Trading Accuracy for Speed in Selective Attention Tasks during Sleep Deprivation

Objective Selective attention is a critical cognitive function in tasks that require careful monitoring of visual information. An increasing amount of night work on jobs that meet such requirements motivates the study of the effects of sleep deprivation (SD) on selective attention. Monitoring tasks in occupational settings can be operationalized for study in the laboratory by visual search tasks, in which the subject has to identify rare target items among distractor items. Visual search under SD leads to a decrease of overall response speed and accuracy. However, an increase of response speed per item with increasing number of items was also found under SD. This might indicate an unfavorable speed-accuracy trade-off: decision speed per item increases, while accuracy decreases in turn. Eye gaze tracking may provide additional indications for the duration of individual decisions during the search process. Methods We recorded gaze behavior during two standard visual search tasks to explore if SD leads to faster and sloppier decisions per item (shorter fixation durations). Visual search performance (response time and sensitivity d’) in the conjunction (CST) and spatial configuration (SST) search task and the associated gaze behavior were tested in two groups of subjects. While sleep deprived participants (N=24, age: 25 ± 5 SD; 12 women) were kept awake during a 24 hour period, participants of a control group (N=24, age: 26 ± 4 SD; 10 women) were allowed to sleep (TIB: 8 hours). Results We found an increase of response time under SD compared to the control group (CST: p = 0.019, MSD= 128 ms ± 220 SD , MC= 1 ms ± 169 SD; SST: p < 0.001, MSD= 65 ms ± 755 SD , MC=- 45 ms ± 604 SD) and a decrease in sensitivity d’ (CST: p = 0.03, MSD= -0.14 ± 0.65 SD , MC= 0.01 ± 0.58 SD; SST: p < 0.001, MSD=-0.43 ± 0.82 SD , MC= 0.10 ± 0.64 SD) . For gaze behavior, we found indications of a cognitive slowing in slower saccade velocities for the SD group compared to the control group (CST: p < 0.001, MSD= -38.8 °/s ± 42.4 SD , MC= -5.1 °/s ± 33.8 SD; SST: p < 0.001, MSD= -29.0 °/s ± 32.4 SD , MC= -2.8 °/s ± 23.7 SD) but not in terms of longer fixation durations. On the contrary, fixation durations decreased in the SD group relative to the control group (CST: p = 0.018, MSD= -16.1 ms ± 30.3 SD , MC= -6.2 ms ± 25.1 SD; SST: p = 0.008, MSD= -11.0 ms ± 23.4 SD , MC= -2.2 ms ± 19.7 SD). Moreover, the change in fixation duration was positively correlated with the change in task accuracy (CST: ρ = 0.29, p = 0.048; SST: ρ = 0.33, p = 0.022), suggesting that subjects that showed the strongest decrease in fixation duration exhibited also the strongest decrease in search accuracy. Further analysis showed that the general decrease of fixation duration was mainly associated with a steep increase of the frequency of very brief fixation durations, so called ‘express fixations’ with durations shorter than 120 ms. Conclusion The decrease in fixation duration under SD is further evidence for a shifted speed-accuracy trade-off as indicated by earlier search studies. SD seemingly leads to an increase of decision speed at the cost of reduced accuracy. Consequently, SD not only results in overall cognitive slowing but also in an impairment of adequate resource allocation in selective attention, potentially due to the absence of prefrontal inhibitory control