3.2 Experimental Study Description

Due to the nature of this study, a high degree of accuracy and repeatability is required.  Consequently, all subject testing is conducted in real-time by two computer systems that will interact with and monitor the subject in place of human operators (see Figure 3.2.1).  One computer is used for multi-channel data acquisition, the other computer is used to control the experiment and view the raw physiological data.  Human operators will then be free to monitor the testing equipment (computers, amplifiers, etc.) to ensure the highest quality of test results.


3.2.1 Subjects

The initial subject group consisted of 8 adults between 21 and 45 years of age (4-males and 4-females).  Subjects were screened to be sure that they have regular sleep/wake schedules and feel well rested after a "normal" night of sleep (i.e. approximately 8 hours).  The screening process is intended to eliminate any potential subjects that have pre-existing sleep disorders and/or poor health conditions.

 

Figure 3.2.2  Subject testing area.

3.2.3 Stimuli

The lighted signals that are presented to the subjects are called simple visual stimulus (SVS) and are easily recognizable when illuminated (similar to a warning light, radar blip, brake light, etc.).  The presentation of the SVS to the subject is called an SVS event.  The time between SVS events is random, so that the subject cannot predict the event times and respond in a habitual manner.  Also, the position of the light on the display board that is illuminated is generated randomly forcing the subject to scan the display board until a signal appears.  The subject responds to each SVS by pressing a button.  This response method was chosen because it is simple, unambiguous, and the action quickly becomes reflexive with little practice.


3.2.4 Test Administration

The data collection for each subject is conducted in three stages over two consecutive days.  During this testing period, all subjects are asked to refrain from using alcohol, caffeine, tobacco, and any other non-essential medication that may have an effect on their EEG waveform.

The experiments as described and human-use protocols used in this study were voluntarily submitted to the Case Western Reserve University Review Committee for Human Studies.  The committee reviewed and approved the study and found this project to be "fully acceptable" with respect to: (1) the rights and welfare of the individuals involved, (2) the appropriateness of the methods to be used to secure informed consent, and (3) the risks and potential benefits of the investigation.

The first day of testing begins with a training session to familiarize the subject with the monitoring task and to help mitigate any first-day arousing effects associated with performing tasks in an unfamiliar environment.  Data is then collected under various eyes open/closed conditions for approximately 5 minutes as a baseline measurement.

After baseline data is collected, the subject performs the test as outlined in sections 3.2.2 and 3.2.3 for approximately 60 minutes.  The subject's performance levels and EEG/EOG signals at maximum alertness are measured.  This information constitutes the alert baseline measurement and this concludes the first day of testing.

On the second day of testing, the subject is requested to wake two hours prior to their usual waking time.  Compliance is confirmed by requiring each subject to place two telephone calls to the lab at both their scheduled wake time and one hour later.  The subjects are instructed to give their name and the time they called on each message (which is verified against a hardware time-stamp on the recorder).  Subjects begin testing (after a full day of work) at their scheduled bedtime, yielding approximately 18 hours of prior wakefulness.

Once again, testing begins with various eyes open/closed conditions as on the first day.  The drowsy subject then performs the SVS test for approximately 75-100 minutes (depending upon test performance).  Immediately following the visual test, the subject is moved to a bed where they are allowed to sleep for approximately one hour.  Data is collected during sleep (lowest point of CNS arousal in the test) to provide an additional baseline measurement for each subject.


3.2.5 Experiment Execution

The entire experiment is performed under computer control.  The control computer runs the experiment by timing the SVS events and sending signals at the appropriate times to illuminate lights on the display board.  The control computer also determines if the subject has responded within a maximum allowable time-period (i.e. successful response).  Signal conditioners provide an interface between the digital input/output of the control and acquisition computers and with the display board lights and response button.  The control computer is also used to synchronize the start of all data acquisition hardware at the beginning of the experiment.

The time between SVS events was selected to be random on the basis of previous studies, which have shown that errors of omission are increased if the subject cannot control or anticipate the time of appearance of the stimulus [Johnson, 1991].  Also, there is no performance feedback to the subject during the experiment since this information can alter the arousal level and performance of the subject, and hence affect the rate of errors of omission [Johnson, 1991].  Thus, in the experiment, missed signals are generally attributable to microsleep, or attention lapses, as can be verified by EOG and video signals.

The simple and automatic response to the visual stimulus is similar to the simplest situation presented to radar operators, pilots, and drivers, where they must identify a situation (e.g. radar blip appearing on a screen, equipment malfunction warning light turns on, brake light in preceding vehicle is illuminated) and respond.  If an operator is incapable of even this extremely simple identification and response scenario, then they are certainly incapable of any actions requiring higher thought (e.g. anticipation or complex reactions).

The time duration of an SVS event will be constant throughout the experiment (e.g. 2 seconds).  The duration of this signal and the maximum allowed response time are based on safety considerations and individual response statistics.  The particular values chosen will depend on the working environment being studied since the attention lapses will have to be of some duration before any significant consequences appear.  As long as a subject cooperates in trying to detect the signal, response omissions should generally be due to a true absence, i.e., to sleep or to a sleep-like state.  Thus, omissions constitute one indication of "extreme behavioral sleepiness" [Torsvall and Akerstedt, 1988].

Most studies dealing with central nervous system (CNS) arousal agree on the fact that performance does not simply degrade continuously until some floor is reached where performance remains.  Rather, performance degrades gradually, but when a threshold is reached, the subject often starts to oscillate between alertness and sleepiness [Torsvall and Akerstedt, 1988; O’Hanlon and Beatty, 1977].  Our test has been designed to capture this oscillatory phenomenon so that it is easy to distinguish transitions in the subject's CNS arousal.

In particular, the testing in this study begins with randomly generated intervals between SVS events as stated previously.  Random intervals with a large mean time between events helps to induce fatigue and amplify the effects of any sleep deprivation.  However, if the subject fails to respond to an SVS event within a specified time period, then the time between SVS events switches from a random interval to a constant, short duration interval.  As long as the subject is in a lapse or microsleep, they are incapable of responding to the signals presented.  As soon as the subject returns to a more alert state by the successful detection of two consecutive signals, the testing system switches back to a randomly generated interval between SVS events.

The control computer continues to switch between random and constant intervals between SVS events as the subject oscillates between alert and impaired states.  This procedure can capture changes in the subject's CNS arousal in the data and this behavioral information can then be correlated with the EEG signal.

Ideally, we want to observe the subject returning to an alert state without system intervention, however, for a subject that does not respond to several consecutive SVS events, the system or test administrator intervenes to return the subject to a higher level of alertness with an artificial stimulus (audible alarm) so that the subject can continue with the testing.  If this situation occurs, the data that is collected can be used to provide an initial understanding of a subject's response to audible stimuli while in a state of extreme sleepiness.


3.2.6 Data Collection

Subject data is collected throughout the test using four channels of EEG (frontal, central, parietal, and occipital), one channel of EOG (vertical), one channel of 500,000X amplification on a 10k dummy load to measure ambient electrical noise, and visual information obtained using infrared cameras are recorded on a VCR.  All EEG and EOG data is amplified and filtered using a Grass Instruments Model 12C Neurodata Acquisition System.  This data together with SVS event and response signals are sampled and stored on an 80486DX2/66 computer equipped with a multi-channel data acquisition board.

The neurophysiological signals are displayed during the data collection period on the control computer's monitor, which is also equipped with a data acquisition board.  A videocassette recorder is used to record the camera video signals, which are synchronized and superimposed with audio calibration signals generated by the control computer to correspond with the neural data and the SVS events.

<PREVIOUS

NEXT>