Clinicians report that patients with traumatic brain injury (TBI) ofte
n have difficulty with tasks requiring sustained attention, and there
are neuroanatomical and neurophysiological reasons to expect such defi
cits. Nevertheless, laboratory measures of sustained attention or vigi
lance in TBI have produced conflicting results. These inconsistencies
may be due to patient heterogeneity as well as the fact that vigilance
performance is dependent on highly specific features of the task desi
gn. We developed a visual vigilance task in which the influence of non
-attentional factors was minimized and task difficulty for patients an
d controls made comparable. Performance was characterized with respect
to vigilance level as well as vigilance decrement, using measures of
perceptual discrimination, response bias, reaction time and reaction t
ime variability. Twenty-six patients with recent TBI and 18 control su
bjects were tested on this task. A MANOVA of ranked scores revealed si
gnificantly different patient and control performance overall. Initial
level of performance (vigilance level) was slower and more variable f
or patients than controls, and patients showed more conservative respo
nse biases. Deterioration over time (vigilence decrement) was also ste
eper for patients than controls for reaction time, reaction time varia
bility, and response bias. Deterioration in accuracy (D') did not diff
er significantly between patients and controls. Performance was not re
lated to available measures of injury severity. Hypotheses relating ar
ousal mechanisms to vigilance performance are discussed.