INTRODUCTION
A large number of adults (approximately 30–50%) infected with
the Human Immunodeficiency Virus (HIV) experience some degree of
cognitive compromise (McArthur & Grant,
1998). The general neuropsychological profile of HIV is
relatively well mapped-out (for reviews see Hinkin
et al., 1998; McArthur & Grant,
1998), with the most frequent symptoms being grouped under the
rubric of attention (Heaton et al., 1995). As
we have recently reviewed (Hardy & Hinkin,
2002), in addition to problems with concentration, HIV+ adults
show deficits in a variety of specific processing aspects of attention
such as set-switching ability (van Gorp et al.,
1989), divided attention (Hinkin et al.,
2000), spatial orienting (Maruff et al.,
1995), inhibition (Martin et al.,
1992), and preparatory processing (Law et
al., 1995). One important aspect of attention that has not yet
been examined in HIV+ adults is perceptual span. Perceptual span tasks,
which involve the discrimination of a target letter among briefly
presented arrays of varying number of nontargets, were originally used
to delineate potential capacity limitations in early visual processing
(Estes & Taylor, 1964). This capacity
limitation was termed the span of apprehension.
Span of apprehension is a useful concept in the neurocognitive study
of HIV because it can delineate a processing deficit (i.e., central
nervous system dysfunction) at a very early point in cognition. It is
early visual processing (e.g., iconic memory capacity) that is
considered with a perceptual span task because of the extreme time
constraints that are imposed on stimulus processing. The unusually
brief presentation of stimuli (relative to most cognitive tasks)
precludes “later” processes that are often involved with
standard test performance (e.g., visual search, full rehearsal of
stimuli—in larger stimulus arrays there is not enough time for
all stimuli information to be transferred to working memory,
double-checking information through visual rescanning or matching with
working memory information, etc.). Furthermore, if cognition is
construed as a series of discrete processing components, which is a
standard view from an information processing or cognitive neuroscience
perspective (e.g., see Kosslyn & Koenig,
1992, p. 406), then an early processing (span of apprehension)
deficit is significant in that it can result in the subsequent
disruption of related and/or later processes located further down
the processing stream. At a more general conceptual level, the
perceptual span task with its severe processing time constraints may be
ideal for the neurocognitive assessment of HIV because cognitive
slowing is a fundamental symptom in HIV-infected individuals.
Perceptual span tasks have also been shown to be highly sensitive to
central nervous system dysfunction. For instance, perceptual span decrements
are evident not only in actively psychotic schizophrenic adults (Neale et al., 1969), but also in schizophrenic adults
in remission (Asarnow & MacCrimmon, 1978)
and first-degree relatives of schizophrenic patients (Wagener et al., 1986). This broad range in task
sensitivity may be ideal for HIV research given the subtle cognitive
symptoms that are frequently observed in HIV-infected individuals.
Thus, the perceptual span task appears well suited to detect subtle as
well as severe HIV-related changes in cognition and central nervous
system functioning.
For the present experiment, a group of HIV+ adults and a control
group of HIV-seronegative adults (HIV−) perform a perceptual span
task where a target stimulus must be identified among a very briefly
presented array of a varying number of nontarget stimuli. The typical
performance measure of interest on such a task is response accuracy in
identifying the target. Therefore, it is predicted that response
accuracy on the perceptual span task will be worse in HIV+ adults
relative to HIV− adults, especially when task demands are greater
where a target must be discriminated among multiple nontargets. Other
aspects of task performance (such as reaction time) will also be
examined.
METHODS
Research Participants
Participants for the present study included 54 HIV+ adults and 19
HIV− controls examined at the West Los Angeles VA Medical Center.
HIV+ participants were recruited from an infectious disease clinic and
from community agencies specializing in services for HIV-infected
patients. Seronegative controls were recruited using posted fliers and
referrals from these sites. Background information and medical history
(including HIV status) of participants was provided through self-report
data based on questionnaires and interviews. Exclusionary criteria
included history of head injury with loss of consciousness in excess of
10 min, history of learning disability, and adverse neurological
history (e.g., stroke or seizure disorder) including secondary
HIV-related central nervous system infection or lymphoma. In addition,
a structured clinical interview composed of the mood, psychotic
spectrum, and substance use sections of the Structured Clinical
Interview for the DSM–IV (First et al.,
1995) was administered by trained psychologists to exclude those
participants with current or past major depression disorder, bipolar
disorder, or psychosis, as well as subjects with current drug or
alcohol abuse/dependence. As can be seen in Table 1, HIV+ and HIV− groups did not
significantly differ in age, education, estimated premorbid IQ as
measured by the American Version of the National Adult Reading Test
(AMNART), self-reported symptoms on the Beck Depression Inventory,
Second Edition (BDI–II), corrected visual acuity, or general
cognitive status as measured by the HIV Dementia Scale. Both groups
included substantial proportions of women, with significantly more
women in the seronegative control group χ2(1,
N = 73) = 6.08, p = .014. A large proportion of
participants were African American, but there was no significant
difference in ethnic composition between subject groups
χ2(4, N = 73) = 1.71, p = .789. Past
alcohol abusers/dependents and past substance
abusers/dependents were present in both groups, with a significant
group difference in substance abusers/dependents. Mean CD4 count
for the HIV+ group was 367 (SD = 266). Sixty-six percent of
HIV+ participants met Centers for Disease Control (CDC,
1992) diagnostic criteria for acquired immunodeficiency syndrome (AIDS)
and all HIV+ participants were currently on highly active antiretroviral
therapy (HAART). Participants provided written informed consent and
were paid 50 dollars for their participation.
Demographics for HIV− and HIV+ Groups
Task and Procedure
Participants completed the perceptual span task (and other parts of
the protocol such as the background questionnaires and interviews,
etc.) as part of a larger neuropsychological study on medication
adherence in HIV-infected adults. All participants completed the entire
protocol according to a uniform procedure, which was always completed
within a single day. The perceptual span task used in the present study
was programmed (by the first author) with SuperLab software (SuperLab Pro Version 2.0, 1999) on a personal
microcomputer and was based on a task used by Estes and Taylor (1964, 1965). Each trial
of the span task began with the presentation of a central fixation
for 1000 ms followed immediately by a stimulus array for a
duration of 50 ms. The stimulus array could be either one, four,
or 12 black letter-characters on a white background (see Figure 1). The letter-characters T and
F were targets and required a left and right button press
respectively (with the left or right index finger) on a response box.
Nontarget letter-characters included randomized selections from the
alphabet with no repeat characters within a single array. Each array
included a single target (except for target-absent catch trials).
Location of characters was randomly assigned on an invisible 4 ×
4 grid subtending approximately 17.0 × 16.2° at the center of
the monitor. Each character subtended approximately 2.6 ×
2.4°. The trial terminated after either a subject response or 2000
ms. The next trial began 2000 ms after either the subject response or
the previous 2000 ms wait period. There were a total of 120 trials with
40 trials (32 target-present trials and 8 target-absent catch trials)
per array size. After 20 trials of practice, participants completed 6
blocks of 20 trials each. Each block contained only one stimulus array
size and blocks were presented in an alternating sequence (1, 4, 12, 1,
4, 12).
An illustration of stimulus events and array size with three example
trials in the perceptual span task. Target = T or F.
Array duration = 50 ms.
RESULTS
Response accuracy for target-present trials was analyzed with a 2
× 3 mixed-model analysis of covariance (ANOVA), with subject
group (HIV− and HIV+) as a between-subjects variable, stimulus
array size (1, 4, and 12) as a within-subjects variable, and history of
either alcohol abuse/dependence or substance abuse/dependence
as a covariate. Responses faster than 100 ms and slower than 2000 ms
were excluded from analyses. A Greenhouse-Geisser correction was
applied to within-subjects effects. Post hoc comparisons were
examined with three one-way ANCOVAs. Because catch trials
(target-absent trials) were presented, an additional 2 × 3
mixed-model ANCOVA was conducted to analyze group differences in false
alarm responses. In addition, to account for possible psychomotor speed
differences between subject groups, a 2 × 3 mixed-model ANCOVA
was conducted on correct reaction times.
For response accuracy, there was a significant main effect of array
size [F(2,140) = 135.5, p < .001], with
mean percent accuracy of 95.7% (SE = 0.8) for array size one,
87.6% (SE = 1.6) for array size four, and 60.6% (SE =
1.8) for array size 12. The main effect for group was significant
[F(1,70) = 4.1, p = .046], with response
accuracy of 83.8% (SE = 2.1) for the HIV− group and
78.8% (SE = 1.3) for the HIV+ group. There was no significant
main effect (p = .143) or interaction (p = .239) for
history of alcohol or substance abuse/dependence. Most importantly,
there was a significant interaction between Group × Array Size
[F(2,140) = 5.1, p = .010], and is
illustrated in Figure 2. To clarify this
interaction, a separate ANCOVA was conducted at each array size to
compare the group difference in response accuracy. There was no
significant difference between the HIV− group and HIV+ group for
array size one (96.2%, SE = 1.4; and 95.1%, SE = 0.8,
respectively) or array size four (89.6%, SE = 2.8; and 85.5%,
SE = 1.7, respectively). There was a significant difference
between groups at array size 12 [65.6%, SE = 3.1; and
55.6%, SE = 1.8, respectively; F(1,70) = 7.5,
p = .008]. Because there was a larger percentage of women
in the HIV− group relative to the HIV+ group, a series of
analyses on response accuracy like those above were conducted but this
time including gender of participant (male or female) as an additional
between-subjects variable. The inclusion of participant gender had no
affect on the pattern of the previously reported results, with no
significant main effect or interactions with gender.
Mean percent correct responses (± SE) for target
identification on the perceptual span task for HIV+ and HIV−
groups.
For false alarm responses, there was a significant main effect of
array size [F(2,140) = 20.6, p < .001],
with mean false alarm rate of 0.08 (SE = 0.05) for array size
one, 0.95 (SE = 0.17) for array size four, and 2.28 responses
(SE = 0.29) for array size 12. The main effect of group
(p = .262) and the interaction between Group × Array
Size (p = .419) were not significant. History of alcohol or
substance abuse/dependence had no significant effects. For reaction
time, there was a significant main effect of array size
[F(2,140) = 50.3, p < .001], with a mean
reaction time of 588 ms (SE = 19) for array size one, 724 ms
(SE = 21) for array size four, and 819 ms (SE = 24)
for array size 12. There was no overall reaction time difference
between the HIV− group (705 ms, SE = 34) and HIV+ group
(716 ms, SE = 20; p = .783) and the interaction
between Group × Array Size was not significant (p =
.731). Again, history of alcohol or substance abuse/dependence had
no significant effects on reaction time. To examine possible
speed–accuracy trade-offs, correlation analyses were conducted
between response accuracy and reaction time for HIV− and HIV+
groups at each array size. As can be seen in Table
2, there was one significant negative correlation coefficient,
in the array size 12 condition for the HIV− group. Thus, there is
no evidence for either group for a speed–accuracy tradeoff.
Correlations between response accuracy and reaction time on the
perceptual span task for HIV− and HIV+ groups
DISCUSSION
As hypothesized, HIV+ adults were less accurate than HIV−
adults in target identification under the most demanding perceptual
condition (array size 12) but not when attentional demands were less. A
standard interpretation of this group difference in span performance is
that HIV+ adults have a restricted span of apprehension, a deficit in
the ability to effectively process large amounts of information at an
early level of visual processing (perhaps due to reduced iconic
storage). This finding is generally compatible with other situations of
information overload in HIV+ adults, such as with working memory (Martin et al., 1995; Worth et
al., 1993) and divided attention (Hinkin et
al., 2000). However, given that perceptual span tasks assess
capacity limitations in very early visual processing, these data show
that HIV+ adults experience information overload (or a capacity
limitation) at an earlier point in processing than previously realized.
An important implication of this finding is that such an early
information processing deficit could modulate or disrupt subsequent
higher-order or more complex levels of cognitive functioning.
Results must be treated with some degree of caution considering the
number of analyses that were conducted. At the same time, although
alpha inflation is a risk here, the p value of .01 for the
interaction between Array Size × Subject Group in the analysis on
response accuracy is relatively small. Conversely, the effect size,
such as the F value of 5.1, is relatively large considering
this is an interaction statistic. In addition, because this is an
initial study of the effects of HIV on perceptual span, an aspect of
cognition that has not received any attention in HIV research, it was
considered important to balance the risks of a type I error with that
of a type II error. Also raising interpretational issues are the
demographic differences between the groups. The seronegatives included
more females than the HIV+ group. In addition, there were more
individuals with a history of substance abuse/dependence in the
HIV+ group. However, both of these variables had no influence, either
as a main effect or interacting with other variables, on perceptual
span performance in any of the analyses. Therefore, it is unlikely that
the worse performance in the HIV+ group was due to these demographic
differences.
The neurophysiological substrate(s) that mediates perceptual span
performance is currently unknown and may in fact differ as a function
of the population being studied. Span deficits in schizophrenia have
been attributed to multiple different, but potentially overlapping,
central nervous system substrates including structures such as
posterior parietal regions and thalamic structures such as the lateral
pulvinar (Asarnow et al., 1991). Given the
known affinity of HIV for subcortical structures, and the link between
these structures and cognition (Martin,
1994), it may be that span deficits among HIV-infected adults
arise as a function of insult to thalamic structures, striatum, or
thalamostriatal projections. Obviously, these same circuits are linked
to cortical regions, including prefrontal cortex, and HIV has been
shown to compromise the integrity of frontal–subcortical
circuits. As such, perceptual span performance may be particularly
sensitive to frontal–subcortical dysfunction in HIV+ adults.
Other factors that could influence or mediate span performance should
also be addressed. For instance, it could be argued that poor span
performance does not have to rely on any processing capacity limitation
but instead is due to the accumulation of decision processing noise
(Duncan, 1980; Pashler,
1998; Tanner, 1961). If the letters in
an array are processed individually and independently on separate
“channels,” which is a reasonable assumption, and there is
a nonzero probability of confusing a nontarget with a target, then the
accumulated chance of making a mistake automatically increases as the
number of letters in the array increases. Therefore, one possible
explanation for the present group difference in span performance is
that HIV+ adults have greater difficulty effectively distinguishing
signals from noise in determining their response. Another factor that
could influence span performance is response strategy. To the degree
that HIV infection leads to deficits in response inhibition, perhaps
the HIV+ adults responded before they had sufficiently processed the
stimulus array information. This is unlikely, however, because there
were no group differences in false alarm responses. Furthermore, it was
only the HIV− group that demonstrated any significant association
between response accuracy and reaction time, a significant negative
correlation in the array size 12 condition indicating more accurate
responses were associated with faster responses. Although this
finding is somewhat counterintuitive, it makes sense when considering
that the duration of information in the visual sensory store (iconic
memory) is very brief, perhaps no greater than 500 ms (e.g., Sperling, 1960). Thus, responses made before array
information rapidly fades away are typically more accurate than slower
responses. A group difference in psychomotor speed is also an unlikely
contributing factor to the HIV-related deficit in span performance
because there were no significant group differences in reaction time.
This lack of HIV-related slowing in reaction time is not unseemly; such
a finding has been reported several times (e.g., Grassi et al., 1999; Miller et
al., 1991; for a review on reaction time in HIV, see Hardy & Hinkin, 2002). In addition, although
subjects were instructed to respond quickly, accurate responding was
the primary goal of subjects. Traditional “reaction time”
tasks are usually easy enough to ensure high levels of accuracy, for
ease of interpretation of reaction time or mental chronometric
differences. Therefore, the emphasis with perceptual span tasks is
typically on response accuracy, as it was in the present study.
In summary, the perceptual span results of the present study support
and extend the finding that HIV-infected adults experience problems
with neurocognitive processing. Specifically, the data indicate that
HIV infection leads to a processing capacity reduction or impairment at
a significantly earlier stage of information processing than has
previously been reported. Future studies will delineate the processing
characteristics of this HIV-related deficit in span of apprehension
(e.g., temporal and perceptual load aspects) as well as examine its
role as a specific “rate limiting factor” (Green, 1993) or mediator for more complex cognition
and behavior in HIV+ adults.
ACKNOWLEDGMENTS
This study was supported by National Institute on Aging Grant No. R03
AG-18549 (awarded to David J. Hardy) and by National Institute of
Mental Health Grant No. R01 MH-58552 (awarded to Charles H. Hinkin). We
gratefully acknowledge Marta Stefaniak, Mona Lam, Karen Mason, and
Bryan Zolnikov for assistance on the present study. Portions of this
paper were presented at the 28th Annual Meeting of the International
Neuropsychological Society, Denver, Colorado, February 2000 and at the
81st Annual Convention of the Western Psychological Association, Maui,
Hawaii, May 2001.
