IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART A: SYSTEMS AND HUMANS, VOL. 36, NO. 3, MAY 2006 611CorrespondenceGroup Path FormationRobert L. Goldstone, Andy Jones, and Michael E. RobertsAbstract—When people make choices within a group, they are fre-quently influenced by the choices made by others. We have experimentallyexplored the general phenomenon of group behavior where an earlyaction facilitates subsequent actions. Our concrete instantiation of thisproblem is group path formation where people travel between destina-tions with the travel cost for moving onto a location inversely related tothe frequency with which others have visited the location. We comparethe resulting paths to optimal solutions [minimal Steiner trees (MSTs)]and the “Active Walker” model of pedestrian motion from biophysics.There were systematic deviations from beeline pathways in the direction ofMST. These deviations showed asymmetries (people took different pathsfrom A to B than they did from B to A) and varied as a function of thetopology of the destinations, the duration of travel, and the absolute scaleof the world. The Active Walker model accounted for many of these results,in addition to correctly predicting the approximate spatial distributionof steps.Index Terms—Conformity, culture, group behavior, group choice, min-imal Steiner trees (MSTs), path dependence, route formation, stigmergy,trail systems.I. INTRODUCTION“Traveler, there is no path. Paths are made by walking.”Antonio MachadoComplex adaptive-system models are typically applied to naturalphenomena, such as the pattern of stripes on zebras or seeds on asunflower [2]. However, these modeling tools are also illuminating forunderstanding group behavior [3]. A complex-systems perspective canliberate us of our customary habit of focusing on individual behaviorand instead encourage attention to emergent social organizations at ahigher level than the individual [4], [5]. Social phenomena such asrumors, the emergence of a standard currency, transportation systems,the World Wide Web, resource harvesting, crowding, and scientificestablishments arise because of individuals’ beliefs and goals, butthe eventual form that these phenomena take is rarely the goal ofany individual.The current research explores spatial group-choice behavior. Whenpeople make choices within a group, they are frequently influenced bythe choices made by others ([6]; see [7] for a review). In particular,there is a striking similarity of choices and behaviors by people ina group, as is revealed by any casual observation of high schools,bars, or scientific meetings. One possible cause of this group-levelsimilarity is that early choices by group members change the en-vironment and, hence, the attractiveness of choices for subsequentgroup members. Many times, initial pioneers reduce the costs forfollowers who pursue similar paths. For example, recent cognitiveneuroscience researchers can perform functional magnetic resonanceimaging (fMRI) experiments rapidly and efficiently because of priorManuscript received September 16, 2004; revised February 18, 2005 andFebruary 25, 2005. This work was supported in part by the National ScienceFoundation under Grant 0527920 and in part by the Department of Educationunder the Institute of Education Sciences Grant R305H050116. This paper wasrecommended by Associate Editor Y. Liu.The authors are with the Psychology Department, Indiana University,Bloomington, IN 47405 USA (e-mail: [email protected]).Digital Object Identifier 10.1109/TSMCA.2005.855779researchers’ developments of brain imaging. Much, if not all, of whatwe know of as culture is built up in this fashion—by following andextending the innovations of predecessors [8].II. DETERMINANTS OF SIMILARITY WITHIN A GROUPThere are several reasons why people in a group may behave simi-larly. People in a group may simply be similar to one another, whichmay be why they formed the group in the first place. Across manykinds of groups, including religious, special interest, occupational,and recreational communities, similar people tend to flock together[9], [10] and stay together once they have collected together [11].A second reason for similarity within a group is that humans areuniquely adept at adopting one others’ innovations. Cultural identityis largely due to the dissemination of concepts, beliefs, and artifactsacross people. Our capacity for imitation has been termed “no-triallearning” by Bandura [12], who stressed that people perform behaviorsthat they would not have otherwise considered by imitating oneanother. Imitation is commonly thought to be the last resort for dulland dim-witted individuals. However, cases of true imitation are rareamong nonhuman animals [13], requiring complex cognitive processesof perception, analogical reasoning, and action preparation. Whencombined with variation and adaptation, imitation is one of the mostpowerful methods for quick and effective learning. Learning theoryhas historically stressed individual learning in solitary conditions.However, there is a growing realization that learning and thinking, ingeneral, are inherently social activities [14].Social psychologists have conducted a considerable amount ofresearch pursuing a third and related possibility: That people aremotivated to conform to the group in order to curry the group’s favor[15]. In social psychology, there has been a long and robust literatureon conformity in groups [7], [16]. To some degree, conformity is foundbecause people desire to obtain social approval from others [17]. Forexample, sometimes, when people give their answers privately, theyare less likely to conform to the group’s opinion than when respondingpublicly [15]. However, at other times, the conformity runs deeperthan this, and people continue to conform to the group’s opinioneven privately [16]. In Ash’s [18] classic experiments on conformity,subjects judged unambiguous stimuli after hearing other opinionsoffering incorrect estimates. Of his subjects, 69% conformed to thebogus majority. Milgram’s [19] subsequent studies of conforming toauthority dictates were directly inspired by Ash’s studies.There is considerable work on how individuals conform or useinformation provided by others. Field work also explores actual smallgroups of people engaged in cooperative problem solving [3]. How-ever, there is less work with laboratory-controlled conditions that ex-plores the dynamics of a group