An experimental study of the coloring problem on human subject networks

Science

Volumn 313, Issue 5788, 2006, Pages 824-827

  Kearns, Michael ^a   Suri, Siddharth ^a   Montfort, Nick ^a  

^a UNIVERSITY OF PENNSYLVANIA   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

BEHAVIORAL RESEARCH; COLORING; GRAPH THEORY; MATHEMATICAL MODELS; PROBLEM SOLVING;

HUMAN SUBJECT NETWORKS; NATURALLY OCCURRING NETWORKS;

SOCIAL ASPECTS;

EMPIRICAL ANALYSIS; EXPERIMENTAL STUDY; HUMAN BEHAVIOR; NETWORK DESIGN;

ARTICLE; BEHAVIOR; COLOR; HUMAN; HUMAN EXPERIMENT; INFORMATION PROCESSING; NORMAL HUMAN; PRIORITY JOURNAL; PROBLEM SOLVING; SOCIAL NETWORK; SOCIAL PSYCHOLOGY; TASK PERFORMANCE;

GAME THEORY; GROUP PROCESSES; HUMANS; INTERPERSONAL RELATIONS; SOCIAL BEHAVIOR; SYSTEMS THEORY;

EID: 33747084763     PISSN: 00368075     EISSN: 10959203     Source Type: Journal    
DOI: 10.1126/science.1127207     Document Type: Article

References (27)

1. J. Travers, S. Milgram, Sociometry 32, 425 (1969).
2. M. Gladwell, The Tipping Point (Little, Brown, Boston, 2000).
3. S. Milgram, Psychol. Today 61, 60 (1967).
4. M. S. Granovetter, AJS 78, 1360 (1973).
5. B. A. Huberman, L. A. Adamic, in Complex Networks, E. Ben-Naim, H. Frauenfelder, Z. Toroczkai, Eds. (Springer, Berlin, 2004), pp. 371-398.
6. O. Sporns, G. Tononi, G. M. Edelman, Cereb. Cortex 10, 127 (2000).
7. A. Wagner, D. Fell, Proc. R. Soc. London B. Biol. Sci. 268, 1803 (2001).
8. A. Broder et al., in Proceedings of the Ninth International World Wide Web Conference (Elsevier, Amsterdam, 2000), pp. 309-320.
9. D. J. Watts, Small Worlds: The Dynamics of Networks Between Order and Randomness (Princeton Univ. Press, Princeton, NJ, 1999).
10. A.-L. Barabási, R. Albert, Science 286, 509 (1999).
11. J. Kleinberg, Nature 406, 845 (2000).
12. M. R. Garey, D. S. Johnson, Computers and Intractability (W. H. Freeman, New York, 1979).
13. J. A. Zoellner, C. L. Beall, IEEE Trans. Electromagn. Compatibil. 19, 313 (1977).
14. J. Janssen, D. Krizanc, L. Narayanan, S. Shende, J. Algorithms 36, 119 (2000).
15. R. M. Karp, in Complexity of Computer Computations, Proc. Sympos. I8M Thomas J. Watson Res. Center, Yorktown Heights, NY (Plenum, New York, 1972), pp. 85-103.
16. S. Khot, in Proceedings of 42nd IEEE symposium on Foundations of Computer Science, (2001).
17. U. Feige, J. Kilian, J. Comput. Syst. Sci. 57, 187 (1998).
18. An earlier session of similarly controlled human subject experiments was conducted in September 2005 as a field test for the system. Although we do not report on these here because they used different network structures, they largely foreshadowed our main findings and influenced the design of the January 2006 experiments. The same was true of a graph-coloring exercise held in a University of Pennsylvania class in February 2005 in which communication between subjects was restricted to obey the network structure (but was otherwise unconstrained, and included open discussion and negotiation). Details are in the supporting material (26).
19. D. J. Watts, S. H. Strogatz, Nature 393, 440 (1998).
20. For each network structure, trials were conducted under three information conditions. For each network, we also examined two different incentive or payment schemes for subjects. The variation across these two design variables yielded 3 × 2 = 6 trials of each network. Two networks were given an additional trial to examine potential learning effects, yielding a total of seven trials. Details in (26).
21. C. F. Camerer, Behavioral Game Theory (Princeton Univ. Press, Princeton, NJ, 2003).
22. One session of experiments used collective or global incentives, in which all 38 participants were each paid $5 for every experiment that resulted in a proper coloring within the 5 min allotted. The other session used individual or local incentives, in which a participant received $5 if an experiment ended (due either to proper coloring or the termination of 5 min) with their own color being different from all their network neighbors. There were no noteworthy differences between the two incentive schemes in any of the measures discussed.
23. There is a downward bias introduced in the average experiment durations because they were capped at 300 s. However, the distribution of unsolved experiments was such that allowing these experiments to continue to solution would only have strengthened the results reported here; see (26).
24. J. Kleinberg, in Proceedings of the 32nd Annual ACM Symposium on Theory of Computing (Association for Computing Machinery, New York, 2000), pp. 163-170.
25. P. Dodds, R. Muhamad, D. J. Watts, Science 301, 827 (2003).
26. Materials and methods are available as supporting material on Science Online.
27. The authors thank D. Watts, C. Camerer, and T. Senator for their encouragement and for many helpful suggestions regarding this research. We are also grateful to H. Ni for his contributions to the early stages of the project. The work was supported by a grant from the Defense Advanced Research Projects Agency.