The GroupSimulator program and documentation were created by David R. Heise. For terms and conditions of use, see http://www.indiana.edu/~socpsy/ACT/legal.htm .
GroupSimulator.nlogo can be downloaded from:
The model runs on your computer within NetLogo, a multi-agent modeling program available without charge at:
http://ccl.northwestern.edu/netlogo/download.shtml . This applet version does not provide all of the functions mentioned below, but all are available in the downloaded version.
Basic documentation concerning the program is given below. More extensive documentation is provided at:
GroupSimulator uses affect control theory to generate interpersonal behaviors, and to track emotions and tensions in multi-agent human groups ranging in size from three to 25 interactants. The program has three kinds of displays: a pictorial view showing interactants’ facial expressions; graphs where group variables are plotted; and a text box for displaying information regarding actions and individuals. Additionally the program provides many controls for defining and analyzing groups in different ways.
The Credits and References section below gives information on affect control theory. The terminology and symbols used in the GroupSimulator program depart somewhat from standard usages in affect control theory. In this program dimensions of Evaluation, Potency, and Activity (EPA) are translated to goodness, dominance, and activation when talking about people, an individual’s fundamental EPA profile is called stable character, and the deflection construct is called tension.
Interactants are selected randomly from a normal distribution centered on specified values of goodness, dominance, and activation. An interactant’s goodness, dominance, and activation constitutes the person’s stable character that he or she tries to maintain in social interaction. Individuals’ stable characters correspond to role-identities combined with personal traits, or to self-sentiments in informal situations like a party. Each individual presumably knows the stable character of every other individual at the scene, either because of a shared definition of the situation, or because the individuals are acquainted with one another personally.
Individual character is displayed graphically in the pictorial view with each individual’s face positioned with regard to two dimensions of individual character, and the size of the individual’s face reflecting the third dimension. The pictorial view’s scatterplot of characters in three dimensions can be viewed from several different perspectives.
A blue image representing the group as a whole also is plotted in the pictorial view, positioned on each dimension at the mean of all individuals’ characters. An action toward the group as a whole impacts impressions of the group entity, and the group image has a facial expression that indicates how the meaning of the group entity is faring. An actor’s action toward the group can be directed at other members individually to change impressions of them, too.
A relative balance of males and females is chosen for the group. Sex partitions the group into subgroups of individuals who can be assigned psychological differences influencing their social interaction.
On each round of interaction, an individual is selected on the basis of some criterion, and that person chooses another individual for dyadic interaction, or else behaves towards the group as a whole. Then another individual chooses a partner for dyadic interaction, or engages the whole group. This process continues until the analyst ends it.
The actor can be chosen to be the one who can produce the least future tension, or the one who has maximum personal tension or minimum personal tension, or the choice can be random.
Alternatively, the actor can be the one whose transient impression is most extreme in one of 26 different directions in the affective space. The directions are defined in the SYMLOG system (see Robert Freed Bales, SOCIAL INTERACTION SYSTEMS, Transaction Publishers, 1999.)
Bales’ Positive-Negative (P-N) dimension is the same thing as the goodness dimension here. Bales’ Forward-Backward (F-B) dimension is a rotation of the dominance and activation dimensions, corresponding to quiet and strong as opposed to noisy and weak. Bales’ Up-Down (U-P) dimension rotates the dominance and activation dimensions to noisy and strong as opposed to quiet and weak (see Bales ’SOCIAL INTERACTION SYSTEMS, pp. 141-3).
The list below shows the SYMLOG label for each of the 26 directions in the affective space, then the values and interests that are characteristic of individuals in that direction, then kinds of behavior that are characteristic, then specific noncorporeal interpersonal acts that characterize the direction, and then the direction’s EPA profile * a point three units away from the origin of the EPA space in the given direction. Interests and Behavior are from Polley (1987; 1989).
Acts characterizing directions were selected from 1,330 behaviors rated on EPA in various English-language databases. Coordinates were computed for each act on all 26 directional axes, and an act was characteristic of a direction if its coordinate on that direction was higher than its coordinates on other directions. (If a direction has less than three characteristic acts, then acts that are somewhat in the given direction are included, marked with a tilde.)
U. Interests: Individual financial success, personal prominence and power.
* Behavior: Active, dominant, talkative.
* Acts: commands, hollers at, disagrees with.
* EPA: 0.00, 2.12, 2.12
UP. Interests: Popularity and social success, being liked and admired.
* Behavior: Supportive, nurturant.
* Acts: asks about something (?), tells something (?) to, answers.
* EPA: 2.12, 1.50, 1.50
UPF. Interests: Active teamwork toward common goals, organizational unity.
* Behavior: Purposeful and considerate.
* Acts: corrects, advises, confers with.
* EPA: 1.73, 2.45, 0.00
UF. Interests: Efficiency, strong impartial management.
* Behavior: Assertive, business-like, impartial.
* Acts: orders, persuades, reprimands.
* EPA: 0.00, 3.00, 0.00
UNF. Interests: Dogmatic enforcement of authority, rules, and regulations.
* Behavior: Authoritarian, controlling.
* Acts: silences, forbids, ~interrogates.
* EPA: -1.73, 2.45, 0.00
UN. Interests: Tough-minded, self-oriented assertiveness.
* Behavior: Egocentric, tough-minded, powerful.
* Acts: criticizes, scolds, quarrels with.
* EPA: -2.12, 1.50, 1.50
UNB. Interests: Rugged, self-oriented individualism, resistance to authority.
* Behavior: Rebellious, provocative.
* Acts: dares, giggles at, taunts.
* EPA: -1.73, 0.00, 2.45
UB. Interests: Active pursuit of change, new and unorthodox ideas.
* Behavior: Unconventional, daring, risk-oriented.
* Acts: joshes, banters with, is saracastic toward.
* EPA: 0.00, 0.00, 3.00
UPB. Interests: Leading group-centered efforts to change and seek new solutions.
* Behavior: Innovative, progressive.
* Acts: chatters to, ~encourages, ~jokes with.
* EPA: 1.73, 0.00, 2.45
P. Interests: Friendship, mutual pleasure, recreation.
* Behavior: Friendly, warm, caring.
* Acts: admits something (?) to, consults,agrees with.
* EPA: 3.00, 0.00, 0.00
PF. Interests: Responsible idealism, collaborative work.
* Behavior: Cooperative, idealistic.
* Acts: explains to, listens to, confides in.
* EPA: 2.12, 1.50, -1.50
F. Interests: Conservative, established, “correct” ways of doing things.
* Behavior: Conforming, conventional.
* Acts: ~prays for, ~assures, ~calms.
* EPA: 0.00, 2.12, -2.12
NF. Interests: Rigid adherence to organizational expections and rules.
* Behavior: Closed-minded, rigid.
* Acts: ~demeans, ~threatens, ~talks down to.
* EPA: -2.12, 1.50, -1.50
N. Interests: Self-protection, self-interest first, self-sufficiency.
* Behavior: Unfriendly, cold, uncaring.
* Acts: nags, ridicules, insults.
* EPA: -3.00, 0.00, 0.00
NB. Interests: Rejection of established procedures, rejection of conformity.
* Behavior: Uncooperative, cynical.
* Acts: wheedles, quibbles with, grouses at.
* EPA: -2.12, -1.50, 1.50
B. Interests: Change to new procedures, different ideas .
* Behavior: Nonconforming, change-oriented.
* Acts: ~babbles to, ~banters with, ~fusses over.
* EPA: 0.00, -2.12, 2.12
PB. Interests: Group-centered approaches to new ideas and new procedures.
* Behavior: Open-minded, flexible.
* Acts: ~chats with, ~cheers up, ~applauds.
* EPA: 2.12, -1.50, 1.50
DP. Interests: Trust in the goodness of others.
* Behavior: Trustful, appreciative, dependent.
* Acts: ~offers something (?) to, ~apologizes to, ~prays with.
* EPA: 2.12, -1.50, -1.50
DPF. Interests: Dedication, faithfulness, loyalty to the organization.
* Behavior: Helpful, accepting of authority.
* Acts: ~makes up with, ~reassures, ~thanks.
* EPA: 1.73, 0.00, -2.45
DF. Interests: Obedience to the chain of command, compliance with authority.
* Behavior: Obiedient, accepting of authority.
* Acts: whispers to, ~murmurs to, ~bows to.
* EPA: 0.00, 0.00, -3.00
DNF. Interests: Grudging self-sacrifice in the interests of the organization.
* Behavior: Martyred, self-punishing.
* Acts: defers to,~fibs to, ~scoffs at.
* EPA: -1.73, 0.00, -2.45
DN. Interests: Passive rejection of popularity, going it alone.
* Behavior: Resentful, uninvolved.
* Acts: begs from. mumbles to, whines to.
* EPA: -2.12, -1.50, -1.50
DNB. Interests: Admission of failure, withdrawal of effort from the task.
* Behavior: Withdrawn, alienated.
* Acts: begs, sucks up to, gibes.
* EPA: -1.73, -2.45, 0.00
DB. Interests: Tolerance for new ideas and different procedures.
* Behavior: Tolerant of change and new ideas.
* Acts: ~drones on at, ~pleads with, ~toadies to.
* EPA: 0.00, -3.00, 0.00
DPB. Interests: Comfort with coworker’s new ideas and suggestions for change.
* Behavior: Accepting of change and innovation.
* Acts: gives in to, ~chit chats with, ~flatters.
* EPA: 1.73, -2.45, 0.00
D. Interests: Giving up personal needs and desires, passivity.
* Behavior: Passive, submissive, quiet.
* Acts: murmurs to, ~whispers to, ~stammers at.
* EPA: 0.00, -2.12, -2.12
Choices of partner are based on similarity in emotion, or similarity in character, or by selecting the other who has maximum personal tension or minimum personal tension, or by a determination of who might generate minimum future tension in an interpersonal event; or as a random choice. Alternatively, the partner can be the one who is most extreme in one of the 26 affective directions.
Each event is displayed in the pictorial view with a green arrow linking the actor to the object of action. Links on past rounds of interaction are displayed as aqua-colored lines. The density of ties among individuals based on actions between them is shown by a dotted line in a graph that records process over time: a dot in the process graph shows the percentage of actual ties relative to the possible number, divided by five. The value of the dot times 20 gives the percentage of occurring links relative to the maximum possible links.
A behavior is generated for the chosen actor and object that minimizes actor, object, and behavior tensiones in accordance with affect control theory. The goodness, dominance, and activation of behaviors can be examined over time in the process graph by selecting “Behavior EPA” as the variables to be plotted. The profile of the behavior generated at a particular moment can be examined by choosing to view outputs.
Emotions are displayed as expressions on the faces of individuals in the pictorial view. The facial expressions consist of just nine variations: neutral, elated, calm, bedazzled, beatific, angry, disgusted, scared, and sad. The average emotion of the two interactants in each acting dyad can be plotted over time in the process graph by selecting “Emotion EPA” as the variables to be plotted. A numerical EPA profile for each interactant’s emotion at a particular moment can be obtained by choosing to view outputs.
“EPA impressions” can be selected as the variables to be plotted in the process graph, in which case the average impressions of each acting dyad’s goodness, dominance, and activation are plotted over time. Impressions of interactants as a result of the current event can be seen by choosing to view outputs. Impressions of individuals are used to select actors on the next round if the basis for choosing actors is “most esteemed,” “most potent,” “most active,” or one of the 26 affective directions.
The total tension produced by the current event is plotted over time in the process graph. Numerical levels of tension for individuals can be seen by choosing to view outputs.
The right side of the display shows output related to Interaction Process Analysis (IPA), a methodology for observing group interactions developed in the middle of the 20th century by Robert Bales. (For more information, see his SOCIAL INTERACTION SYSTEMS,ibid.) One graph shows interactants’ relative frequencies of performing and receiving actions. Another graph shows the distribution of simulated acts categorized into IPA categories relating to instrumental and expressive aspects of group process. A text box shows the actor and recipient of the current event, the behavior’s SYMLOG direction translated into words for behaviors of that type, and a possible topic that the actor might be forwarding, based on the SYMLOG value direction of the actor’s character.
Clicking the SETUP button populates the pictorial view with a set of individuals. Clicking the ON-OFF button generates continuous rounds of social interaction among the individuals until the button is clicked again. The SPEED slider above the pictorial view controls the rate at which rounds of interaction occur. (You can edit the GO button so that clicking the button produces just one round of interaction.) Clicking the NEXT button causes just one action to be generated.
The emotional expressions on agents’ faces show their current emotions. If the INITIAL-TENSION slider is set to zero, then the faces initially show characteristic emotions experienced when reality perfectly confirms characters. Otherwise the beginning emotions reflect individuals’ characters along with tensions randomly assigned to each individual.
You can view the group from different perspectives by making a selection on the VIEW menu, and then clicking the RE-PLOT button. (You also can use the RE-PLAY or the RE-START button, but this will erase interactions that occurred previously.) The VIEW menu’s option of “E x A (P)” shows individuals arrayed in terms of goodness on the vertical dimension and activation horizontally, with size of face representing dominance. The “P x A (E)” option shows dominance versus activation, with size representing goodness. “P x E (A)” shows dominance versus goodness, with size representing activation. The “SYMLOG (UD)” option shows individuals distributed in terms of Forward-Backward versus Positive-Negative, with size representing Up-Down. “MAX SYMLOG” shows the same thing but expanded to maximally fill the graph, as suggested in Bales and Cohen’s (1979) drawing instructions.
Clicking the RE-PLAY button erases effects of all rounds of interaction and returns individuals to their starting states, with each individual retaining the same stable character and the same initial tension. Clicking ON-OFF then reproduces the prior rounds of interaction. Clicking the RE-START button also returns individuals to their starting states, but clicking ON-OFF produces fresh rounds of interaction in the same group, with individuals perhaps choosing affiliations in a different random order than on the previous run.
You determine the number of individuals�from three to 25�using the NUMBER-OF-INDIVIDUALS slider. Faces become smaller as more individuals are selected.
The proportion of individuals who are female is set with the PROPORTION-FEMALES slider.
The number of individuals and the proportion who are female must be decided before clicking the SETUP button.
The expected character of males or of females can be set using the GOODNESS, DOMINANCE, and ACTIVATION sliders for each sex. Individuals of a given sex are drawn from a multivariate normal distribution centered on the numbers set in the character sliders for that sex. Individuals with character values beyond realistic limits of -4 and +4 on each scale are discarded, and a replacement is obtained by drawing again.
The ACTOR-CHOICE drop-down menus, one for each sex, determine the basis for deciding which person in the group will act next. The “least future tension” option is implemented by computing the optimal event among every possible pairing of actors and objects. Maximal or minimal personal tension relate to the distance between an individual’s fundamental character and the prevailing impression of the individual in the group. The criteria specified by SYMLOG directions are applied to impressions created by the last event (not to stable character variables).
The basis for attraction between individuals is set on the OBJECT-CHOICE drop-down menu.
The EQUATIONS drop-down menu determines which impression-formation equations are used in calculations: male equations for everyone, female equations for everyone, or the average of male and female equations for everyone (unisex). The option of “random acts” obtains EPA profiles for behaviors by drawing from a multinormal distribution centered at the origin of the EPA space, with a standard deviation set on the INITIAL-TENSION slider; unisex equations are used to compute impressions and emotions. The option of “identity echos” generates EPA profiles for an agent’s behavior by random draws from a multinormal distribution centered in the EPA space at the agent’s character, or EPA sentiment, with a standard deviation set on the INITIAL-TENSION slider; impressions and emotions are disabled with this option because they explode to infinity for certain types of characters when behaviors are chosen this way. The option of “one individual” uses an equation set describing impression processes of a specific person, and this is used for everyone in the group.
The RECIPROCAL-ACTION-PROBABILITY slider determines how frequently a dyad exchanges actions back and forth, at the expense of actions involving others. Dyadic exchanges of this sort occur in question-answer sequences, and other conversational structures. The probability should be set higher in groups experiencing conflict (Pincus et al. 2008). The ADDRESS-GROUP-PROBABILITY slider determines how frequently actors address the group as a whole, even if interacting with some individual would produce less tension. The group entity can be removed as a possible object of interaction by setting the slider to a negative value. The GRP-ACT-TO-ALL switch determines whether actions on the group affect impressions of the group entity alone, or whether the actor’s behavior is directed to all other members of the group as well.
Green arrows in the pictorial view show which dyads are operative on the current round of interaction. (The faces of some interactants may hide the faces of others, in which case arrows for events pairing the interactants also may be hidden.) Interacting dyads initially are linked with dim connecting lines, but the lines gradually become brighter if interaction occurs over and over in the dyad. Dyads with pleasant relations are connected with aqua or white lines, those with hostile relations are connected with red lines.
Right-clicking a face in the pictorial view raises a menu of options. One line of this menu specifies the sex and identification number of the topmost face. Other lines may specify the sex and identification numbers of faces hidden by the top one. Moving the mouse pointer over an identification line produces a sub-menu. Choosing the “follow” option on the sub-menu will keep the selected face at the middle of the pictorial view in future rounds of interaction. The “watch” option highlights the selected face in future rounds of interaction. The “inspect” option opens a window with a blow-up of the individual’s face on top and various kinds of information below. This window can be left open to track technical details regarding an individual during subsequent rounds of interaction. The window also can be used to change an individual’s character by entering new values for the individual’s fundamental EPA profile and pressing Enter. (Changing most other variables would be futile because the new values would be discarded by the program on the next round of interaction.)
The process graph shows over-time development of average tension and density of links. Exact coordinates of any point on the graph can be seen by positioning the mouse pointer over the point. Three additional lines (identified as E, P, and A in the legend) display over-time averages of Evaluation, Potency, and Activity components of individuals’ emotions, or of impressions of the individuals, or of behaviors being enacted in dyads. You select which you want to see with the GRAPH-VARIABLES drop-down menu. Note While the scales of the horizontal and vertical axes of the graph are initialized to fit a variety of analyses, the scales change automatically when greater ranges are needed.
Rank-frequency histograms display the proportion of times that each individual has been an actor, and the proportion of times that each individual has been the recipient of others’ actions. The group as a whole is included in the recipient graph, colored orange.
The IPA histogram shows the proportions of actions in Bales’ twelve IPA categories. The drop-down menu named IPA-CODING-BASIS provides different bases for coding behavior EPAs into IPA categories. For example, the option of “2004 sentiments” defines each IPA category with mean EPA profiles for three signature acts in that category, averaged over ratings obtained from Indiana University males and females in 2004.
The INPUT-OUTPUT text box gives fundamentals and initial transients for each actor at the top. Thereafter each line shows the actor and the target of an implemented action, and extreme representatives of actions in the given SYMLOG direction, drawn from 1,330 behaviors rated on EPA in various English dictionaries. A possible topic is indicated, corresponding to interests associated with the SYMLOG value direction of the actor’s stable character. The BEHAVIOR-TYPE drop-down menu allows selecting between verbal acts and acts that encroach on the personal space or privacy of the target person.
File input or output cannot be employed with the applet version of GroupSimulator, only with the downloaded version.
Clicking the SAVE-IO switch “on” causes simulation data to be written to two files named Data_GroupSimulator_actions.txt and Data_GroupSimulator_runs.txt. The files are written in the same directory as the GroupSimulator.nlogo file. The first line of each file identifies the model that produced the results. The “actions file contains information on each simulated action. The ”runs" file contains summary results for each complete run, with run size set using the RUN-SIZE slider.
Clicking the FAST switch to “on” stops updating of all visual displays in order to speed up processing when conducting multiple runs. Data continue to be saved in files, and the “ticks” counter above the pictorial view continues to operate, showing progress of the current simulation.
Clicking the READ-FILE button opens a file directory window so that you can load a set of interactants from a file that you have prepared in a text processing program. You also can re-load a group that you analyzed previously, if you saved the contents of the INPUT-OUTPUT box for that analysis to a file.
Bales, Robert Freed. 1999. Social Interaction Systems: Theory and Measurement. New York: Transaction.
Bales, Robert Freed and Stephen P. Cohen. 1979. SYMLOG: A System for the Multiple Level Observation of Groups. New York: Free Press.
Heise, David R. 2007. Expressive Order: Confirming Sentiments in Social Actions. New York: Springer. (The book has three sections. Part 1 provides a plain-language exposition of the theory. Part 2 presents the mathematical derivations that define sentiment-confirming behavior, labeling, attribution, and emotion. Part 3 describes the research program associated with the theory, and the Interact computer simulation program that is used in research.)
Heise, David R. 2010. SURVEYING CULTURES: DISCOVERING SHARED CONCEPTIONS AND SENTIMENTS, Hoboken NJ: Wiley. (Covers measurement of sentiments and the development of equations describing impression formation.)
Pincus, David, Kristen M. Fox, Katherine A. Perez, Jaclyn S. Turner, and Andrew R. McGeehan. 2008. “Nonlinear Dynamics of Individual and Interpersonal Conflict in an Experimental Group.” Small Group Research 39:150-178.
Polley, Richard Brian. 1987. “The dimensions of social interaction: A method for improving rating scales.” Social Psychology Quarterly 50:72-82.
Polley, Richard Brian. 1989. “On the dimensionality of interpersonal behavior: A reply to Lustig.” Small Group Behavior 20:270-278.