visual cognition in social...

EN56CH22-Giurfa ARI 14 October 2010 14:25

Visual Cognitionin Social InsectsAurore Avargues-Weber,1,2 Nina Deisig,1,2

and Martin Giurfa1,2

1Centre de Recherches sur la Cognition Animale, Universite de Toulouse,F-31062 Toulouse Cedex 9, France2Centre de Recherches sur la Cognition Animale, CNRS, F-31062 Toulouse Cedex 9,France; email: [email protected]

Annu. Rev. Entomol. 2011. 56:423–43

First published online as a Review in Advance onSeptember 24, 2010

The Annual Review of Entomology is online atento.annualreviews.org

This article’s doi:10.1146/annurev-ento-120709-144855

Copyright c© 2011 by Annual Reviews.All rights reserved

0066-4170/11/0107-0423$20.00

Key Words

learning, nonelemental learning, Hymenoptera, vision

Abstract

Visual learning admits different levels of complexity, from the formationof a simple associative link between a visual stimulus and its outcome,to more sophisticated performances, such as object categorization orrules learning, that allow flexible responses beyond simple forms oflearning. Not surprisingly, higher-order forms of visual learning havebeen studied primarily in vertebrates with larger brains, while simplevisual learning has been the focus in animals with small brains suchas insects. This dichotomy has recently changed as studies on visuallearning in social insects have shown that these animals can master ex-tremely sophisticated tasks. Here we review a spectrum of visual learningforms in social insects, from color and pattern learning, visual atten-tion, and top-down image recognition, to interindividual recognition,conditional discrimination, category learning, and rule extraction. Weanalyze the necessity and sufficiency of simple associations to accountfor complex visual learning in Hymenoptera and discuss possible neuralmechanisms underlying these visual performances.

423

Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


INTRODUCTION

Visual learning refers to an individual’s capac-ity of acquiring experience-based informationpertaining to visual stimuli so that adaptiveresponses can be produced when viewing suchstimuli again. This capacity admits differentlevels of complexity, from the establishmentof a simple associative link connecting a visualstimulus (e.g., a specific color) and its outcome(e.g., a reward or a punishment) to moresophisticated performances such as learning tocategorize distinct objects (e.g., animal versusnonanimal) or apprehending abstract rulesapplicable to unknown visual objects (e.g.,“larger than,” “on top of,” or “inside of”).

The first situation, the establishment ofunivocal, unambiguous links between a visualtarget and its outcome, constitutes a case of el-emental learning. For instance, what is learnedfor a color is valid only for that color and notfor different ones. In contrast, learning aboutcategories or rules constitutes a case of nonele-mental learning, as appropriate responses canbe transferred to unknown stimuli for whichthe subject has no personal experience, as longas the stimuli satisfy the learned category orrule. In these cases, the subject’s responseis flexible and relatively independent of thephysical nature of the stimuli considered.

Social Hymenoptera, particularly bees (Apisspp. and Bombus spp.), ants, and wasps (severalgenera), which are at the center of this article,are interesting models for the study of visuallearning because in their natural context theyhave to solve a diversity of visual problems ofvarying complexity. For instance, these insectslearn and memorize the local cues characteriz-ing the places of interest, which are essentiallythe hive and the food sources (27, 28, 68, 71, 85,112). Honey bees (Apis mellifera), and to a mi-nor extent bumble bees (Bombus terrestris), areflower constant, which means that they forageon a unique floral species as long as it offersa profitable nectar and/or pollen reward (14,40, 46). This capacity is based partly on visualcues provided by flowers such as colors or pat-terns. Learning and memorizing the visual cues

of the exploited flower through their associationwith a nectar and/or pollen reward are what al-low a bee forager to track a particular flowerspecies in the field (68). Similarly, learning abil-ities for landmark constellations, complex natu-ral scenes, and celestial cues used in navigation(e.g., azimuthal position of the sun, polarizedlight pattern of the blue sky) ensure a safe re-turn to the nest and enhance foraging efficiency(15, 16, 24, 74, 103).

Visual capacities are highly developed in so-cial Hymenoptera. Bees, wasps, and some antspecies see the world in color (3, 8, 9, 12, 21, 60,61, 63, 70), perceive shapes and patterns (23, 33,60, 61, 84, 102), and resolve movements with ahigh temporal resolution (86). One of the rea-sons why bees, ants, and wasps constitute anattractive model for the study of visual learningresides precisely in the existence of controlledexperimental methods for the study of these ca-pacities in the laboratory.

VISUAL CONDITIONINGOF BEES

Visual conditioning of honey bees (98) has un-covered the perceptual capabilities of these in-sects and has been used to this end for nearlya century. This protocol exploits the fact thatfree-flying honey bees learn visual cues suchas colors, shapes, patterns, depth, and motioncontrast, among others (33, 34, 59, 98, 102),when these cues are presented together with areward of sucrose solution. Each bee is individ-ually marked by means of a color spot on thethorax or the abdomen so that individual per-formances can be recorded. The marked bee isgenerally displaced by the experimenter towardthe experimental site, where it is rewarded withsucrose solution to promote its regular return.Such pretraining is performed without present-ing the training stimuli in order to avoid un-controlled learning. When the bee starts visit-ing the experimental place actively (i.e., withoutbeing displaced by the experimenter), the train-ing stimuli are presented and the choice of theappropriate visual target rewarded with sucrose.This basic protocol has been used to study visual

424 Avargues-Weber · Deisig · Giurfa

Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


learning in other bee species such as bumblebees (24, 63, 82), leaf-cutter bees (9), stinglessbees (6, 72), and wasps (3, 60).

In studies on pattern vision, the plane ofstimulus presentation is extremely important.Early works (47–49, 98) presented stimuli onthe horizontal plane of an experimental table,and the position of the stimuli were variedto verify that bees were indeed choosing arewarded pattern rather than using positioninformation. In this situation it is difficult todetermine which specific information con-tained in the patterns is used by the insects tomake their choices because they can approachthe patterns from any possible direction. Later,vertical presentation was preferred because itconstrained the approach direction to the pat-terns (99), thereby forcing a frontal approachand perception. In this way, it was possible todetermine whether bees are sensitive to differ-ent pattern cues such as orientation, bilateraland radial symmetry, and center of gravity. Inall cases, insects have to be trained and testedindividually to achieve a precise control of theexperience of each subject. It is also importantto control the distance at which a choice is madebecause visual orientation and choice are me-diated by different visual cues at different dis-tances or angles subtended by the target (33, 34,37). The time between visits to the experimen-tal place also has to be recorded, as it reflects theappetitive motivation of the bee (75) and thusits motivation to learn. The associations builtin this context can be classical, operant, or both;i.e., they may link a visual stimulus (conditionedstimulus, or CS) and a sucrose reward (un-conditioned stimulus, or US), the response ofthe animal (e.g., landing) and the US, or both,respectively. The experimental framework isnevertheless mainly operant, because the bee’sbehavior is a determinant for obtaining or notobtaining the sucrose reinforcement.

Visual conditioning of freely flying insectsdoes not allow visual learning to be studied atthe cellular level. Because bees freely fly duringthe experiment, studying neural activity invisual centers in the brain simultaneously re-mains impossible thus far. Recently, however,

a protocol for visual conditioning of harnessedbees has been developed (50, 51). This proto-col, based on pioneer studies by Kuwabara (56),consists of training a harnessed bee to extendits proboscis to colors (50) or motion cues(51) paired with sucrose solution. Hungry beesreflexively extend the proboscis when theirantennae are touched with sucrose solution, theequivalent of a nectar reward. In this protocol,colors or patterns are paired with a sucrose re-ward to create a Pavlovian association in whichthe visual stimuli are the CS and sucrose is theUS. Learning, however, is poor in this protocol.It takes 2 days to reach an acquisition level thatis approximately 40%, and this is possible onlyif bees have their antennae previously cut. Thereasons for the apparent interference of theantennae on visual learning remain unknown.Cutting the antennae may affect the generalmotivation of the bee so that the sucrosereward is not as attractive as expected by theexperimenter (19). Improving this protocol isa priority for future research on visual learningas it will allow combining behavioral quantifi-cation with access to the nervous system.

VISUAL CONDITIONINGOF ANTS

Visual learning in ants has been studied mostlyin the context of insect navigation. Experimentswith the Sahara desert ant, Cataglyphis bicolor,have increased our understanding of navigationstrategies based on celestial cues and landmarks(see Reference 103 for a review). In other antspecies such as, Melophorus bagoti (10, 74), Gi-gantiops destructor (64), and Formica rufa (39, 45),similar questions have been investigated, thusplacing importance on determining how antsuse visual cues to negotiate their spatial envi-ronment. These works are not examined herebecause they rarely focus on the learning pro-cess itself, which is the main framework of ourreview. Although learning of visual cues under-lies navigation processes studied in these ants,learning curves or memory retention tests aregenerally absent from these works, thus makingdifficult any analysis in terms of the associative

www.annualreviews.org • Visual Cognition in Social Insects 425

Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


links established during spatial learning in ants.It is therefore difficult to determine what is andwhat is not elemental in these performances.

Nevertheless, a recent work on Cataglyphisaenescens and Formica cunicularia has used anexperimental design that reproduces basic fea-tures of visual training in bees (8), as it focuseson color learning and discrimination. Theseants were trained in a Y-maze to choose anddiscriminate monochromatic lights of constantintensity associated with a food reward. Usingthis kind of design could help researchers dis-sect in a more controlled way the nature of as-sociative learning in ants.

ATTENTIONAL ANDEXPERIENCE-DEPENDENTMODULATION OF VISUALLEARNING

The first study on bee learning and memorythat used controlled protocols for characteriz-ing individual learning and memory employedcolors as rewarding stimuli (66). Free-flyinghoney bees were trained to choose a rewardedmonochromatic light and were then presentedin dual-choice situations with the rewardedlight versus an alternative color on a horizon-tal plane. This study reported learning curvesfor different wavelengths and showed that, un-der these experimental conditions, bees learnedall wavelengths after few learning trials. Somewavelengths, particularly 413 nm, were learnedfaster than others, requiring only one to threeacquisition trials (66; but see below). This re-sult argued in favor of innate biases in colorlearning, probably reflecting the intrinsic bio-logical relevance of the color signals that arelearned faster (66). Indeed, color-naıve honeybees in their first foraging flight prefer thosecolors that experienced bees learn faster (35),and those colors seem to correspond to floralcolors highly associated with profitable nectarreward (35).

Visual learning, as studied in these color-conditioning experiments, is elemental, as beesare just presented with a single color targetpaired with sucrose solution. It was supposed to

be a fast form of learning (66; see above), com-pared, for instance, to learning of visual pat-terns, which usually takes longer (20 or moretrials). Recent studies on bumble bee and honeybee color learning (21, 29) have neverthelessintroduced a new twist to these conclusions. Itwas long thought that what an animal sees andvisually learns is constrained by its perceptualmachinery with no or less place for experience-dependent modulations of perception. Studieson honey bees (22, 29) and bumble bees (21)have shown that this idea is wrong: In somecases, learning one and the same color may oc-cur after few trials but in other cases it may takemore than 20 trials (Figure 1a). The criticalfeature is how the bees learn the task. Abso-lute conditioning, in which a subject is trainedwith a single color rewarded with sugar water,yields generally fast learning. Differential con-ditioning, in which the same subject learns todiscriminate a rewarded from a nonrewardedcolor, takes more trials, even if the rewardedcolor is the same as in absolute conditioning.When these animals are asked to discriminatecolors in a test, their performance differs dra-matically. Whereas bees trained in differentialconditioning discriminate colors that are verysimilar (Figure 1c), bees trained in absoluteconditioning do not discriminate the same pairof colors (Figure 1b) (21, 29). Similar resultswere obtained in ants trained to discriminatecolors in a Y-maze (8).

Comparable results were obtained in astudy on pattern learning and discriminationby honey bees (32, 89). Whereas differentialconditioning results in a visual recognitionstrategy that uses the cues present in the wholepattern, absolute conditioning results in arecognition strategy that restricts cue samplingmainly to the lower half of the pattern. In otherwords, bees recognize a pattern differently,depending on the kind of learning implicit tothe conditioning task. In both cases (color andpattern learning), however, differential condi-tioning increases the demands imposed on theperceptual system of the bees, which must notonly go where a rewarded stimulus is presented(absolute conditioning) but also discriminate


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


ed

1.0

a c

0.8

0.6

Co

rre

ct c

ho

ice

s

Trial

Fruit fly brain

Honey bee brain

Training Test

0.4

0.2

01 2 3 4 5 6 7 8 9 10 11

Absolute conditioning

12 13 14 15 ABS DIFFDsmallTr

DIFFDlarge

memelop

Opticlobe

Antennal lobe

Antennallobe

Opticlobe

safbeb

lo

lo

ca

peml

dl

Mushroombody

Mushroombody

Centralbody

caca

pe

LPR

CB

Differential conditioning

b

ABSTr DIFFDsmall

DIFFDlarge

200 μm

50 μm

Figure 1(a, b, c) Attention-like processes in honey bees. Performance of the free-flying bees trained with colors under absolute and differentialconditioning. (a) Acquisition along 15 trials (mean ± standard error; n = 15 bees for each curve). (b) Tests of the group trained withabsolute conditioning. The red bar depicts the test presenting the trained situation (ABSTr), i.e., the single color that was previouslyrewarded. The white bar depicts the test presenting a novel differential situation (DIFF Dsmall), i.e., the color that was previouslyrewarded versus a new color that was similar to the trained one. The gray bar depicts the test presenting a novel differential situation(DIFF Dlarge), i.e., the color that was previously rewarded versus a new color that was different from the trained one. (c) Tests of thegroup trained with differential conditioning. The blue bar depicts the test presenting the trained situation (DIFF DsmallTr), i.e., thepreviously rewarded and the nonrewarded colors that were similar. The white bar depicts the test presenting just the previouslyrewarded color (ABS). The gray bar depicts the test presenting the previously rewarded color versus a novel color different from therewarded one (DIFF Dlarge). (d ) Three-dimensional reconstruction of the fruit fly (Drosophila melanogaster) brain (courtesy of ArminJenett). (e) Three-dimensional reconstruction of a honey bee brain. Abbreviations: me, medulla; lo, lobula; lop, lobula plate; ca, calyx;dl, dorsal lobe; ml, medial lobe; pe, peduncle; CB, central body; eb, ellipsoid body; fb, fan-shaped body; sa, superior arch; LPR, lateralprotocerebrum. Adapted from Reference 29.

it from a nonrewarding alternative (differentialconditioning). The difference in performancetherefore suggests that attentional processesare involved because in differential condition-ing the bee has to focus on the difference and

not on the mere presence of a visual target,thus making learning slower. In any case, theresult goes against the idea that the differencebetween two colors is an immutable propertyconstrained by the visual machinery.


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


When Menzel characterized color learn-ing (66, 67; see above), studies on patternperception were simultaneously performed byWehner (99–101) and others (e.g., 1), continu-ing the tradition started by von Frisch’s students(47–49, 105–107). In contrast to Menzel’s work,these studies focused not on learning but on theperceptual capabilities of bees confronted withpattern discrimination tasks. Visual condition-ing was also used in these and later works onpattern perception (for reviews, see References59, 84, and 102), but a quantification of acquisi-tion curves and/or a characterization of patternmemory was absent from these works. This tra-dition was continued in the 1970s, 1980s, andeven 1990s, as visual learning was used mainlyas a tool to answer questions on visual percep-tion and discrimination. Yet, some experimentsshowed that in pattern vision, as in color vision,what a bee perceives depends on its previous vi-sual experience and of possible attentional pro-cesses. Zhang & Srinivasan (115) showed, forinstance, that the previous visual experience ofa bee can speed up the analysis of the retinalimage when a familiar object or scene is en-countered. They first attempted to train beesto distinguish between a ring and a disk wheneach shape was presented as a textured figureplaced a few centimeters in front of a similarlytextured background. The figures are, in princi-ple, detectable through the relative motion thatoccurs at the figure borders, which are at a dif-ferent distance than the background when beesfly toward the targets. Despite intensive train-ing, the bees were incapable of learning the dif-ference between a ring and a disk, a discrimina-tion that usually poses no problems when thebees experience these stimuli as plain (nontex-tured) shapes. Zhang & Srinivasan then traineda group of bees to this “easy” problem, present-ing a plain black disk and ring a few centime-ters in front of a white background. The beescould, as expected, easily learn the task. Theywere then confronted with the difficult problemof learning the textured disk versus the ring,and this time they immediately solved the dis-crimination. Thus, pretraining with plain stim-

uli primed the pattern recognition system insuch a way that it was able to detect shapes thatotherwise could not be distinguished. It may bethat such pretraining triggers attentional pro-cesses that allow better focusing on the targetsthat have to be discriminated.

Uncovering how attentional processes andlearning modulate visual perception constitutesan unexplored and promising research field.The existence of attentional processes in insectbrains is not far-fetched, and recent researchhas located such processes in precise structuresof the insect brain. In the fruit fly, Drosophilamelanogaster, attention can be demonstratedand characterized at the physiological level (96).A fruit fly fixed stationary within a circulararena, and tracking a visual object (a verti-cal black bar) moving at a constant frequencyaround it, exhibits anticipatory behavior con-sistent with attention for the bar tracked. Suchan anticipatory tracking has a neural correlatein the form of a transient increase in a 20–30 Hz local-field potential recorded in a regionof the brain called the medial protocerebrum(Figure 1d ) (96). In other words, the 20–30 Hzresponse in the fly brain is correlated with tran-sitions to behavioral tracking. This response isnot only anticipatory, but also selective to thestimulus presented, increased by novelty andsalience and reduced when the fly is in a sleep-like state (96). Moreover, the use of mutantsshowed that a subset of neurons of the mush-room bodies, which are a higher-order struc-ture of the insect brain (Figure 1d,e), is re-quired for both the tracking response and the20–30 Hz response (96). This result is consis-tent with the finding that mushroom bodies arerequired for choice behavior of D. melanogasterfacing contradictory visual cues (90). In thiscase, individually tethered flies flying station-ary are trained in a circular arena in which onekind of visual stimulus (say, a T pattern) rep-resents a permitted flight direction, while an-other kind of visual stimulus (say, an inverted Tpattern) represents a forbidden flight directionassociated with a displeasing heat beam on thethorax. Tang & Guo (90) conditioned flies


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


to choose one of two directions in responseto color and shape cues; after the training,flies were tested with contradictory cues. Wild-type flies made a discrete choice that switchedfrom one alternative to the other as the rela-tive salience of color and shape cues graduallychanged, but this ability was greatly diminishedin mutant flies with miniature mushroom bod-ies or with chemically ablated mushroom bod-ies. In other words, mushroom bodies medi-ate the assessment of the relative saliency ofconflicting visual cues (90, 110) and are alsoinvolved in improving the extraction of vi-sual cues after pretraining in D. melanogaster(77). The mushroom bodies of hymenopteransmay play similar roles (Figure 1e), thus favor-ing attention and better problem solving anddiscrimination.

Yet, visual learning and the neural circuitsmediating it are still poorly understood in thefruit fly. The mushroom bodies, which arethe main site for olfactory memories, are notdirectly involved in visual learning because inD. melanogaster, contrary to hymenopterans,there is no direct input from the visual areasof the brain to these structures (108). Recentstudies have succeeded in identifying the pre-cise neuronal substrates of two forms of visualmemory in the D. melanogaster brain outsidethe mushroom bodies (62). Memories forpattern elevation and orientation were retracedto different neuronal layouts of the centralcomplex, a median structure of the insect brain(Figure 1d ). Liu et al. (62) showed that twoneuronal layers of the central complex are re-quired to achieve visual discriminations basedon pattern elevation or orientation. In all cases,visual short-term memory was studied, thusleaving open the question of the localization ofvisual long-term memory. In bees and wasps,the localization of visual memories may differfrom that of D. melanogaster. In contrast to thefruit fly, visual areas of the hymenopteran brainprovide direct input to the mushroom bodies(26), thus making these structures a candidatefor the localization of visual memories inaddition to the central complex (Figure 1d ).

COMPLEX FORMS OF VISUALLEARNING (THAT MAY NOTBE SO COMPLEX)Only in the 1990s did researchers becomeinterested in the existence of cognitive process-ing in insects, and the honey bee was the modelchosen to address most of the works performedin that direction. Such a delay with respectto the cognitive revolution, which flourishedfrom the late 1970s to the early 1980s (73), canonly be explained by the reluctance to view in-vertebrates, and therefore insects, as organismscapable of nonelemental, higher-order formsof learning. For instance, the main idea withrespect to visual pattern learning, which is stillsometimes defended, was that insects can onlyview isolated spots, blobs, and bars withoutthe capacity of integrating them into a givenconfiguration (52–55). Even a basic capacity ofrecognition systems such as generalization, theability of individuals to respond to stimuli that,despite being different from a trained target,are nevertheless perceptually similar to it (81,83), was and is considered by some researchersas too high-level for a honey bee (54, 55). Yet,dozens of works had already shown that honeybees generalize their choice of visual patternsto novel figures that have some similarity tothose that have been trained (e.g., 1, 100).This refusal of generalization capacities isconsistent with the preconception that insectshave limited plasticity and should be viewedrather as reflex machines reacting to specificfeatures in the environment to which they aretuned.

In the past decade, however, researchershave found evidence showing that bees are notreflex machines and that they exhibit visuallearning capabilities that were only suspected insome vertebrates. Some of these capacities aresurprising and may be viewed as nonelemental.However, an alternative view could argue thatit is possible to explain them based on simple,elemental associations. These experiments, re-viewed in the next section, were not conceivedto address these opposite views, so that we arecurrently unable to determine whether these


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


performances are forms of elemental or higher-order learning.

Visually Based Individual Recognitionin Wasps

The capacity of individuals to recognize theirdistinctive identity has long been dismissed insocial insects due to the cognitive requirementsthat such performance may impose on coloniesmade of thousand of individuals. For instance,Wilson (104) stated that “. . . insect societies are,for the most part, impersonal . . . The sheer sizeof the colonies and the short life of the mem-bers make it inefficient, if not impossible, to es-tablish individual bonds.” However, not all so-cial insects live in huge, overcrowded societies.Small, relatively primitive colonies of bumblebees, wasp, and some ant species are based ondominance hierarchies where individual recog-nition may be crucial for responding appropri-ately to a conspecific. Indeed, recent studieshave shown that Pachycondyla villosa ant queensrecognize each other using olfactory, cuticu-lar cues (20). In the visual domain, studies onthe paper wasp, Polistes fuscatus, have shownthat individual recognition is achieved throughlearning the yellow-black patterns of the waspfaces and/or abdomens (92). In another species,Polistes dominulus, more variable patterns withlarger black components were carried by indi-viduals ranking higher in the nest hierarchy.Altering these facial and/or abdominal colorpatterns induced aggression against such ani-mals, whether or not their patterns were madeto signal higher or lower ranking. These results,however, were challenged by another study(9a), which could not find evidence support-ing the hypothesis that the facial patterns ofP. dominulus act as hierarchy or quality signals.Size, on the contrary, was highly correlated withsocial dominance in this wasp (9a). Althoughthese results question the validity of the hypoth-esis that posits that visual facial patterns thatcontain more black areas or more black spotsare associated with dominant wasps, they donot exclude the possibility that visual patternsare used as individual identity markers rather

than as status markers. In this scenario, waspswould recognize each other on the basis of theirfacial features, and each individual mask, re-gardless of its amount of black areas or spots,would have an unambiguous outcome in termsof its ranking in the social structure (i.e., maskA → α individual and mask B → β individual).Wasps would learn a series of elemental associa-tions between mask patterns and social ranking.Given the small size of colonies, in which 5 to10 individuals can coexist, storing several mem-ories, one for each individual, seems plausible.If this were the case, a fundamental goal wouldbe to characterize the storage capacity of thevisual memory as related to colony size.

Observatory Learning in Bumble Bees

Recent studies on bumble bees (57, 58, 109)have shown that these insects copy other bees’learned foraging preferences by observing theirchoices of visual, rewarded targets. Bumblebees, B. terrestris, are influenced by other con-specifics when sampling unfamiliar flowers,such that they land on unknown flowers whereother bees have been (57). This occurs evenwhen naıve bees are separated from experiencedforagers by a transparent screen so that they canneither sample the flowers by themselves norinteract with their foraging conspecifics (109).Similarly, naıve bees abandon an unrewardingspecies and switch to a more rewarding alter-native more quickly if accompanied by experi-enced foragers (58).

As surprising as this performance mayappear, it can be accounted for by an elementalform of associative learning called second-order conditioning (76), which involves twoconnected associations. In this scenario, ananimal first learns an association between a CSand an US (CS1 + US) and then experiencesa pairing between a new conditioned stimulus,CS2 and CS1 (CS1 + CS2). In this way, CS2becomes meaningful, directly through itsassociation with CS1 and indirectly with theUS. How would this apply to the observa-tional learning of bumble bees? One couldpropose that naıve bumble bees would first


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


associate the presence of a conspecific withreward (CS1 + US) simply by foraging close toexperienced foragers. Afterward, observing aconspecific landing on a given color may allowan association between color and conspecific(CS2 + CS1) to be established (58). Theseconnected elemental links may thus underliethe observational learning of bees. This hy-pothesis is supported by the fact that honeybees can learn second-order associations whilesearching for food. They learn to connect bothtwo odors (odor 1 + sucrose reward; Odor 2 +Odor 1) (7) and one odor and one color (odor+ sucrose reward; color + odor) (42).

Symbolic Matching to Sampleand Other Forms of ConditionalDiscrimination in Honey Beesand Bumble Bees

Symbolic matching to sample is a term usedto describe an experimental situation in whichthe correct response to a problem dependson a specific background or condition. Inother words, animals have to learn, for in-stance, that given condition A, response C iscorrect, while for condition B, response Dis correct. Symbolic matching to sample is aform of conditional discrimination becausea given stimulus, the sample (also called theoccasion setter), sets the condition for thenext choice. Using this design, Zhang et al.(117) trained honey bees to fly though acompound Y-maze consisting of a series ofinterconnected cylinders. The first cylindercarried the sample stimulus (e.g., a vertical ora horizontal black-and-white grating). Thesecond and third cylinders had two exits apiece.Each exit presented a visual stimulus so that thebee had to choose between them. In the secondcylinder, bees had to choose between a blueand a green square. In the third cylinder, theyhad to choose between a radial sectored patternand a ring pattern. Correct sequences of choiceswere “Vertical–Green–Ring” and “Horizontal–Blue–Radial.” Only after making a succession ofcorrect choices (i.e., in both the second and thethird cylinders) could a bee reach a feeder with

sucrose solution. The bees learned to masterthese successive associations between differentkinds of visual cues (117). This finding was alsoextended to other sensory modalities, as thesame principle applied when visual cues werecombined with odors in a similar protocol (87).

Conditional learning allows other variantsthat, depending on the number of occasion set-ters and discriminations involved, have receiveddifferent names. For instance, another form ofconditional discrimination involving two occa-sion setters is the so-called transwitching prob-lem. In this problem, an animal is trained differ-entially with two stimuli, A and B, and with twodifferent occasion setters, C and D. When C isavailable, stimulus A is rewarded, whereas stim-ulus B is not (A+ versus B-); the opposite occurs(A- versus B+) when D is available. The tran-switching problem is also considered a form ofcontextual learning because the occasion settersC1 and C2 can be viewed as contexts determin-ing the appropriateness of each choice. Bum-ble bees have been trained in a transwitchingproblem to choose a 45◦ grating and to avoida 135◦ grating to reach a feeder, and to do theopposite to reach their nest (28). Here, the nestand the feeder provide the appropriate contextsdefining what has to be chosen. Bumble beescan also learn that an annular or a radial discmust be chosen, depending on the disc’s asso-ciation with a 45◦ or a 135◦ grating either at thefeeder or the nest entrance: At the nest, accesswas allowed by the combinations 45◦ + radialdisc and 135◦ + annular disc, but not by thecombinations 45◦ + annular disc and 135◦ +radial disc; at the feeder, the opposite applied(27). In both cases, the potentially competingvisuomotor associations were insulated fromeach other because they were set in differentcontexts.

Solving this kind of problem can be viewedas a form of nonelemental learning and thus as asophisticated form of cognitive visual process-ing. Indeed, as for other forms of conditionaldiscrimination, one could describe this protocolas CA+, CB- (if C then A, but not B), and DA-,DB+ (if D then B, but not A). Each stimulus,A, B, C, and D, therefore is rewarded as often


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


as it is nonrewarded, such that solutions cannotbe based on the mere outcome of A, B, C or D.A higher-order solution then would be to learnthe outcome of each particular configuration,CA, CB, DA, or DB. However, an alternativeexplanation could argue that what the insectsdo is establish hierarchical simple associationsas the ones underlying second-order condition-ing (see above). Indeed, one could imagine thatbees learn to associate a radial disc with sucrosereward and that they then learn to associate a45◦ grating with the radial disc. This is a rela-tively simple strategy that is probably used bybees for navigational purposes (116, 117) whenthey are confronted with successions of differ-ent landmarks en route to the goal.

A critical factor determining one strategy orthe other therefore may be the temporal orderof stimulus presentation. If these are presentedserially, learning chains of simple associationscould be primed, while simultaneous presenta-tion of stimuli may prime learning of configu-rations and their specific outcome. An exampleof the latter is the case of honey bees trainedto solve a biconditional discrimination, AC+,BD+, AD-, BC-, in which all four stimuli werepresented simultaneously and were rewarded asoften as they were nonrewarded (80). Four dif-ferent gratings combining one color (yellow orviolet = A or B) with one orientation (hori-zontal or vertical = C or D) were used in sucha way that bees had to learn that, for instance,yellow-horizontal (AC) and violet-vertical (BD)were rewarded while yellow-vertical (AD) andviolet-horizontal (BC) were nonrewarded. Beeslearned to choose the rewarded stimuli, al-though the colors and orientations were am-biguous when considered alone. Thus, theylearned the configurations and not the specificoutcome of the elements (80). Again, cumu-lative experience is a critical factor promotingconfigural learning (36). Although few learningtrials promote processing of a compound colorstimulus made of A and B elements as the sumof A and B, increasing number of trials resultsin bees treating the compound AB as a uniqueentity that is different from its composingelements (36).

NONELEMENTAL VISUALLEARNING

The visual performances of the previous sec-tion could be accounted for by elemental asso-ciations despite their sophistication. A higherlevel of complexity, however, is reached whenanimals respond in an adaptive manner to novelstimuli that they have never encountered beforeand that do not predict a specific outcome perse based on the animals’ past experience. Such apositive transfer of learning (78) is therefore dif-ferent from elemental forms of learning, whichlink known stimuli or actions to specific rewards(or punishments). In the cases considered in thissection, the insects’ responses have in commonthe transfer to novel stimuli, which cannot beexplained based on the previous knowledge thatthe animal has of these stimuli.

Categorization of Visual Stimuliin Honey Bees

Visual categorization refers to the classificationof visual stimuli into defined functional groups(44). It can be defined as the ability to group dis-tinguishable objects or events on the basis of acommon feature or set of features and thereforeto respond similarly to them (94, 113). A typi-cal categorization experiment trains an animalto extract the basic attributes of a category andthen tests it with novel stimuli that were neverencountered before and that may or may notpresent the attributes of the category learned.If the animal chooses the novel stimuli on thebasis of these attributes, it classifies them as be-longing to the category and exhibits thereforepositive transfer of learning.

Several studies have shown the ability of vi-sual categorization in free-flying honey beestrained to discriminate different patterns andshapes. For instance, van Hateren et al. (95)trained bees to discriminate two given grat-ings presented vertically with different orienta-tions (e.g., 45◦ versus 135◦) by rewarding one ofthese gratings with sucrose solution. Each beewas trained with a changing succession of pairsof different gratings, one of which was always


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


rewarded and the other not. Despite the differ-ence in pattern quality, all the rewarded pat-terns had the same edge orientation and all thenonrewarded patterns also had a common ori-entation, which was perpendicular to the re-warded one. Under these circumstances, thebees had to extract and learn the orientationthat was common to all rewarded patterns tosolve the task. This was the only cue predict-ing reward delivery. In the tests, bees were pre-sented with novel patterns, all of which wereall nonrewarded, and exhibited the same stripeorientations as the rewarding and nonreward-ing patterns employed during the training. Insuch transfer tests, bees chose the appropriateorientation despite the novelty of the structuraldetails of the stimuli. Thus, bees could catego-rize visual stimuli on the basis of their globalorientation.

They can also categorize visual patternsbased on their bilateral symmetry. Whentrained with a succession of changing patternsto discriminate bilateral symmetry from asym-metry, they learn to extract this informationfrom different figures and to transfer it to novelsymmetrical and asymmetrical patterns (31).Similar conclusions apply to other visual fea-tures such as radial symmetry, concentric pat-tern organization and pattern disruption (seeReference 5 for a review), and even photographsbelonging to a given class (e.g., radial flower,landscape, plant stem) (118).

How could bees appropriately classify dif-ferent photographs of radial flowers if they varyin color, size, outline? An explanation was pro-vided by Stach et al. (88), who showed that dif-ferent coexisting orientations can be consideredat one time and that they can be integrated intoa global stimulus representation that is the ba-sis for the category (88). Thus, a radial flowerwould be, in fact, the conjunction of five ormore radiating edges. In addition to focusingon a single orientation, honey bees assembleddifferent features to build a generic pattern rep-resentation, which could be used to respondappropriately to novel stimuli sharing such abasic layout. Honey bees trained with a seriesof complex patterns sharing a common layout

comprising four edges oriented differently re-membered these orientations simultaneously intheir appropriate positions and transferred theirresponse to novel stimuli that preserved thetrained layout. These results show that honeybees extract regularities in their visual envi-ronment and establish correspondences amongcorrelated features. Therefore, they may gen-erate a large set of object descriptions from afinite set of elements.

This capacity can explain recent findingsshowing that honey bees learn to recognizehuman faces if trained to do so (23). In thiscase, bees were rewarded with sugar water tochoose and distinguish people’s faces from pho-tographs. Bees chose the appropriate photo-graph, thus showing a capacity to discriminatethis particular kind of stimuli. Does this meanthat bees realize that two different persons arebehind two discriminated photographs? Notreally. To the bees, the photographs weremerely strange flowers. The question thenwould be what information from the pho-tographs was used to recognize the right stim-ulus. This question was tackled by a work thatstudied whether bees can bind the features ofa face-like stimulus (two dots in the upper partrepresenting the eyes, a vertical line below asthe nose, and a horizontal line in the lower partas the mouth) and recognize faces using thisbasic configuration (2). Bees did indeed distin-guish between different variants of the face-likestimuli, thus showing that they discriminate be-tween these options, but grouped together andreacted, therefore, similarly to faces if trained todo so. Stimuli made of the same elements (twodots, a vertical line, and a horizontal line) butnot preserving the configuration of a face werenot recognized as positive, thus showing thatbees learn that the rewarded stimulus consistsof a series of elements arranged in the specificspatial configuration of a face (2). Furthermore,if trained with real faces, bees can learn to rec-ognize novel views of a face by interpolatingbetween or averaging views they have experi-enced (25).

In any case, honey bees show positivetransfer of learning from a trained to a novel


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


set of stimuli, and their performances are con-sistent with the definition of categorization.Visual stimulus categorization therefore is nota prerogative of certain vertebrates. However,this result might not be surprising as it allowsan elemental learning interpretation. Tounderstand this elemental interpretation, thepossible neural mechanisms underlying catego-rization should be considered. If we allow thatvisual stimuli are categorized on the basis ofspecific features such as orientation, the neuralimplementation of category recognition couldbe relatively simple. The feature(s) allowingstimulus classification would activate specificneuronal detectors in the optic lobes, the visualareas of the bee brain. Examples of such featuredetectors are the orientation detectors whoseorientation and tuning have already beencharacterized by means of electrophysiologicalrecordings in the honey bee optic lobes (111).Thus, responding to different gratings havinga common orientation of, say, 60◦ is simple,because all these gratings elicit the same neuralactivation in the same set of orientation detec-tors despite their different structural quality.

In the case of category learning, the activa-tion of an additional neural element is needed.Such an element would be a reward neuronwhose activity would substitute for sucrose re-ward. Such a neuron has been identified in thehoney bee brain, VUMmx1 (ventral unpairedmedian neuron located in the maxillar neu-romere 1) (43). VUMmx1 mediates olfactorylearning in the honey bee because it contactsthe olfactory circuit at its key processing stagesin the brain. In other words, when an odor acti-vates the olfactory circuit, concomitant sucrosestimulation activates VUMmx1, thus providingthe basis for neural coincidence between odorand reward. The branching of VUMmx1 makesit specific for the olfactory circuit and thus forolfactory learning (43). Other VUM neuronswhose function is still unknown are present inthe bee brain (79). Some of them could providethe neural basis of reward in associative visuallearning.

Category learning could thus be re-duced to the progressive reinforcement of an

associative neural circuit relating visual-codingand reward-coding neurons, similar to thatunderlying simple associative (e.g., Pavlovian)conditioning. From this perspective, even if cat-egorization is viewed as a nonelemental learn-ing form because it involves positive transferof learning, it may simply rely on elementallinks between conditioned and unconditionedstimuli.

An even simpler alternative may account forthis performance. The mechanism explainedabove could be viewed as a form of supervisedlearning, in which a visual network is instructedby the external signal of the reinforcement neu-ron to respond to the right combination of fea-tures. Recent modeling work on the vertebratevisual system has shown that visual networkscan learn to extract the distinctive features of acategory without any kind of supervision (65).The model relies on spike-timing-dependentplasticity (STDP), which is a learning rule thatmodifies synaptic strength as a function of therelative timing of pre- and postsynaptic spikes.When a neuron is repeatedly presented withsimilar inputs, STDP has the effect of concen-trating high synaptic weights on afferents thatsystematically fire early, while postsynapticspike latencies decrease. Masquelier et al. (65)showed that a network that exhibits STDP andis repeatedly presented with natural images of agiven category becomes progressively tuned torespond better to the features that correspondto prototypical patterns of the category. Thosefeatures that are both salient and consistentlypresent in the images are highly informativeand enable robust object recognition. Testingwhether similar neural mechanisms underlieobject categorization in the insect visual systemwould be a fascinating endeavor.

Rule Learning in Honey Bees

In rule learning, the animal learns relations be-tween objects, not the objects themselves. Typ-ical examples are the so-called rules of samenessand of difference. Sameness and differencerules are demonstrated through the protocolsof delayed matching to sample (DMTS) and


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


delayed nonmatching to sample (DNMTS),respectively. In DMTS, animals are presentedfirst with a sample and then with a set of stimuli,one of which is identical to the sample andis reinforced. Because the sample is changedregularly, animals must learn the samenessrule, i.e., “always choose what is shown to you(the sample), independent of what else is shownto you.” In DNMTS, the animal must learn theopposite, i.e., “always choose the opposite ofwhat is shown to you (the sample).” Honey beesforaging in a Y-maze learn both rules (38). Beeswere trained in a DMTS problem in which theywere presented with a changing nonrewardedsample (i.e., one of two different coloreddisks or one of two different black-and-whitegratings, vertical or horizontal) at the entranceof a maze (Figure 2). The bees were rewardedonly if they chose the stimulus identical tothe sample once within the maze. Bees trainedwith colors and presented in transfer tests withunknown black-and-white gratings solved theproblem and chose the grating identical to thesample at the entrance of the maze. Similarly,bees trained with the gratings and tested withcolors in transfer tests also solved the problemand chose the novel color corresponding tothat of the sample grating at the maze entrance.Transfer was not limited to modalities withinthe visual domain (pattern versus color) butcould also operate between different domainssuch as olfaction and vision (38). Furthermore,bees also mastered a DNMTS task, thus show-ing that they learn a rule of difference betweenstimuli (38). The capacity of honey bees to solvea DMTS task has recently been analyzed withrespect to the working memory underlying it(114). It was found that the sample is stored forapproximately 5 s (114), a period that coincideswith the duration of other visual and olfactoryshort-term memories characterized in simplerforms of associative learning in honey bees(69; see above). Moreover, bees trained in aDMTS task can learn to pay attention to oneof two different samples presented successivelyin a flight tunnel and can transfer the learningof this sequence weight to novel samples(114).

Despite the honey bees’ evident capacity tosolve relational problems such as the DMTSor the DNMTS tasks, such capacities are notunlimited. In some cases, biological constraintsmay impede the bee from solving a particu-lar problem for which rule extraction is nec-essary. It is therefore interesting to focus ona different example of rule learning that beescould not master, the transitive inference prob-lem (4). In this problem, animals must learna transitive rule, i.e., A > B, B > C, thenA > C. Preference for A over C in this contextcan be explained by two strategies: (a) deduc-tive reasoning (97) in which the experimentalsubjects construct and manipulate a unitary andlinear representation of the implicit hierarchyA > B > C, and (b) responding as a functionof reinforced and non-reinforced experiences(91), in which animals choose among stimuli onthe basis of the effective number of reinforcedand nonreinforced experiences (A is always re-inforced whereas C is always nonreinforced).

To determine whether bees can learn a tran-sitive rule, they were trained with five differentvisual stimuli (A, B, C, D, and E) in a multiplediscrimination task: A+ versus B-, B+ versusC-, C+ versus D-, D+ versus E- (4). Train-ing involved overlapping adjacent premise pairs(A > B, B > C, C > D, D > E), which under-lie a linear hierarchy A > B > C > D > E.After training, bees were tested with B ver-sus D, a nonadjacent pair of stimuli that werenever explicitly trained together. In theory, Band D have equivalent associative strengths be-cause they are, in principle, equally associatedwith reinforcement or no reinforcement duringtraining. Thus, if bees were guided by the stim-ulus’ associative strength, they should chooserandomly between B and D. If, however, beesused a transitive rule, they should prefer B toD.

Honey bees learned the premise pairs as longas these were trained as uninterrupted, consec-utive blocks of trials (4). But if shorter and in-terspersed blocks of trials were used, such thatbees had to master all pairs practically simul-taneously, performance collapsed and bees didnot learn the premise pairs. The bees’ choice


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


Sample

Transfer tests with colors(training with patterns)

Transfer tests with patterns(training with colors)

Vertical Horizontal

Preferencefor vertical

Preferencefor horizontal

d

Blue YellowSample

Preferencefor blue

Preferencefor yellow

a

c

b

1.0

0.8

0.6

Co

rre

ct c

ho

ice

s

0.4

0.2

0

Figure 2Rule learning in honey bees. Honey bees trained in a delayed matching-to-sample task to collect sugarsolution in (a) a Y-maze on a series of (b) patterns or colors learn a rule of sameness. (c, d ) Transfer tests withnovel stimuli. (c) In Experiment 1, bees trained on the colors were tested on the patterns. (d ) In Experiment2, bees trained on the patterns were tested on the colors. In both cases, bees chose the novel stimulicorresponding to the sample, although they had no experience with such test stimuli. Adapted fromReference 38.

was significantly influenced by their experiencewith the last pair of stimuli (D+ versus E-) suchthat they preferred D and avoided E. In thetests, no preference for B to D was found. Al-though this result agrees with an evaluation ofstimuli in terms of their associative strength (seeabove), during training bees more often visited

B when it was rewarding (B+ vs. C-) than Dwhen it was rewarding (D+ vs. E-), such that apreference for B should have been expected ifonly the associative strength were guiding thebees’ choices. It was then concluded that beesdo not establish transitive inferences betweenstimuli but rather guide their choices by the


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


joint action of a recency effect (preference ofthe last rewarded stimulus, D) and by an eval-uation of the associative strength of the stimuli(in which case preference for B should be evi-dent). As the former supports choice of D whilethe latter supports choice of B, equal choice ofB and D in the tests could be explained (4). Inany case, memory constraints (in this case thefact that simultaneous mastering of the differentpremise pairs was not possible and the fact thatthe last excitatory memory seems to predomi-nate over previous memories) impeded learningthe transitive rule. Recently, Cheng & Wignall(11) demonstrated that failure to master severalconsecutive visual discriminations is due to re-sponse competition occurring when animals aretested. This may explain why bees in the tran-sitive inference protocol were unable to masterthe successive short blocks of training with dif-ferent premise pairs.

Counting

Counting could be useful in navigation taskswhere the number of landmarks encounteredduring a foraging trip or near the hive maycontribute to efficient orientation of free-flyingbees. Furthermore, it could also improve for-aging through evaluation of food source prof-itability (e.g., number of flowers in a patch).Whether or not honey bees estimate numeros-ity has been addressed recently by two dif-ferent works that reached similar conclusions(18, 41).

Dacke & Srinivasan (18) were inspired byChittka & Geiger’s (13) pioneering work sug-gesting that bees may count landmarks en routeto the feeder. In Dacke & Srinivasan’s protocol,bees were trained to fly into a tunnel to finda food reward after a given number of land-marks. The shape, size, and positions of thelandmarks were changed in the different test-ing conditions in order to avoid any confound-ing factor(s). Bees showed a stronger preferenceto land after the correct number of landmark innonrewarded tests. This behavior was observedwhen bees were trained to collect reward af-ter one, two, three, or four landmarks but not

more, thus indicating a limit in their countingcapacity.

A similar limit was found in a DMTS proto-col (41; see above) in which bees had to choosethe stimulus containing the same number ofitems as a sample. The authors controlled forlow-level cues such as cumulated area and edgelength, configuration identity, and illusionaryshape similarity formed by the elements. Theirresults showed that honey bees have the capac-ity to match visual stimuli in a DMTS task aslong as the number of items does not exceedfour. Together with Dacke & Srinivasan’s work(18), this result indicates that the bee brain candeal with a real numerosity concept, even if itis limited to a number of four. Interestingly,the same limit was found in humans when timeexposure of items to be counted is limited (17).

CONCLUSION

Almost one hundred years of research on visuallearning in bees and other social Hymenoptera,starting with Karl von Frisch’s (98) first demon-strations on color and pattern learning in bees,have yielded an impressive amount of informa-tion about how honey bees, bumble bees, andwasps see the world and learn about visual cuesin their environment. New discoveries in thisfield have shown that, in addition to simpleforms of visual learning, social Hymenopteraalso master complex forms of visual learning,from conditional discriminations and observa-tional learning to rule learning. Although thecognitive capabilities of bees and wasps maysurprise owing to their sophistication, limita-tions related to natural life seem inescapable.For instance, in the case of wasps learning fa-cial mask patterns of conspecifics, one couldimagine that interindividual recognition is cer-tainly possible but probably has limitations interms of the number of individuals that canbe learned and remembered. Similarly, mas-tering several different associations simultane-ously would be facilitated if these are organizedserially or hierarchically in chains of associa-tions that can mediate successful navigation in acomplex environment. But if these associations


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


have to be mastered simultaneously at the sameplace, learning them would probably be diffi-cult given the bees’ biological specialization asa serial forager. In this case, learning configu-rations of stimuli may be more adaptive thanlearning each component separately.

If bees and wasps exhibit such a high degreeof complex forms of visual learning, which kindof limitation do they present as a model forunraveling the mechanisms of these phenom-ena? The main limitation resides so far in theimpossibility of addressing questions relatedto the cellular and molecular mechanismsunderlying these learning forms. Learningprotocols have exploited the advantage of notrestraining the animals’ movements so that thebehaviors recorded express all the potential ofthe insect brain. However, they are limitingbecause no access to the brain is so far possiblein a flying bee. As mentioned above, new

protocols in which bees learn color-rewardand motion cues–reward associations underrestrained conditions (50, 51) are promisingbecause they allow the neural circuits involvedin these learning forms to be accessed (30). Thecritical question would be then to what extentdo restraining conditions limit the expressionof more complex forms of visual learning?

Why should bees and wasps continue to beattractive for research on visual cognition de-spite this technical limitation? The answer issimple: They exhibit sophisticated visual per-formances that cannot at this time be uncoveredin a fruit fly. Future research in social insectsshould benefit from a comparative analysis be-tween the visual performances and mechanismsof bees and flies and overcome the historic bur-den of not having a window open to the neuraland molecular basis of visual learning, irrespec-tive of the level of complexity considered.

SUMMARY POINTS

1. Visual learning admits different levels of complexity, from the formation of a simpleassociative link between a visual stimulus and its outcome (elemental learning), to moresophisticated performances such as object categorization or rules learning, which allowflexible responses beyond simple forms of learning (nonelemental learning). Social insectsexcel at visual learning in a foraging and navigation context.

2. Honey bees, ants, and bumble bees learn to associate different kinds of visual cues suchas colors or patterns with food reward. Even if these associations remain elemental, per-formance can be modulated by the complexity of the task, thus suggesting attentionalprocesses in these insects. Attention- and experience-dependent changes in visual dis-crimination can be traced to the neural level. Studies in the fruit fly suggest that theseprocesses can be located at the level of mushroom bodies, a structure of the insect braininvolved in learning and memory.

3. Social insects exhibit sophisticated visual abilities. Wasps recognize each other on thebasis of facial marks, bumble bees learn to choose profitable food sources by observingthe choice of other bees, and honey bees and bumble bees learn to solve different kindsof conditional discriminations. These performances admit both elemental and nonele-mental explanations.

4. Depending on their past experience, honey bees respond in an adaptive manner to novelstimuli that they have never encountered before. Such a positive transfer of learning,characteristic of nonelemental forms of learning, has been shown in studies demonstrat-ing categorization, learning of rules such as sameness or difference, and basic countingabilities in bees.


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


DISCLOSURE STATEMENT

The authors are not aware of any affiliations, memberships, funding, or financial holdings thatmight be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS

We thank the support of the French National Research Agency (ANR; Project APICOLOR), theFrench Research Council (CNRS), and the University Paul Sabatier (Project APIGENE). AuroreAvargues-Weber and Nina Deisig contributed equally to this review. This review is dedicated toProf. Dr. Randolf Menzel on his seventieth anniversary.

LITERATURE CITED

1. Anderson AM. 1972. The ability of honey bees to generalize visual stimuli. In Information Processing inthe Visual Systems of Arthropods, ed. R Wehner, pp. 207–12. Berlin: Springer

2. Avargues-Weber A, Portelli G, Benard J, Dyer A, Giurfa M. 2010. Configural processing enables dis-crimination and categorization of face-like stimuli in honeybees. J. Exp. Biol. 213:593–601

3. Beier W, Menzel R. 1972. Untersuchungen uber den Farbensinn der deutschen Wespe (Paravespulagermanica F., Hymenoptera, Vespidae): verhaltensphysiologischer Nachweis des Farbensehens. Zool. Jb.Physiol. 76:441–54

4. Benard J, Giurfa M. 2004. A test of transitive inferences in free-flying honeybees: unsuccessful perfor-mance due to memory constraints. Learn. Mem. 11:328–36

5. Benard J, Stach S, Giurfa M. 2006. Categorization of visual stimuli in the honeybee, Apis mellifera. Anim.Cogn. 9:257–70

6. Biesmejer JC, Giurfa M, Koedam D, Potts SG, Joel DM, Dafni A. 2005. Convergent evolution: floralguides, stingless bee next entrances, and insectivorous pitchers. Naturwissenchaften 92:444–50

7. Bitterman ME, Menzel R, Fietz A, Schafer S. 1983. Classical conditioning of proboscis extension inhoneybees (Apis mellifera). J. Comp. Psychol. 97:107–19

8. Camlitepe Y, Aksoy V. 2010. First evidence of fine colour discrimination ability in ants (Hymenoptera,Formicidae). J. Exp. Biol. 213:72–77

9. Campan R, Lehrer M. 2002. Discrimination of closed shapes by two species of bee, Apis mellifera andMegachile rotundata. J. Exp. Biol. 205:559–72

9a. Cervo R, Dapporto L, Beani L, Strassmann JE, Turillazzi S. 2008. On status badges and quality signalsin the paper wasp Polistes dominulus: body size, facial colour patterns and hierarchical rank. Proc. R. Soc.B 275:1189–96

10. Cheng K, Narendra A, Sommer S, Wehner R. 2009. Traveling in clutter: navigation in the centralAustralian desert ant Melophorus bagoti. Behav. Process. 80:261–68

11. Cheng K, Wignall AE. 2006. Honeybees (Apis mellifera) holding on to memories: Response competitioncauses retroactive interference effects. Anim. Cogn. 9:141–50

12. Chittka L, Beier W, Hertel H, Steinmann E, Menzel R. 1992. Opponent colour coding is a universalstrategy to evaluate the photoreceptor inputs in Hymenoptera. J. Comp. Physiol. A 170:545–63

13. Chittka L, Geiger K. 1995. Can honey bees count landmarks? Anim. Behav. 49:159–6414. Chittka L, Thomson JD, Waser NM. 1999. Flower constancy, insect psychology, and plant evolution.

Naturwissenschaften 86:361–7715. Collett TS, Collett M. 2002. Memory use in insect visual navigation. Nat. Rev. Neurosci. 3:542–5216. Collett TS, Graham P, Durier V. 2003. Route learning by insects. Curr. Opin. Neurobiol. 13:718–2517. Cowan N. 2001. The magical number 4 in short-term memory: a reconsideration of mental storage

capacity. Behav. Brain Sci. 24:87–11418. Dacke M, Srinivasan MV. 2008. Evidence for counting in insects. Anim. Cogn. 11:683–8919. de Brito Sanchez M, Chen C, Li J, Liu F, Gauthier M, Giurfa M. 2008. Behavioral studies on tarsal

gustation in honeybees: sucrose responsiveness and sucrose-mediated olfactory conditioning. J. Comp.Physiol. A 194:861–69


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


20. D’Ettorre P, Heinze J. 2005. Individual recognition in ant queens. Curr. Biol. 15:2170–7421. Dyer AG, Chittka L. 2004. Fine colour discrimination requires differential conditioning in bumblebees.

Naturwissenschaften 91:224–2722. Dyer AG, Neumeyer C. 2005. Simultaneous and successive colour discrimination in the honeybee (Apis

mellifera). J. Comp. Physiol. A 191:547–5723. Dyer AG, Neumeyer C, Chittka L. 2005. Honeybee (Apis mellifera) vision can discriminate between and

recognise images of human faces. J. Exp. Biol. 208:4709–1424. Dyer AG, Rosa MGP, Reser D. 2008. Honeybees can recognise images of complex natural scenes for

use as potential landmarks. J. Exp. Biol. 211:1180–8625. Dyer AG, Vuong QC. 2008. Insect brains use image interpolation mechanisms to recognize rotated

objects. PLoS ONE 3:e408626. Ehmer B, Gronenberg W. 2002. Segregation of visual input to the mushroom bodies in the honeybee

(Apis mellifera). J. Comp. Neurol. 451:362–7327. Fauria K, Colborn M, Collett TS. 2000. The binding of visual patterns in bumblebees. Curr. Biol.

10:935–3828. Fauria K, Dale K, Colborn M, Collett TS. 2002. Learning speed and contextual isolation in bumblebees.

J. Exp. Biol. 205:1009–1829. Giurfa M. 2004. Conditioning procedure and color discrimination in the honeybee Apis mellifera. Natur-

wissenschaften 91:228–3130. Giurfa M. 2007. Behavioral and neural analysis of associative learning in the honeybee: a taste from the

magic well. J. Comp. Physiol. A 193:801–2431. Giurfa M, Eichmann B, Menzel R. 1996. Symmetry perception in an insect. Nature 382:458–6132. Giurfa M, Hammer M, Stach S, Stollhoff N, Muller-Deisig N, Mizyrycki C. 1999. Pattern learning by

honeybees: conditioning procedure and recognition strategy. Anim. Behav. 57:315–2433. Giurfa M, Lehrer M. 2001. Honeybee vision and floral displays: from detection to close-up recognition.

In Cognitive Ecology of Pollination, ed. L Chittka, JD Thomson, pp. 61–82. Cambridge, UK: CambridgeUniversity Press

34. Giurfa M, Menzel R. 1997. Insect visual perception: complex abilities of simple nervous systems. Curr.Opin. Neurobiol. 7:505–13

35. Giurfa M, Nunez JA, Chittka L, Menzel R. 1995. Colour preferences of flower-naive honeybees.J. Comp. Physiol. A 177:247–59

36. Giurfa M, Schubert M, Reisenman C, Gerber B, Lachnit H. 2003. The effect of cumulative experienceon the use of elemental and configural visual discrimination strategies in honeybees. Behav. Brain Res.145:161–69

37. Giurfa M, Vorobyev M, Kevan P, Menzel R. 1996. Detection of coloured stimuli by honeybees: minimumvisual angles and receptor specific contrasts. J. Comp. Physiol. A 178:699–709

38. Giurfa M, Zhang S, Jenett A, Menzel R, Srinivasan MV. 2001. The concepts of ‘sameness’ and‘difference’ in an insect. Nature 410:930–33

39. Graham P, Collett TS. 2002. View-based navigation in insects: how wood ants (Formica rufa L.) look atand are guided by extended landmarks. J. Exp. Biol. 205:2499–509

40. Grant V. 1951. The fertilization of flowers. Sci. Am. 12:1–641. Gross HJ, Pahl M, Si A, Zhu H, Tautz J, Zhang S. 2009. Number-based visual generalisation in the

honeybee. PLoS ONE 4:e426342. Grossmann KE. 1971. Belohnungsverzogerung beim Erlernen einer Farbe an einer kunstlichen Futter-

stelle durch Honigbienen. Z. Tierpsychol. 29:28–4143. Hammer M. 1993. An identified neuron mediates the unconditioned stimulus in associative olfactory

learning in honeybees. Nature 366:59–6344. Harnard S. 1987. Categorical Perception. The Groundwork of Cognition. Cambridge, UK: Cambridge Uni-

versity Press45. Harris RA, Hempel de Ibarra N, Graham P, Collett TS. 2005. Ant navigation: priming of visual route

memories. Nature 438:30246. Heinrich B. 1979. “Majoring” and “Minoring” by foraging bumblebees, Bombus vagans: an experimental

analysis. Ecology 60:245–55


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


47. Hertz M. 1929. Die Organisation des optischen Feldes bei der Biene I. Z. Vergl. Physiol. 8:693–74848. Hertz M. 1929. Die Organisation des optischen Feldes bei der Biene II. Z. Vergl. Physiol. 11:107–4549. Hertz M. 1933. Uber figurale Intensitaten und Qualitaten in der optischen Wahrnehmung der Biene.

Biol. Zb. 53:11–4050. Hori S, Takeuchi H, Arikawa K, Kinoshita M, Ichikawa N, et al. 2006. Associative visual learning, color

discrimination, and chromatic adaptation in the harnessed honeybee Apis mellifera L. J. Comp. Physiol. A192:691–700

51. Hori S, Takeuchi H, Kubo T. 2007. Associative learning and discrimination of motion cues in theharnessed honeybee Apis mellifera L. J. Comp. Physiol. A 193:825–33

52. Horridge A. 2000. Seven experiments on pattern vision of the honeybee, with a model. Vis. Res. 40:2589–603

53. Horridge A. 2003. The effect of complexity on the discrimination of oriented bars by the honeybee (Apismellifera). J. Comp. Physiol. A 189:703–14

54. Horridge A. 2006. Visual discriminations of spokes, sectors, and circles by the honeybee (Apis mellifera).J. Insect Physiol. 52:984–1003

55. Horridge A. 2009. Generalization in visual recognition by the honeybee (Apis mellifera): a review andexplanation. J. Insect Physiol. 55:499–511

56. Kuwabara M. 1957. Bildung des bedingten Reflexes von Pavlovs Typus bei der Honigbiene, Apis mellifica.J. Fac. Sci. Hokkaido Univ. Ser. VI Zool. 13:458–64

57. Leadbeater E, Chittka L. 2005. A new mode of information transfer in foraging bumblebees? Curr. Biol.15:R447–48

58. Leadbeater E, Chittka L. 2007. The dynamics of social learning in an insect model, the bumblebee(Bombus terrestris). Behav. Ecol. Sociobiol. 61:1789–96

59. Lehrer M. 1997. Honeybee’s visual orientation at the feeding site. In Orientation and Communication inArthropods, ed. M Lehrer, pp. 115–44. Basel: Birkhauser

60. Lehrer M, Campan R. 2004. Shape discrimination by wasps (Paravespula germanica) at the food source:generalization among various types of contrast. J. Comp. Physiol. A 190:651–63

61. Lehrer M, Campan R. 2005. Generalization of convex shapes by bees: What are shapes made of? J. Exp.Biol. 208:3233–47

62. Liu G, Seiler H, Wen A, Zars T, Ito K, et al. 2006. Distinct memory traces for two visual features in theDrosophila brain. Nature 439:551–56

63. Lotto RB, Chittka L. 2005. Seeing the light: illumination as a contextual cue to color choice behavior inbumblebees. Proc. Natl. Acad. Sci. USA 102:3852–56

64. Macquart D, Latil G, Beugnon G. 2008. Sensorimotor sequence learning in the ant Gigantiops destructor.Anim. Behav. 75:1693–701

65. Masquelier T, Thorpe SJ. 2007. Unsupervised learning of visual features through spike timing dependentplasticity. PLoS Comput. Biol. 3:e31

66. Menzel R. 1967. Untersuchungen zum Erlernen von Spektralfarben durch die Honigbiene (Apis mellifica).Z. Vergl. Physiol. 56:22–62

67. Menzel R. 1968. Das Gedachtnis der Honigbiene fur Spektralfarben. I. Kurzzeitiges und langzeitigesBehalten. Z. Vergl. Physiol. 60:82–102

68. Menzel R. 1985. Learning in honey bees in an ecological and behavioral context. In Experimental Behav-ioral Ecology and Sociobiology, ed. B Holldobler, M Lindauer, pp. 55–74. Stuttgart, Ger.: Fischer

69. Menzel R. 1999. Memory dynamics in the honeybee. J. Comp. Physiol. A 185:323–4070. Menzel R, Backhaus W. 1991. Colour vision in insects. In Vision and Visual Dysfunction. The Perception of

Colour, ed. P Gouras, pp. 262–88. London: MacMillan71. Menzel R, Greggers U, Hammer M. 1993. Functional organization of appetitive learning and memory

in a generalist pollinator, the honey bee. In Insect Learning: Ecological and Evolutionary Perspectives, ed.D Papaj, AC Lewis, pp. 79–125. New York: Chapman & Hall

72. Menzel R, Steinmann E, de Souza J, Backhaus W. 1988. Spectral sensitivity of photoreceptors and colourvision in the solitary bee, Osmia rufa. J. Exp. Biol. 136:35–52

73. Miller GA. 2003. The cognitive revolution: a historical perspective. Trends Cognit. Sci. 7:141–44


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


74. Narendra A, Si A, Sulikowski D, Cheng K. 2007. Learning, retention and coding of nest-associated visualcues by the Australian desert ant, Melophorus bagoti. Behav. Ecol. Sociol. 61:1543–53

75. Nunez JA. 1982. Honeybee foraging strategies at a food source in relation to its distance from the hiveand the rate of sugar flow. J. Apicult. Res. 21:139–50

76. Pavlov IP. 1927. Lectures on Conditioned Reflexes. New York: International Publishers77. Peng Y, Xi W, Zhang W, Zhang K, Guo A. 2007. Experience improves feature extraction in Drosophila.

J. Neurosci. 27:5139–4578. Robertson I. 2001. Problem Solving. Hove, UK: Psychology79. Schroter U, Malun D, Menzel R. 2007. Innervation pattern of suboesophageal ventral unpaired median

neurones in the honeybee brain. Cell Tissue Res. 327:647–6780. Schubert M, Giurfa M, Francucci S, Lachnit H. 2002. Nonelemental visual learning in honeybees. Anim.

Behav. 64:175–8481. Shepard RN. 1987. Towards a universal law of generalisation for psychological science. Science 237:1317–

2382. Spaethe J, Tautz J, Chittka L. 2001. Visual constraints in foraging bumblebees: Flower size and color

affect search time and flight behavior. Proc. Natl. Acad. Sci. USA 98:3898–90383. Spence KW. 1937. The differential response to stimuli varying within a single dimension. Psychol. Rev.

44:430–4484. Srinivasan MV. 1994. Pattern recognition in the honeybee: recent progress. J. Insect Physiol. 40:183–9485. Srinivasan MV. 2010. Honey bees as a model for vision, perception, and cognition. Annu. Rev. Entomol.

55:267–8486. Srinivasan MV, Poteser M, Kral K. 1999. Motion detection in insect orientation and navigation. Vis.

Res. 39:2749–6687. Srinivasan MV, Zhang SW, Zhu H. 1998. Honeybees link sights to smells. Nature 396:637–3888. Stach S, Benard J, Giurfa M. 2004. Local-feature assembling in visual pattern recognition and general-

ization in honeybees. Nature 429:758–6189. Stach S, Giurfa M. 2005. The influence of training length on generalization of visual feature assemblies

in honeybees. Behav. Brain Res. 161:8–1790. Tang S, Guo A. 2001. Choice behavior of Drosophila facing contradictory visual cues. Science 294:1543–4791. Terrace HS, McGonigle B. 1994. Memory and representation of serial order by children, monkeys and

pigeons. Curr. Dir. Psychol. Sci. 3:180–8592. Tibbets EA. 2002. Visual signals of individual identity in the wasp Polistes fuscatus. Proc. Biol. Sci. 269:1423–

2893. Tibbets EA, Dale J. 2004. A socially enforced signal of quality in a paper wasp. Nature 432:218–2294. Troje F, Huber L, Loidolt M, Aust U, Fieder M. 1999. Categorical learning in pigeons: the role of

texture and shape in complex static stimuli. Vis. Res. 39:353–6695. Van Hateren JH, Srinivasan MV, Wait PB. 1990. Pattern recognition in bees: orientation discrimination.

J. Comp. Physiol. A 197:649–5496. Van Swinderen B, Greenspan RJ. 2003. Salience modulates 20–30 Hz brain activity in Drosophila. Nat.

Neurosci. 6:579–8697. Von Fersen L, Wynne CDL, Delius JD. 1990. Deductive reasoning in pigeons. Naturwissenschaften

77:548–4998. Von Frisch K. 1914. Der Farbensinn und Formensinn der Biene. Zool. Jb. Physiol. 37:1–23899. Wehner R. 1967. Pattern recognition in bees. Nature 215:1244–48

100. Wehner R. 1971. The generalization of directional visual stimuli in the honeybee, Apis mellifera. J. InsectPhysiol. 17:1579–91

101. Wehner R. 1972. Dorsoventral asymmetry in the visual field of the bee, Apis mellifera. J. Comp. Physiol.A 77:256–77

102. Wehner R. 1981. Spatial vision in arthropods. In Invertebrate Visual Centers and Behavior. Handbook ofSensory Physiology, ed. H Autrum, VII/6C:287–616. Berlin: Springer

103. Wehner R. 2003. Desert ant navigation: how miniature brains solve complex tasks. J. Comp. Physiol. A189:579–88


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.


104. Wilson EO. 1971. The Insect Societies. Cambridge, MA: Belknap Press105. Wolf E. 1933. Critical frequency of flicker as a function of intensity of illumination for the eye of the

bee. J. Gen. Physiol. 17:7–19106. Wolf E. 1934. Das Verhalten der Biene gegenuber flimmernden Feldern und bewegten Objekten.

Z. Vergl. Physiol. 20:151–61107. Wolf E, Zerrahn-Wolf G. 1935. The effect of light intensity, area, and flicker frequency on the visual

reactions of the honey bee. J. Gen. Physiol. 18:853–63108. Wolf R, Wittig T, Liu L, Wustmann G, Eyding D, Heisenberg M. 1998. Drosophila mushroom bodies

are dispensable for visual, tactile, and motor learning. Learn. Mem. 5:166–78109. Worden BD, Papaj DR. 2005. Flower choice copying in bumblebees. Biol. Lett. 1:504–7110. Xi W, Peng Y, Guo J, Ye Y, Zhang K, et al. 2008. Mushroom bodies modulate salience-based selective

fixation behavior in Drosophila. Eur. J. Neurosci. 27:1441–51111. Yang EC, Maddess T. 1997. Orientation-sensitive neurons in the brain of the honey bee (Apis mellifera).

J. Insect Physiol. 43:329–36112. Zeil J, Kelber A, Voss R. 1996. Structure and function of learning flights in ground-nesting bees and

wasps. J. Exp. Biol. 199:245–52113. Zentall TR, Galizio M, Critchfield TS. 2002. Categorization, concept learning and behavior analysis:

an introduction. J. Exp. Anal. Behav. 78:237–48114. Zhang S, Bock F, Si A, Tautz J, Srinivasan M. 2005. Visual working memory in decision making by

honey bees. Proc. Natl. Acad. Sci. USA 102:5250–55115. Zhang S, Srinivasan MV. 1994. Prior experience enhances pattern discrimination in insect vision. Nature

368:330–33116. Zhang SW, Bartsch K, Srinivasan MV. 1996. Maze learning by honeybees. Neurobiol. Learn. Mem.

66:267–82117. Zhang SW, Lehrer M, Srinivasan MV. 1999. Honeybee memory: navigation by associative grouping

and recall of visual stimuli. Neurobiol. Learn. Mem. 72:180–201118. Zhang SW, Srinivasan MV, Zhu H, Wong J. 2004. Grouping of visual objects by honeybees. J. Exp.

Biol. 207 3289–98


Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.

EN56-Frontmatter ARI 28 October 2010 7:29

Annual Review ofEntomology

Volume 56, 2011Contents

Bemisia tabaci: A Statement of Species StatusPaul J. De Barro, Shu-Sheng Liu, Laura M. Boykin, and Adam B. Dinsdale � � � � � � � � � � � � � 1

Insect Seminal Fluid Proteins: Identification and FunctionFrank W. Avila, Laura K. Sirot, Brooke A. LaFlamme, C. Dustin Rubinstein,

and Mariana F. Wolfner � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �21

Using Geographic Information Systems and Decision Support Systemsfor the Prediction, Prevention, and Control of Vector-Borne DiseasesLars Eisen and Rebecca J. Eisen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �41

Salivary Gland Hypertrophy Viruses: A Novel Group of InsectPathogenic VirusesVerena-Ulrike Lietze, Adly M.M. Abd-Alla, Marc J.B. Vreysen,

Christopher J. Geden, and Drion G. Boucias � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �63

Insect-Resistant Genetically Modified Rice in China: From Researchto CommercializationMao Chen, Anthony Shelton, and Gong-yin Ye � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �81

Energetics of Insect DiapauseDaniel A. Hahn and David L. Denlinger � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 103

Arthropods of Medicoveterinary Importance in ZoosPeter H. Adler, Holly C. Tuten, and Mark P. Nelder � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 123

Climate Change and Evolutionary Adaptations at Species’Range MarginsJane K. Hill, Hannah M. Griffiths, and Chris D. Thomas � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 143

Ecological Role of Volatiles Produced by Plants in Responseto Damage by Herbivorous InsectsJ. Daniel Hare � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 161

Native and Exotic Pests of Eucalyptus: A Worldwide PerspectiveTimothy D. Paine, Martin J. Steinbauer, and Simon A. Lawson � � � � � � � � � � � � � � � � � � � � � � � � 181

vii

Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.

EN56-Frontmatter ARI 28 October 2010 7:29

Urticating Hairs in Arthropods: Their Nature and Medical SignificanceAndrea Battisti, Goran Holm, Bengt Fagrell, and Stig Larsson � � � � � � � � � � � � � � � � � � � � � � � � � � 203

The Alfalfa Leafcutting Bee, Megachile rotundata: The World’s MostIntensively Managed Solitary BeeTheresa L. Pitts-Singer and James H. Cane � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 221

Vision and Visual Navigation in Nocturnal InsectsEric Warrant and Marie Dacke � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 239

The Role of Phytopathogenicity in Bark Beetle–Fungal Symbioses:A Challenge to the Classic ParadigmDiana L. Six and Michael J. Wingfield � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 255

Robert F. Denno (1945–2008): Insect Ecologist ExtraordinaireMicky D. Eubanks, Michael J. Raupp, and Deborah L. Finke � � � � � � � � � � � � � � � � � � � � � � � � � � � 273

The Role of Resources and Risks in Regulating Wild Bee PopulationsT’ai H. Roulston and Karen Goodell � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 293

Venom Proteins from Endoparasitoid Wasps and Their Rolein Host-Parasite InteractionsSassan Asgari and David B. Rivers � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 313

Recent Insights from Radar Studies of Insect FlightJason W. Chapman, V. Alistair Drake, and Don R. Reynolds � � � � � � � � � � � � � � � � � � � � � � � � � � � 337

Arthropod-Borne Diseases Associated with Political and Social DisorderPhilippe Brouqui � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 357

Ecology and Management of the Soybean Aphid in North AmericaDavid W. Ragsdale, Douglas A. Landis, Jacques Brodeur, George E. Heimpel,

and Nicolas Desneux � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 375

A Roadmap for Bridging Basic and Applied Researchin Forensic EntomologyJ.K. Tomberlin, R. Mohr, M.E. Benbow, A.M. Tarone, and S. VanLaerhoven � � � � � � � � 401

Visual Cognition in Social InsectsAurore Avargues-Weber, Nina Deisig, and Martin Giurfa � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 423

Evolution of Sexual Dimorphism in the LepidopteraCerisse E. Allen, Bas J. Zwaan, and Paul M. Brakefield � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 445

Forest Habitat Conservation in Africa Using Commercially ImportantInsectsSuresh Kumar Raina, Esther Kioko, Ole Zethner, and Susie Wren � � � � � � � � � � � � � � � � � � � � � � 465

Systematics and Evolution of Heteroptera: 25 Years of ProgressChristiane Weirauch and Randall T. Schuh � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 487

viii Contents

Ann

u. R

ev. E

ntom

ol. 2

011.

56:4

23-4

43. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by P

rof.

Dr.

Mar

tin G

iurf

a on

12/

11/1

0. F

or p

erso

nal u

se o

nly.

visual cognition in social...

Documents