Social Insect Networks

Author(s): Jennifer H. Fewell

Source: Science, New Series, Vol. 301, No. 5641 (Sep. 26, 2003), pp. 1867-1870
Published by: American Association for the Advancement of Science
Stable URL: http://www.jstor.org/stable/3835174
Accessed: 28-09-2017 16:11 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3835174?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to

digitize, preserve and extend access to Science

This content downloaded from 143.106.201.11 on Thu, 28 Sep 2017 16:11:06 UTC
All use subject to http://about.jstor.org/terms
 NETWORKS IN BIOLOGY

I
tion processing task such as filtering out spuri- engineering, because prototypes often do not 5. Y. Setty, A. E. Mayo, M. G. Surette, U. Alon, Proc. Natl.
Acad. Sci. U.S.A. 100, 7702 (2003).
ous input fluctuation (25), generating temporal work and need to be understood in order to
6. D. Thieffry, A. M. Huerta, E. Perez-Rueda, J. Collado-
programs of expression (3, 25) or accelerating correct their design. The program of molecular
Vides, Bioessays 20, 433 (1998).
the throughput of the network (2, 26). Recently, biology is reverse-engineering on a grand scale. 7. T. I. Lee et al., Science 298, 799 (2002).
the same network motifs were also found in the Reverse engineering a nonmodular network of 8. E. H. Davidson et al., Science 295, 1669 (2002).
transcription network of yeast (7, 27). It is im- a few thousand components and their nonlinear 9. S. H. Strogatz, Nature 410, 268 (2001).

portant to stress that the similarity in circuit interactions is impossible (exponentially hard 10. L. H. Hartwell, J. J. Hopfield, S. Leibler, A. W. Murray,
Nature 402, C47 (1999).
structure does not necessarily stem from circuit with the number of nodes). However, the spe-
11. J. Ihmels et al., Nature Genet. 31, 370 (Aug, 2002).
duplication. Evolution, by constant tinkering, cial features of biological networks discussed
12. E. Ravasz, A. L. Somera, D. A. Mongru, Z. N. Oltvai,
appears to converge again and again on these here give hope that biological networks are A.-L. Barabasi, Science 297, 1551 (2002).
circuit patterns in different nonhomologous sys- structures that human beings can understand. 13. C. R. Myers, arXiv: cond-mat/0305575 (2003).

tems (25, 27, 28), presumably because they Modularity, for example, is at the root of the 14. J. J. Hopfield, Proc. Natl. Acad. Sci. US.A. 79, 2554
( 1 982).
carry out key functions (see Perspective (29) success of gene functional assignment by ex-
15. D. Bray,J. Theor. Biol. 143, 215 (1990).
STKE). Network motifs can be detected by pression correlations (11, 34). Robustness to
16. H. Lipson, J. B. Pollack, N. P. Suh, Evolution 56, 1549
algorithms that compare the patterns found in component tolerances limits the range of pos- (2002).
the biological network to those found in suitably sible circuits that function on paper to only a 17. J. Gerhart, M. W. Kirschner, Cells, Embryos, and Evo-
randomized networks (25, 27). This is analo- few designs that can work in the cell. This can lution: Toward a Cellular and Developmental Under-
standing of Phenotypic Variation and Evolutionary
gous to detection of sequence motifs as help theorists to home in on the correct design
Adaptability (Blackwell Science Inc, Oxford, 1997).
recurring sequences that are very rare in with limited data (21-23). Network motifs de-
18. C. H. Waddington, Nature 150, 563 (1942).
random sequences. fine the few basic patterns that recur in a net- 19. H. Kacser, J. A. Burns, Symp. Soc. Exp. Biol. 32, 65
Network motifs are likely to be also work and, in principle, can provide specific ( 1 973).

found on the level of protein signaling net- experimental guidelines to detennine whether 20. M. Savageau, Nature 229, 542 (1971).

works (30). Once a dictionary of network they exist in a given system (25). Ihese con- 21. N. Barkai, S. Leibler, Nature 387, 913 (1997).
22. U. Alon, M. G. Surette, N. Barkai, S. Leibler, Nature
motifs and their functions is established, cepts, together with the current technological
397, 168 (1999).
one could envision researchers detecting revolution in biology, may eventually allow
23. A. Eldar et al., Nature 419, 304 (2002).
network motifs in new networks just as charactenzation and urlderstanding of cell-wide 24. D. Fell, Understanding the Control of Metabolism
protein domains are currently detected in networks, with great benefit to medicine. The (Portland Press, London, 1997).

the sequences of new genes. Finding a se- similarity between the creations of tinkerer and 25. S. S. Shen-Orr, R. Milo, S. Mangan, U. Alon, Nature
Genet. 31, 64 (2002).
quence motif (e.g., a kinase domain) in a engineer also raises a fimdamental scientific
26. N. Rosenfeld, M. B. Elowitz, U. Alon,J. Mol. Biol. 323,
new protein sheds light on its biochemical challenge: understanding the laws of nature that 785 (2002).
function; similarly, finding a network motif uxiite evolved and designed systems. 27. R. Milo et al., Science 298, 824 (2002).

in a new network may help explain what 28. G. C. Conant, A. Wagner, Nature Genet. 34, 264 (2003).
29. A. Wagner, Sci. STKE 2003, pe41 (2003).
systems-level function the network per- References and Notes
30. C. V. Rao, A. P. Arkin, Annu. Rev. Biomed. Eng. 3, 391
forms, and how it performs it. 1. F. Jacob, Science 196, 1 161 (1977).
(2001 ).
2. M. Savageau, Biochemical Systems Analysis: A Study
Will a complete description of the biological 31. M. E. Csete, J. C. Doyle, Science 295, 1664 (2002).
of Function and Design in Molecular Biology (Addi-
networks of an entire cell ever be available? 32. A. Arkin, P. Shen, J. Ross, Science 277, 1275 (1997).
son-Wesley, Reading, MA, 1976).
3. M. Ronen, R. Rosenberg, B. I. Shraiman, U. Alon, Proc. 33. T. S. Gardner, D. di Bernardo, D. Lorenz, J. J. Collins,
The task of mapping an unknown network is
Natl. Acad. Sci. U.S.A. 99, 10555 (2002). Science 301,102 (2003).
known as reverse-engineering (3, 3]-33). 4. C. H. Yuh, H. Bolouri, E. H. Davidson, Science 279, 34. M. B. Eisen, P. T. Spellman, P. O. Brown, D Botstein,
Much of engineering is actually reverse- 1896 (1998). Proc. Natl. Acad. Sci. U.S.A. 95, 14863 (1998).

V I E W P O I N T

Social Insect Networks

Jennifer H. Fewell

Social insect colonies have many of the properties of adaptive networks. The simple How can viewing insect societies as net-
rules governing how local interactions among individuals translate into group be- works shape our understanding of social orga-
haviors are found across social groups, giving social insects the potential to have a nization and evolution? First, they have become
profound impact on our understanding of the interplay between network dynamics one of the central model systems for exploring
and social evolution. self-organization: the process by which interac-
tions occurring locally between individuals
The formal exploration of social insect col- groups generally) have key network attributes produce group-level attributes. Self-organi-
onies as networks is in its infancy. Howev- that appear consistently in complex biological zation in a social insect colony produces
er, social insects such as wasps, ants, and systems, from molecules through ecosystems; emergent properties: social phenotypes that
honeybees provide a powerful system for these include nonrandom systems of connectiv- are greater than a simple summation of
examining how network dynamics contrib- ity and the self-organization of group-level individual worker behaviors (2). The basic
ute to the evolution of complex biological phenotypes (1-3). Colonies exhibit multi- rules generating these dynamics are broad-
systems. Social insect colonies (and social ple levels of organization, yet it is still ly applicable across taxa whose members
possible to track individuals, making these show social behavior, and they produce
societies more accessible to experimen- ubiquitous patterns of social organization,
School of Life Sciences, Arizona State University,
Tempe, AZ 85287-1501, USA. E-mail: j.fewell@asu. tal manipulation than many other com- including mass action responses, division
edu plex systems. of labor, and social hierarchies (2, 4).

www.sciencemag.org SCIENCE VOL 301 26 SEPTEMBER 2003 1867

This content downloaded from 143.106.201.11 on Thu, 28 Sep 2017 16:11:06 UTC
All use subject to http://about.jstor.org/terms
 NETWORKS IN BIOLOGY

Second, the social insects provide an oppor- more nestmates than do others. The most obvi- the scout who discovers a foraging trail can
tunity to explore how behavior evolves within ous of these is the queen, who, in honeybees, completely block the retrieval of a resource, yet
complex systems. This has led to a shift in secrets a pheromone that represses reproduction removing the recruits who follow the scout has
focus from variation among individuals to how in workers and maintains colony cohesion. little effect on overall foraging (15). Social
interactions among individuals and groups Queen pheromone is transmitted to workers as insect networks are similar in this way to other
shape that variation. Most of the well-studied they groom her, then is rapidly transmitted biological networks, from food webs with key-
social insects are eusocial (only one or a few through the hive via trophallaxis and deposits stone species (16) to metabolic pathways, in
individuals in the colony reproduce), and the on nest wax (11). Key individuals are also which a few key molecules are involved in
colony is considered an adaptive unit made up present within worker task groups, where they most reactions (9, 17).
of related individuals (5). Because of this, we stimulate performance of a task or provide a With these global attributes in place, how
are comfortable in relating group dynamics to central point around which performance is or- does information transfer within a social col-
fitness effects at both the individual and group ganized (12). For example, foraging task ony actually occur? Unfortunately, we do not
levels. However, multilevel selection acts on groups often include scouts or dancers. They yet have enough empirical data to answer this
social insect colonies, not just because their communicate most of the information about question well. Models to date have explored
members are highly related but also because resource location and availability and, in ants, networks in the context of task regulation: the
they are densely connected networks. This often maintain the cohesion of groups of re- amount of effort by individuals or groups that
emerging view of social groups as networks cruits that go out to forage (10, 12, 13). is allocated to different tasks. One approach
contributes to a growing awareness has been to consider the colony as
of how the fitness of individuals a regular network (9), in which
and groups is generated interactive- individuals performing the same
ly across levels of biological orga- task form clusters of high connec-
nization (3, 6, 7). tivity, with weaker links across
To explore the relationships be- \ tasks (18, 19). In a model of re-
tween complexity and selection in -eeding \ cruitment to alternate resource
social systems, we first need to de- =/ pathways, Bonabeau et al. (19)
scribe the social group as a net- R
showed that colonies can balance
work. A network is simplistically a / I I
efficient utilization of a single re-
system of interacting elements, or /;;; ;7\ / l l source with flexible allocation
> F | | F
nodes, that communicate with each \ Dancing across resources by a mixed strat-
1 1 1

other [see (8, 9) in this issue]. So- n J egy of within-cluster information

rT | |

\' I I _

cial insect colonies are dense net- transfer coupled with global in-
F " | v
works in which individuals have formation transfer across clusters.
Pollen:
multiple points of contact (1, 10). An important finding of this mod-
0 Resources
As dense networks, colonies dis- el [and the Pacala et al. model
tribute information rapidly, allow- Fig. 1. The network pathways modulating pollen foraging in a I honeybee (18) on which it was based] is the
ing them to respond flexibly and colony (des veloped with T. Taylor). Nodes are the tasks linked to pollen importance of cross-cluster links
ectors are the individuals transmitting information: F F, forager; in maintaining flexibilitY for
efficiently to the dynamic environ- foraglng; ve
B, brood; R, recruit. Foragers returning with polle n receive .
ment in which they live An ex- in'fornnati;on n about pollen storage levels as they place pollen l oads into movlng lndlvlduals from one task
treme example is the alarm re- cells. The 4 amount of stored pollen is negative feedback f or pollen or cluster to another.
sponse of African honeybees, in foraging. Pc ollen is removed from cells by nurse bees, who feed it first to The assumptions of the Bona-
which an initial release of alarm developing brood and give excess to pollen foragers. Receivi ng pollen beau et al. model (19) fit well
pheromone by a few guards cas- from nurse zs is negative feedback for foraging. Foragers als o receive into the context of trail selection
n about pollen stores from brood, who produce
cades within a minute to stinging Informatlon
pheromone X when they are not fed; brood care reduces hung Xer levegis during foraging, where the sig-
responses by thousands of bees. Information n on pollen availability and location is transmitted by pollen nals are well defined. However,
Like many biological systems, dancers. Da 3ncing elicits recruitment to foraging by workers no wt actively expanding the model more wide-
social insect colonies are also dis- engaged in [ foraging (2, 20-22). ly to multiple tasks has been
tributed networks (2). Although the problematic. One reason is that
colony generally has a single queen, she does The importance of these rare individuals contacts between workers are extremely flu-
not centrally control colony function. Instead, makes it likely that for many functions the id. Connections between workers in a social
workers make decisions based on local infor- colony network becomes scale-free, which insect colony are ephemeral, and signals
mation and perform behaviors in parallel (10). means that variation in connectivity is best de- themselves can outlive connections. Signal
This is the case, to some degree, even for scribed by a power law rather than a Poisson systems- are also highly diverse in informa-
hierarchical systems such as the wasp network, distribution (14). This is important to colony tion content and include large-scale signals,
where the queen controls the reproductive out- resiliency, because it means that the loss of any such as alarm pheromones, that target the
put of the colony but does not individually ofthe vast majority of workers would have little colony globally (10).
direct many aspects of day-to-day colony func- effect. In contrast, removing nodes within a Social insect networks are traditionally
tion. We lack sufficient data to accurately char- randomly distributed network can quickly frag- modeled with workers as nodes. However,
acterize the connections that occur between any ment the system. Although scale-free networks because worker interactions are so fluid, we
two individuals within a colony, much less the are buffered from the effects of random loss, the can alternatively map the system from the
connections across the society. However, it is removal of key nodes can severely disrupt the perspective of treating tasks as nodes and
clear that connections among nestmates are system (1, 14). The colony has long-term individual workers as connectors (symbolic
mechanisms to replace any element, including
nonrandomly distributed for many, if not most, dynamics). Figure 1 describes such a map for
colony functions. A few key individuals, or the queen, but the removal of key individuals the short-term modulation of pollen foraging
hubs, distribute information (connect) to many does have immediate disruptive effects. Loss of in honeybees. It is clear from this map that

1 868 26 SEPTEMBER 2003 VOL 301 SCIENCE www.sciencemag.org

This content downloaded from 143.106.201.11 on Thu, 28 Sep 2017 16:11:06 UTC
All use subject to http://about.jstor.org/terms
 NETWORKS IN BIOLOGY

I
cross-task connections are the primary path- environment at points of choice between tasks. Used dishes pile up in the sink, producing
ways for regulating pollen foraging. Pollen alternate trails is changed. Ants reaching a continuously increasing stimulus. The dishes
foraging is homeostatically regulated around these points preferentially choose the trail go unnoticed until the threshold of the one most
pollen storage levels and is positively regu- with more pheromone and add to it, creat- sensitive to them is met, and he or she washes
lated around brood, for which it is the main ing a positive feedback loop. Meanwhile, them. This removes the dishes as a stimulus,
nutrient source (20). Pollen foragers collect the pheromone marks on the alternate trail further reducing the likelihood that the other
most of their information about colony pollen decay. As more foragers repeat this pro- group members will ever wash them. The result
need and/or intake either indirectly via cess, one trail becomes the primary and is a dishwashing specialist (much to his/her
changes in pollen stores, from nurse bees who often the only route (2, 24). These simple dismay), and a set of nondishwashers. Similar
feed foragers when excess pollen is available, rules underlie trail-making in multiple ant interactions across other chores, from cleaning
or from brood who emit hunger signals (21, species (2). Similar rules describe conver- the bathroom to taking out the garbage, gener-
22). The map is not consistent with the as- gent group behaviors in other social spe- ate a division of labor for the household.
sumption of high within-cluster connectivity, cies, such as migrating social spiders who The realization that individuals within a
but it does support the assertion that con- choose a direction of travel based on the social group are linked as a network is
nections across tasks are important to allo- accumulation of draglines from others in important to our understanding of how se-
cation (18, 19). They may, in fact, be the the group (25). lection acts on sociality. The fitness of
primary links modulating task regulation The minimal rule set for divergence can every individual in the group is produced in
globally. If so, information flow in social be condensed to two components: (i) per- part as a result of their interactions with
insect colonies has an important similarity formance of a behavior by one individual other group members. The emergence of
to that in human social networks, where reduces the probability that others will per- collective behaviors via self-organization
weak ties across social clusters play an form the same behavior, and (ii) stimulus also produces phenotypes at the colony lev-
important role in regulating society as a levels for the behavior increase in the ab- el that are themselves subject to selection
large-scale network (23). sence of performance. Most divergence (7). These interactions set the stage for
Although the complexities of the whole- models also include a positive feedback multilevel selection (32). Network-level
colony network have not yet been well de- loop, in which performance of the behavior properties, including group size, connectiv-
scribed, large strides have been made in the increases the probability that the individual ity, and even variation in individual respon-
analysis of how local interactions within the will perform the behavior again. This self- siveness to signals can all shape the adap-
network affect global colony dynamics. As reinforcement generates divergence even tive function of the social group (18, 28).
dense networks, social insect colonies have a with initially small random differences in As an example, as described above, the
high potential for the emergence of large-scale behavior and produces a faster and more emergence of division of labor is based in
phenomena via self-organization (1). Self-orga- stable system of divergence (26). However, part on intrinsic variation in worker re-
nization pervades all aspects of colony function, divergence can emerge in the absence of sponse thresholds. Honeybee colonies with
including foraging, nest defense, resource stor- self-reinforcement if individuals initially more diversity in worker thresholds for for-
age, nest construction, site selection, thermo- differ intrinsically in their response thresh- aging are able to respond better to changes
regulation, and division of labor (2). olds: the stimulus level at which they re- in the availability and need for resources.
The growing body of theoretical and empir- spond by performing a behavior (27, 28). This diversity is generated by the extreme
ical work on self-organization is one of the This rule set forms the basis for the re- polyandry of honeybee queens, who mate
more important contributions of social insect sponse threshold models of division of labor with a dozen or more males (22).
research to understanding biocomplexity (2). (27). These models begin with the initial Network interactions also have a pro-
What is perhaps most important about self- assumption that individuals perform a task found influence on individual behavior and
organization in social insects is that it is not when environmental stimuli reach a level that fitness. The fitness of each individual in a
based on derived characteristics unique to the matches the individual's threshold for re- social group is dependent on the pheno-
taxon. Instead, it is driven by a limited set of sponse. That individual performs the task; in types of the other group members (7); they
nonlinear dynamics that should occur across doing so, she reduces the stimulus levels are each other's social environments. These
social systems, from insects to humans (2, 4). encountered by others and thus reduces their reciprocal fitness effects are generated by
As an example, a majority of the emergent probability of performing the task also. Em- nonlinear interactions within the social net-
components of social behavior can be catego- pirical tests on solitary bees and on ant work. In some systems, self-organization
rized as "convergent," in which individuals queens during colony founding have shown can actually generate conflicting fitness ef-
become behaviorally more similar, or "diver- that division of labor can emerge even with- fects at the individual and group levels. For
gent," in which the behavior of one individual out a history of direct selection (29). When ant queens, when division of labor sponta-
reduces the likelihood that the second individ- normally solitary ant queens are forced into neously emerges from small initial differ-
ual will perform the same behavior. artificial social groups, one individual takes ences in behavior (29), it produces associ-
The minimal components (or minimal over the task of excavation, whereas the other ated fitness disparities, because the queen
rule set) for convergence can be condensed individual remains in the nest and tends who takes over the task of nest excavation
to (i) a positive stimulus for the behavior as brood. The dynamics of this division of labor is more likely to die. Whether an individual
a result of its performance; (ii) amplifica- fit well with the predictions of the response becomes the excavator, and suffers the as-
tion of the stimulus through successive it- threshold model. sociated fitness consequences, depends on
erations; and (iii) a decay component, so Similar patterns of divergence occur across which group they land in and the thresholds
that signals and cues must be regenerated. other social taxa. Social hierarchies within of everyone in that group.
A beautiful example of behavioral conver- bumblebees and primates can be modeled by a What should be done next in the explora-
gence via these minimal rules is found in similar minimal rule set for divergence, coupled tion of social groups as networks? We need to
the trail marking system of the Argentine with reinforcement (30, 31). Division of labor expand our models from elegant descriptions
ant Linepithema humile. Workers traveling also appears frequently within social species, of single behaviors to incorporate the more
to and from resources lay a pheromonal including humans. As an example, we can complex dynamics of the group as a whole.
. .

trail. Each time a trail is laid, the local lmagme an apartment where housemates share We also need to test those models empirically

www.sciencemag.org SCIENCE VOL 301 26 SEPTEMBER 2003 1869

This content downloaded from 143.106.201.11 on Thu, 28 Sep 2017 16:11:06 UTC
All use subject to http://about.jstor.org/terms
 I NETWORKS IN BIOLOGY

on a wider range of social systems. Finally, to 7. A. J. Moore, E. D. Brodie, J. B. Wolf, Evolution 51, 1352 19. E. Bonabeau, G. Theraulaz, J.-L. Deneubourg,J. Theor.
( 1 997). Biol. 195, 157 (1998).
understand the evolutionary significance of
8. U. Alon, Science 301, 1866 (2003). 20. J. H. Fewell, M. Winston, Behav. Ecol. Sociobiol. 30,
network dynamics, we must explicitly mea- 9. D. Bray, Science 301, 1864 (2003). 387 (1992).
sure their fitness effects on the social group 10. B. Holidobler, E. O. Wilson, The Ants (BeEknap Press of 21. S. Camazine, Behav. Ecol. Sociobiol. 32, 265
(1 993).
(7). This interplay between network dynam- Harvard Univ. Press, Cambridge, MA, 1990).
11. K. Naumann, M. Winston, K. Slessor, G. Prestwich, F. 22. R. E. Page, J. Erber, Naturwissenschaften 89, 91 (2002).
ics and selection is just beginning to be ex- 23. M. Granovetter, Am. J. Sociol. 78, 1360 (1973).
Webster, Behav. Ecol. Sociobiol. 29, 321 (1991 ).
plored, and social insects have the potential 12. S. K. Robson, J. F. A. Traniello, in Information Process-
24. S. Goss, S. Aron, J. L. Deneubourg, J. M. Pasteels,

to be on the leading edge. Naturwissenschaften 76, 579 (1989).

ing in Social Insects, C. Detrain, J. L. Deneubourg, J. M.
Pasteels, Eds. (Birkhauser, Basel, Switzerland, 1999), 25. F. Saffre, R. Furey, B. Krafft, J. L. Deneubourg,J. Theor.

pp. 239-259. Biol. 198, 507 (1999).

References
13. T. D. Seeley, S. Camazine, J. Sneyd, Behav. Ecol. 26. G. Theraulaz, E. Bonabeau, J. L. Deneubourg, Proc. R.
1. A.-L. Barabasi, Linked: The New Science of Networks
Sociobiol. 28, 277 (1991). Soc. London Ser. B 265, 327 (1998).
(Perseus, Cambridge, MA, 2002).
14. R. Albert, A.-L. Barabasi, Rev. Mod. Phys. 74, 47 27. S. N. Beshers, J. H. Fewell, Annu. Rev. Entomol. 46,
2. S. Camazine et al., Self-Organization in Biological
(2002). 413 (2001).
Systems (Princeton Univ. Press, Princeton, NJ,
15. D. M. Gordon, Am. Nat. 159, 509 (2002). 28. R. E. Page, S. D. Mitchell, Apidologie 29, 171
2001).
16. R. Sole, J. Montoya, Santa Fe Working Paper 00-11- (1 998).
3. R. Sole, B. C. Goodwin, Signs of Life: How Complexity
060 (2000). 29. J. H. Fewell, R. E. Page, Evol. Ecol. Res. 1, 537
Pervades Biology (Basic Books, New York, 2000).
4. C. K. Hemelrijk, Ethology 108, 655 (2002). 17. H. Jeong, B. Tombor, R. Albert, Z. Oltvai, A.-L. Bara- (1 999).

5. T. D. Seeley, Am. Nat. 150, S22 (1997). basi, Nature 407, 651 (2000). 30. P. Hogeweg, B. Hesper, Behav. Ecol. Sociobiol. 12,

6. D. S. Wilson, L. A. Dugatkin, Am. Nat. 149, 336 18. S. W. Pacala, D. M. Gordon, H. C. J. Godfray, Evol. 271 (1983).
(1 997). Ecol. 10, 127 (1996). 31. C. K. Hemelrijk, Biol. Bull. 202, 283 (2002).
32. D. S. Wilson, Am. Nat. 150, S 1 (1997).

R E V I E W

Communication in Neuronal Networks

Simon B. Laughlinl and Terrence J. Sejnowski23*

Brains perform with remarkable efficiency, are capable of prodigious computation, wiring
wiring costs.
costs.We
Wethen
thenexamine
examine
a constraint
a constraint
that
that
and are marvels of communication. We are beginning to understand some of the impinges on all aspects of neural function but
geometric, biophysical, and energy constraints that have governed the evolution of has only recently become apparent energy
cortical networks. To operate efficiently within these constraints, nature has opti- consumption. Next we look at energy-efficient
mized the structure and function of cortical networks with design principles similar neural codes that reduce signal traffic by ex-
to those used in electronic networks. The brain also exploits the adaptability of ploiting the relationships that govern the repre-
biological systems to reconfigure in response to changing needs. sentational capacity of neurons. We end with a
brief discussion on how synaptic plasticity may
Neuronal networks have been extensively stud- synapses and can combine and process synaptic reconfigure the cortical network on a wide
ied as computational systems, but they also inputs, both linearly and nonlinearly, to imple- range of time scales.
serve as communications networks in transfer- ment a rich repertoire of operations that process
nng large amounts of information between information (1). Neurons can also establish and Geometricat and Biophysical
brain areas. Recent work suggests that their change their connections and vary their signal- Constraints on Wiring
structure and fimction are governed by basic ing properties according to a variety of rules. Reducing the size of an organ, such as the
principles of resource allocation and constraint Because many of these changes are driven by brain, while maintaining adequate function is
minimization, and that some of these principles spatial and temporal patterns of neural signals, usually beneficial. A smaller brain requires
are shared with human-made electronic devices neuronal netsvorks can adapt to circumstances, fewer materials and less energy for construc-
and communications networks. The discovety self-assemble, autocalibrate, and store informa- tion and maintenance, lighter skeletal ele-
that neuronal networks follow simple design tion by changing their properties according ments and muscles for support, and less
rules resembling those found in other networks to experience. energy for carriage. The size of a nervous
is striking because nervous systems have many The simple design rules improve efficien- system can be reduced by reducing the num-
unique properties. cy by reducing (and in some cases minimiz- ber of neurons required for adequate function,
To generate complicated patterns of ing) the resources required to implement a by reducing the average size of neurons, or by
behavior, nervous systems have evolved prodi- given task. It should come as no surprise that laying out neurons so as to reduce the lengths
gious abilities to process information. Evolution brains have evolved to operate efficiently. of their connections. The design principles
has made use of the rich molecular repertoire, Economy and efficiency are guiding princi- governing economical layout have received
versatility, and adaptability of cells. Neurons ples in physiology that explain, for example, the most attention.
can receive and deliver signals at up to 105 the way in which the lungs, the circulation, Just like the wires connecting components
and the mitochondria are matched and co- in electronic chips, the connections between
regulated to supply energy to muscles (2). To neurons occupy a substantial fraction of the
1Department of Zoology, University of Cambridge, identify and explain efficient design, it is total volume, and the wires (axons and den-
Downing Street, Cambridge CB2 3EJ, UK. 2Howard necessary to derive and apply the structural drites) are expensive to operate because they
Hughes Medical Institute, Salk Institute for Biological
and physicochemical relationships that con- dissipate energy during signaling. Nature has an
Studies, La Jolla, CA 92037, USA. 3Division of Biolog-
ical Sciences, University of California, San Diego, La nect resource use to performance. We con- important advantage over electronic circuits be-
Jolla, CA 92093, USA. sider first a number of studies of the geomet- cause components are connected by wires in
To whom correspondence should be addressed. E- rical constraints on packing and wiring that three-dimensional (3D) space, whereas even the
mail: terry@salk.edu show that the brain is organized to reduce most advanced VLSI (very large scale integra-

1870 26 SEPTEMBER 2003 VOL 301 SCIENCE www.sciencemag.org

This content downloaded from 143.106.201.11 on Thu, 28 Sep 2017 16:11:06 UTC
All use subject to http://about.jstor.org/terms