19 healthy non-shiftworking right-handed participants (11 males) aged between 18 and 39 years (M ± SD: 24.6 ± 5.3 years) were recruited from the general population. Initial eligibility screening involved gathering information regarding age, self-reported height/weight, smoking status, health status, and sleep patterns via telephone screening. Exclusion criteria were the following: presence of medical conditions, psychiatric disorders, sleep disorders, abnormal blood chemistries, habitual sleep duration < 7 h or > 9 h, BMI outside the normal to overweight range (18.5–27 kg/m2), regular medication use (oral contraception was allowed), drug and/or alcohol abuse, methamphetamine abuse, more than 2 h of structured high impact activity/exercise per week (to limit the extreme metabolic changes influencing results related to being sedentary in the lab for 7 days), food allergies, pregnancy, and any trans meridian travel in the 60 days prior to the study, or history of shiftwork, to ensure stable sleep/wake patterns and no circadian misalignment. If deemed eligible, following written informed consent to participate in the study, participants attended two further screening sessions in person to complete questionnaires regarding psychological and physical health, had a blood test completed, and wore an activity monitor for 7 consecutive days prior to the study commencing (Phillips Respironics Actiwatch, Murrysville, PA, USA). Participants also completed a 7-day sleep diary and a 4-day food diary to assess habitual sleeping and eating patterns, respectively. Female participants were only scheduled to participate in the luteal phase of their menstrual cycle, as confirmed by self-report, to control for changes in sleep quality, hormonal factors, and basal metabolism during the menstrual cycle29. Participants refrained from over-the-counter medications, caffeine, alcohol, napping, and were instructed to keep a strict habitual sleep schedule (22:00–23:00 to 06:00–07:00), which was verified by actigraphy and a subjective sleep diary during the 7 days prior to study commencement. Urine toxicology was performed to verify the absence of any illicit substances immediately prior to study commencement. These criteria are consistent with the laboratory’s previous sleep studies standard exclusion criteria30, 31. Handedness was assessed using the Flinders Handedness Survey (FLANDERS)32 and was not used as inclusion or exclusion criteria. Data collection for the present study took place within a wider study, which received approval by the University of South Australia Human Research Ethics Committee (#0000033621) and is registered with the Australian New Zealand Clinical Trials Registry (ANZCTR12615001107516). All procedures were in accordance with the National Statement on Ethical Conduct in Human Research.
The study was conducted at the Sleep and Chronobiology Laboratory at the University of South Australia. Participants spent six nights and seven days consecutively in a laboratory-controlled environment. The laboratory temperature was set to 22 ± 1 °C. Light intensity was < 100 lx at eye-level during wakefulness and was fixed at < 0.03 lx during sleep periods. On the first night, participants had an 8-h sleep opportunity, which followed a baseline day. Prior to the first night shift, participants were awake for 28 h to simulate the common pattern observed in shiftworkers prior to their first shift4. From days three to six, during the simulated night shifts, participants were given a 7-h daytime sleep opportunity on each day between 1000 to 1700 h. The last day (day 7) involved recovery sleep, where participants were given an 8-h sleep opportunity between 2200 to 0600 h to adjust their sleep/wake cycle from the shiftwork manipulation. Further information on the protocol can be found from a previously published study33. Data for the current study was collected at six different time points across six days. Refer to Fig. 1 for an outline of the testing protocol.
Two measures of alertness were employed in the current study, one assessing objective (behavioural) alertness, and the other assessing subjective (self-reported) alertness.
Psychomotor vigilance task (PVT)
A 10-min hand-held PVT device, which is highly sensitive to sleep loss was used to measure alertness in the current study. The PVT is regarded as a reliable and valid assay of alertness34, 35. The PVT required participants to respond as quickly as possible to the presentation of a visual stimulus (i.e., a red circle) which changes to a number as a timer counts from 0 to 500 ms. PVT performance at six time points were obtained in the current study: days two and seven at 1400 h, days four and five at 0300 h, and days four and five at 0500 h were examined. The PVT reciprocal of the mean reaction time (mean RRT) and lapses were used for analysis. Mean RRT was calculated as the mean of the reciprocal of the reaction time (1/RT) in milliseconds. A lapse was counted when reaction time was greater than 500 ms36. Lower mean RRT scores and higher lapses indicate poorer performance and lowered alertness.
Stanford sleepiness scale (SSS)
A 7-point rating scale was used to assess subjective sleepiness, ranging from 1 “feeling active, vital, alert, or wide awake” to 7 “no longer fighting sleep, sleep onset soon; having dream-like thoughts”37. The SSS has been found to predict performance on tasks that assess alertness, such as reaction time and vigilance tests38, 39. The SSS was used to assess subjective sleepiness throughout the protocol at the same six time points as the PVT (i.e., days two and seven at 1400 h, days four and five at 0300 h, and on days four and five at 0500 h). Higher SSS scores indicate higher levels of sleepiness.
Spatial attention measures
Two measures assessing spatial attention, more specifically, one assessing judgement and the other assessing detection, were used in the present study. The judgment task entailed judgment of the spatial properties of lines, whilst the detection task required speeded responses to presented stimuli. Stimulus presentation for both tasks were presented using E-prime 2.0 software on a DELL desktop computer with a screen resolution of 1920 X 1080 and a diagonal screen size of 58 cm. Participants were seated approximately 60 cm from the centre of the screen, with their heads rested on a chinrest to maintain participants’ head position towards the centre of the screen.
A commonly used spatial attention task, the Landmark task18, 20, 40 was employed in the current study. One hundred and eight vertically pre-bisected horizontal lines were presented against a grey background. Each pre-bisected horizontal line contained diagonally opposite pairs of black and white segments on each side of the vertical divider. The top segment on one side was white whilst the bottom segment of the same side was black, and vice versa for the opposite side. Side of longer segment (left/right), polarity (upper left black, upper left white) and deviation from the midline (0.5, 1, or 2 mm) differed between trials. Half the total number of stimuli were pre-bisected towards the left, and half the stimuli contained upper left black polarity. The factors side, polarity and deviation were counterbalanced between the 108 stimuli, such that each factorial combination was presented on nine occasions. Participants were instructed to indicate the side of the pre-bisected line that was longer, responding using the letter ‘z’ and ‘m’ on the keyboard to indicate left and right, respectively. Each trial started with a blank screen for 1000 ms followed by a stimulus presentation for 500 ms and a blank response window for a maximum of 2000 ms (see Fig. 2). If the participants responded within 2000 ms, the following trial was presented. However, if participants did not respond within the allocated response time window, the trial was presented again at a later stage, so each participant responded to 108 stimuli before task termination. The task lasted approximately eight minutes. Bisection bias on the Landmark Task was calculated as the number of right responses minus the number of left responses divided by the total number of trials for each participant41. Hence, negative and positive bias scores indicate a bias towards the left and right side, respectively. Landmark Task spatial bias on days two and seven at 1400 h, and days four and five at 0300 h were analysed.
Participants were presented with 5 mm circular dark grey stimuli at varying eccentricities on a light grey screen. Each circular stimulus was presented randomly at one of four eccentricities on each side of the fixation point at 2, 7, 12, and 17 cm from the centre of the screen with a height of 28 cm and width of 51 cm. A total of 120 stimuli were presented, with 15 stimuli at each of the eight screen positions. A fixation cross was presented throughout the trials. Each trial started with the presentation of the central fixation cross for a randomized presentation time ranging between 1500 to 4500 ms, followed by the stimulus presentation for 100 ms. The response window was set to a maximum of 1500 ms (see Fig. 3). The trial terminated once the participant responded or if no response was made after 1500 ms. If participants did not respond to a stimulus, that trial was considered an omission of response. If participants responded within 100 ms of stimulus presentation, this response was deemed an anticipatory response and removed from analysis. The task lasted approximately 12 min. Detection Task performance was assessed on days two and seven at 1400 h, and days four and five at 0500 h. The number of omissions (i.e., the trials that participants did not respond to stimuli) and the reaction time following stimulus presentation were used for analysis.
Data was analysed using jamovi version 1.0.7.042. Alertness measures (PVT & SSS) and spatial bias measures (bisection bias on the Landmark Task & reaction time and omissions on the Detection Task) corresponding to the time of testing (i.e., D2 1400 h, D4 0300 h, D4 0500 h, D5 0300 h, D5 0500 h, D7 1400 h) were used in analysis. Missing data was not imputed or corrected for in the analyses as mixed effects models allow for parameters to be estimated using available data43.
Alertness changes with time of testing
Three mixed effects models with ID as a random intercept were run to determine whether alertness changed across the different time points of the simulated shiftwork protocol. All three models included the categorical variable of time (D2 1400 h, D4 0300 h, D4 0500 h, D5 0300 h, D5 0500 h, D7 1400 h) as a fixed effect. The continuous variables PVT lapses, PVT mean reciprocal reaction time (mean RRT), or SSS were entered as the outcome variable for each model. F-values from fixed effects, degrees of freedom, and post hoc analyses are reported in Table 1. Holm corrections were used to account for multiple post hoc comparisons. Partial eta2 effect sizes were also calculated independently from the mixed effects modelling.
Spatial attention changes with time of testing
Three mixed effects models with ID as a random intercept were run to investigate whether the Landmark Task bisection bias, and Detection Task reaction time and omissions related to time. All models had time as a fixed effect, and the continuous variables bisection bias, reaction time or omissions as the outcome variable for each model. Side and location were further predictors in the Detection Task models. F values, degrees of freedom, and post hocs comparisons with Holm corrections are reported in Table 1.
The alertness and spatial attention relationship
A series of mixed effect models were conducted to investigate the relationship between alertness and spatial attention during the simulated shiftwork protocol. For the Landmark Task, two linear mixed effects models with fixed effects of objective alertness, as measured by PVT lapses and PVT mean RRT, were conducted. PVT lapses and PVT mean RRT were predictors in separate models. Bisection bias was the outcome variable in these models. Another mixed effects model, with subjective alertness, as measured by the SSS, with the same time points as the previous model was conducted, and bisection bias as the outcome variable was run. ID was entered as a random intercept in all models.
For the Detection Task, six linear mixed effects models with fixed effects of objective alertness (i.e., PVT lapses, PVT mean RRT), or subjective alertness (i.e., SSS), were conducted to determine the relationship between alertness and spatial bias. The outcome variable was reaction time for three models and omissions for the remaining three models. Time, side of stimulus presentation (left/right), and location of stimulus presentation (2, 7, 12 and 17 cm) was entered as a fixed effect and ID was entered as a random effect in all models. Post hoc comparisons with Holm corrections were also conducted.