the scope of the separate papers.
Original language | English |
---|---|
Awarding Institution |
|
Supervisors/Advisors |
|
Place of Publication | Brussels |
Publication status | Published - 2007 |
Keywords
- functional measurement