"AIDS" models the spread of the human immunodeficiency virus
(HIV), via sexual transmission, through a small isolated
human population.  As is well known now, HIV is spread in a
variety of ways of which sexual contact is only one.  HIV
can also be spread by needle-sharing among injecting drug
users, through blood transfusions [although this has become
very uncommon in countries like the United States in which
blood is screen for HIV antibodies], or from HIV-infected
women to their babies either before or during birth, or
afterwards through breast-feeding.  This model focuses only
on the spread of HIV via sexual contact.  It therefore
illustrates the effects of certain sexual practices across a

The model examines the emergent effects of three aspects of
sexual behavior.  The user controls the population's
tendency to practice abstinence, the amount of time an
average "couple" in the population will stay together, and
the population's tendency to use condoms.  Exploration of
the first and second variables may illustrate how changes in
sexual mores in our society have contributed to increases in
the prevelance of sexually transmitted diseases, while
exploration of the third may provide a contemporary solution 
to the problem. 

The model also examines the effects of the population's
tendency to seek out testing for HIV.  By allowing the user
to control how often an average individual will choose to be
tested, the user can explore an important dimension of HIV's
threat to public health.  Because the virus does not make
itself immediately known in its host, individuals are often
infected for some time before a test or immune difficiency
symptoms (which leads to a test) identifies them as such. 
Regular identification of individuals infected by the virus
could have significant public health impacts if knowledge of
the infection positively affected sexual behaviors.  This
model explores this possibility by making all individuals
who know of their positive HIV status always practice
safe sex.  

The model uses "couples" to represent two people engaged in
sexual relations.  Individuals wander about the screen when
they are not in couples.  Upon coming into contact with a
suitable partner, there is a chance the two individuals will
"couple" together.  When this happens, the two individuals
no longer move about, they stand next to each other holding
hands as a representation of two people in a sexual
relationship.  The presence of the virus in the population
is represented by the colors of individuals who are
infected.  Three colors are used.  Green individuals are
uninfected.  Pale blue individuals are infected, but their
infection is unknown.  Red individuals are infected and
their infection is known.

A button SETUP creates individuals with particular
behavioral tendencies according to the values of the
interface's five sliders (described below).  Once the
simulation has been setup, there are two ways to run it, by
either pushing the button GO or the button EXPERIMENT.  GO
starts the simulation and runs it continuously until GO is
pushed again.  During a simulation initiated by GO,
adjustments in sliders can affect the behavioral tendencies
of the population.  The button EXPERIMENT, however, starts
the simulation, runs it for 2 years and then stops.  At the
end of a simulation initiated by the button EXPERIMENT, a
report of the simulation's results are sent to the output
window.  During a simulation initiated by EXPERIMENT, the
use of sliders has no effect on the population. 

The model's interface window also contains two monitors. 
The first monitor shows the percent of the population which
is infected by HIV.  The second monitor shows how much time
has transpired during the running of the model.  This model
measures each time-step as one week.  Fifty-two time-steps
equals a year, and years are shown in the second monitor.  

The total number of individuals in the simulation is
controlled by the slider INIT-PEOPLE (initialized to vary
between 100 - 1000), which must be set before SETUP is

During the model's setup procedures, all individuals are
given "tendencies" according to four additional sliders. 
These tendencies are generally assigned in a normal
distribution, so that, for instance, if a tendency slider is
set at 8, the most common value for that tendency in the
population will be 8.  Less frequently,  individuals will
have values 7 or 9 for that tendency, and even less
frequently will individuals have values 6 or 10 (and so on). 

The slider ABSTNCE (0 - 10) determines the tendency of the
individuals to become involved in couples (as stated
earlier, all couples are presumed to be sexually active). 
When the setup procedures are run with a high ABSTNCE slider
value, individuals are less likely to couple than when the
slider is set at a low value.  When the ABSTNCE slider is
set at ten, coupling is unlikely, although (because
tendencies are assigned in a normal distribution) it is
still possible.  Note that when deciding to couple, both
individuals involved must be "willing" to do so, so high
abstinence tendencies in two individuals are actually
compounded (i.e. two individuals with a 50% chance of
coupling actually only have a 25% of coupling in a given

The slider DURATION (1 - 200) determines how long
individuals are likely to stay in a couple (in weeks). 
Again, the tendencies of both individuals in a relationship
are considered.  The relationship only lasts as long as is
allowed by the tendency of the partner with a shorter
committment tendency.

The slider PROTECTION (0 - 10) determines the tendency of
individuals in the population to use condoms.  If an
individual uses a condom, it is assumed in this model that
no infection by HIV is possible.  Note that this tendency
(like the others) is probablistic at several levels.  For
instance, when PROTECTION is set to 9, most of the
individuals have a protection value of 9, although some have
protection values of 8 or 10, and fewer yet have protection
values of 7 or 11 (11 would be off-scale and the same as 10
for all practical purposes).  Also, though, an individual
with a protection value of 9 will still sometimes not choose
to use a condom (10% of the time, roughly).  

Simulation of condom-use is further complicated by the fact
that if one partner "wants" to use a condom while the other
partner does not, the couple does not use condoms.  This
characteristic of the model is representative of the
dynamics of some sexual relations, but not others.  The
decision was somewhat arbitrary, and the user is invited to
play with this characteristic and others in the "Extending
the Model" section of this window. 

The slider TESTING (0 - 10) is the final slider of the
interface.  It determines the tendency of individuals in the
population to get tested for HIV.  Set at 10, the average
turtle will get tested for HIV once a year.  Set at 2, the
average individual will only get tested every five years. 
This tendency has significant impact because the model
assumes that individuals who know that they are infected
always practice safe sex, even if their tendency as healthy
individuals was different.  Again, this characteristic of
the model is only represented of the behaviors of some
individuals.  The model was designed in this way to
highlight the public health effects associated with frequent
testing *and* appropriate responses to knowledge of one's
HIV status.  To explore the impact of alternative behaviors
on public health, the model code can be changed relatively
painlessly.  These changes are described in the section, 
"Extending the Model." 

The model's plot window draws a three column histogram
showing the total number of uninfected individuals, infected
individuals (whose positive status is not known), and
infected individuals (whose positive status is known).

Set the INIT-PEOPLE slider to 300, DURATION to 100, and the
other three sliders to 0. Push SETUP and then push GO. 
Notice that many individuals rapidly pair up into stationary
"couples" with the patches behind them stamped a dark gray. 
These couples represent sexual activity between the two
individuals.  Individuals that continue to move about (and 
do not have a gray patch behind them) are not engaging in
sexual relations.  With DURATION set to 100, an average
individual will have monogamous sexual relations with a
partner for about 100 weeks (approximately 2 years) before
the ending the sexual relationship and searching out a new

Stop the simulation (by pressing the GO button once again),
move the ABSTNCE slider to 10, push SETUP, and start the
simulation again (with the GO button).  Notice that this
time, couples are not forming.  With a high tendency towards
abstinence, individuals do not choose to have sex, which in
this model is depicted by the graphical representation of
couplehood.  Notice that with these settings, HIV does not
typically spread.  However, spreading could happen since the
population's tendency is set according to a normal
distribution and a few turtles will probably have abstinence
values below 10 at this setting. 

Again reset the simulation, this time with ABSTNCE set back
to 0 and DURATION set to 1.  Notice that while individuals
do not stand still next to each other for any noticeable
length of time, coupling is nevertheless occurring.  This is
indicated by the occasional flash of dark gray behind
individuals that are briefly next to one another.  Notice
that the spread of HIV is actually faster when duration is
very short compared to when it is very long.

Now run a simulation with DURATION equal to 1, ABSTNCE set
to 10, PROTECTION set to 10, and TESTING set to 10.  With
negligable couple formation and high condom-use anyway,
there should be no spread of HIV.  Notice how pale blue
"infection unknown" individuals turn red much quicker in
this simulation.  When the individuals turn red, their HIV
status becomes known to them.  Some individuals turn red
because they have been infected for long enough that they
develop symptoms which alerts them to the need for an HIV
test.  Others become red because they choose to be tested. 
With TESTING set to 10, healthy individuals are also being 
tested, but there color does not change since the tests come
back negative.  

When all the individuals in the simulation are either green
or red, change the sliders without stopping the simulation. 
Set DURATION to 100, ABSTNCE to 0, PROTECTION to 0, and
TESTING to 0.  Notice that despite the immediate formation
of couples and the fact that condom-use tendency is
presumably very low, some between healthy (green)
individuals and infected (red) individuals, no spread of HIV
occurs.  This is because while the model expects HIV
positive individuals to continue to engage in sexual
relations, it presumes that those individuals will always
use condoms and that the condoms will always work.  The
rational for this design is discussed briefly in the "What
is it?" section of this window.  If all individuals are
either healthy or known-HIV-positive, the simulation stops
because no further change in the prevelance of HIV is

Finally, set INIT-PEOPLE to 1000 to notice that couples can
form on top of each other.  Watch the simulation until you
see individuals by themselves, but standing still and with a
gray patch behind them indicating coupling.  Stop the
simulation and use the mouse to drag away individuals that
surround the apparently "single" (but coupled) individual. 
Underneath one of its neighbors, you will find the
individuals partner.  This apparent bug in the program is
necessary because individuals need to be able to couple
fairly freely.  If space constraints prohibited coupling,
unwanted emergent behavior would occur with high population

It's possible that you will not immediately notice that the
couples are formed by a partnership of "righty" shapes and
"lefty" shapes.  These shapes are not intended to represent
genders in any fashion, but merely to provide a useful way
to depict good couple graphics.  In order for the hands of a
righty and lefty to match up, both must be off-centered in
their patch.  Without this feature, two couples next to each
other would appear to be a line of four individuals (instead
of two groups of two) because the turtle-shapes would hold
hands on both sides of their patches.  It is intended that
the differences between righty and lefty shapes not be
especially apparent in order to prevent unintended
associations with gender.  Any righty and lefty shape can be
thought of as male or female or neither.

      2)  THINGS TO TRY
Run a number of experiments with the EXPERIMENT button to
find out the effects of different variables on the spread of
HIV.  Try using good controls in your experiment.  Good
controls are when only one variable are changed between each
trial.  For instance, to find out what effect the average
duration of a relationship has, run four experiments with
the DURATION slider set at 1 the first time, 2 the second
time, 10 the third time, and 50 the last.  How much does the
prevelance of HIV increase in each case?  Does this match
your expectations?  

Are the effects of some slider variables mediated by the
effects of others?  For instance, if lower duration values
seem to increase the spread of HIV when all other sliders
are set at 0, does the same thing occur if all other sliders
are set at 10?  You can run many experiments to test
different hypotheses regarding the emergent public health
effects associated with individual sexual behaviors.

Like all computer simulations of human behaviors, this model
has necessarily simplified its subject area substantially. 
The model therefore provides numerous opportunities for lots
of extensions... 

The model graphically depicts sexual activity as two people
standing next to each other.  This suggests that all couples
necessarily have sex and that abstinence is only practiced
in singlehood.  The model could be changed to reflect a more
realistic view of what couples are.  Individuals could be in
couples without having sex.  To show sex, then, a new
graphical representation would have to be employed.  Perhaps
sex could be symbolized by having the patches beneath the
couple flash briefly to a different color.

The model does not distinguish between genders.  This is an
obvious over-simplification chosen because making an
exclusively heterosexual model was untenable while making
one that included a variety of sexual preferences might have
distracted from the public health issues which it was
designed to explore.  However, extending the model by adding
genders would make the model more realistic.

The model assumes that individuals who are aware that they
are infected always practice safe sex.  This portrayal of
human behavior is clearly not entirely realistic, but it
does create interesting emergent behavior which has genuine
relevance to certain public health debate.  However, an
interesting extension of the model would be to change
individuals' reactions to knowledge of HIV status.

The model assumes that condom use is always 100% effective. 
In fact, responsible condom use is actually slightly less
than ideal protection from the transmission of HIV.  Add a
line of code to the INFECT procedure to check for a slight
random chance that a particular episode of condom-use is not
effective.  Another change that can be made in the INFECT
procedure regards a couple's choice of condom-use.  In this
model, when the two partners of a couple "disagree" about
whether or not to use a condom, the partner that doesn't
wish to use a condom makes the choice.  The opposite
possibility is also possible. 

Finally, certain significant changes can easily be made in
the model by simply changing the value of certain global
variables in the procedure SETUP-GLOBALS.  For instance, the
variable EXPERIMENT-LENGTH is set to 2 (years).  The
variable can be changed for longer or shorter experiments. 
Two other variables set in this procedure that are
especially worthy of investigation are INFECTION-CHANCE and
SYMPTOMS-SHOW.  The former controls what chance an infection
has of spreading from an infected to an uninfected partner
if no protection is used.  The variable is currently set to
50, meaning that in a week of sexual relations, the chance
of infection occurring is 50%.  It is not clear at this time
how realistic that figure is.  SYMPTOMS-SHOW is the variable
that controls how long, on average, a person will have the
HIV virus before symptoms begin to appear alerting that
person to the presence of some health problem.  It is
currently set to 200 (weeks) in this model. 

Notice that the four procedures which assign the different
tendencies generate many small random numbers and add them
together.  This produces a normal distribution of tendency
values.  A random number between 0 and 100 is as likely to
be 1 than it is to be 50 as it is to be 99.  However, the sum
of 20 numbers between 0 and 5 is much more likely to be 50
as it is to be 99.

Notice that the global variables SLIDERCHECK1, SLIDERCHECK2,
and so on, are assigned with the values of the various
sliders so that the model can check the sliders against the
variables while the model is running.  Every time-step, a
slider's value is checked against a global variable that
holds the value of what the slider's value was the time-step
before.  If the slider's current value is different than the
global variable, StarLogoT knows to call procedures which
adjust the population's tendencies.  Otherwise, those
adjustment procedures are not called.