Behavior and cognition: what a mini brain teaches us

The bee’s mini brain, which contains about 10⁶ cells (10¹¹ in humans), enables it not only to manage stereotyped behaviors such as foraging, but also—thanks to its plasticity—to adapt and allow the insect to respond to new problems through often complex learning. The bee’s brain is capable of providing “intelligent” solutions to a wide range of ecological or other problems, as is the case in vertebrates and in humans.

By Martin Giurfa

Insects have usually been regarded as simple little reflex machines. According to this particular view, their behavior would be governed essentially by stereotyped reactions, leaving little room for phenomena of plasticity. This view—which therefore excludes the possibility of addressing questions related to animal cognition in insects—has inspired a large body of work in robotics. It ignores, however, that insects, like most animals, process information from their environment in an adaptive and flexible way, which allows them to respond to a changing environment. It also ignores the remarkable evolutionary success of insects, a success that has enabled them to enter and conquer virtually all available habitats on the planet and to surpass all other multicellular organisms in absolute numbers and in number of species. These facts suggest that the insect brain must be capable of providing “intelligent” solutions to a wide range of ecological problems in order to ensure such success. These problems are similar or identical to those that other vertebrates—including humans—must face in their respective environments.

It is therefore relevant to ask to what extent insects are only capable of rigid behaviors and simple forms of learning, and whether additional levels of cognitive complexity involving non-elementary forms of learning must be acknowledged in order to explain the behavioral richness observed in insects. In brief, as in most animals, we will find in the bee elementary types of learning, i.e., simple associations linking two specific stimuli (for example, a neutral stimulus and a reinforcer) or a response and a reinforcer. By virtue of their very specificity, these associations are limited and do not allow the animal to cope with situations in which it is confronted with new stimuli. We therefore ask whether the plastic behavior observed in the bee can be explained exclusively in terms of elementary associations, or whether a higher level of complexity must be admitted in order to explain it.

In this chapter, we try to answer this question on the basis of studies carried out on an insect model, the honeybee Apis mellifera. This model appears particularly appropriate for this type of issue because it combines remarkable behavioral richness with a relatively simple nervous system that is accessible through new techniques allowing brain activity to be measured in vivo and during learning.

<h2>Bee behavior in a natural context</h2>

The behavioral richness of the bee (von Frisch 1967) justifies the use of this model in studies of insects’ cognitive abilities. The bee lives in society and cannot survive long in the absence of the other members of the hive. Despite its small size, it is able to navigate efficiently over distances of up to about ten kilometers between the hive and food sources, i.e., flowers. It organizes its nectar- and/or pollen-foraging activities according to an extremely rapid and efficient “assembly-line” work pattern: it visits and exploits successively flowers belonging always to the same species; when that species no longer offers nectar or pollen, it turns to another species. This is the phenomenon of “floral constancy.” Sensory capacities and motor performances are highly developed. Bees see the world in color, perceive and discriminate shapes and patterns, and are able to detect motion with very high temporal resolution. Their olfactory sense allows them to distinguish a wide spectrum of odors, and their mechanosensory perception is also extremely rich thanks to the presence of thousands of mechanosensory hairs around the insect’s body and internal proprioceptors. Complex behaviors may have innate bases (for example, the construction of comb cells) or depend entirely on experience (for example, the effective manipulation of certain floral structures in order to extract their pollen and nectar). Natural selection has highlighted in particular the learning of specific information characteristic of interesting locations, namely the hive and food sources. Learning landmark cues and celestial information used in the context of navigation (azimuthal position of the sun, pattern of polarized light in the sky) guarantees return to the nest and thus optimizes foraging efficiency. Bees communicate the presence of food sources around the hive through ritualized body movements known as the “waggle dance,” a communication system that conveys information about the direction and distance of the exploited food source (von Frisch 1967). Bees that follow a forager’s dance in the darkness of the hive and on the vertical surface of the combs where they are located obtain from the dancer’s movements the information needed to find the exploited food source: the speed of the movements informs them about the distance, and the angle of the waggle phase relative to the vertical informs them about the flight direction relative to the sun. Under natural conditions, several dances occur in parallel within the colony. Different decision-making processes at the individual and collective levels can be activated and become effective on the basis of partial knowledge of the full range of available options.

<h2>Bee behavior in an experimental context</h2>

Understanding and analyzing the mechanisms underlying these complex behaviors require an experimental approach in which the essential traits of bee behavior are preserved while, at the same time, the number of variables is considerably reduced. Different approaches have been developed in this spirit, starting from the pioneering work of Karl von Frisch, who was the first to establish experimental scientific work devoted to understanding bee behavior (1967). Two essential paradigms can be cited here because of their importance in the study of different facets of learning in the bee: 1) olfactory conditioning of the proboscis extension response in immobilized bees and 2) conditioning of the approach flight of freely flying bees to an explicitly rewarded visual stimulus.

Olfactory conditioning of the proboscis extension response

Immobilized bees can be conditioned to respond to olfactory stimuli (Takeda 1961; Bittermann et al. 1983). In this paradigm, each bee is fixed in a small metal tube such that only its head protrudes. The only movements available to the insect are thus those of the antennae and the mouthparts (mandibles and proboscis). The bees’ antennae are their chemosensory organs; when the antennae of a hungry bee are touched with a sugar solution (for example with a toothpick soaked in sugar solution), it extends its proboscis in order to reach and lick the sugar solution (Fig. 1).

Figure 1. Classical olfactory conditioning of the proboscis extension response in the honeybee. When the antennae of a hungry bee immobilized in a metal tube come into contact with a drop of sugar solution, the insect extends its proboscis and drinks the solution. Odors delivered to the antennae do not elicit this response in naive animals (left: before conditioning). If, however, an odor precedes the arrival of the sugar solution (forward conditioning), an association is created that allows the odor to elicit the proboscis extension response (PER) in subsequent tests (middle). The acquisition curves (right) show a differential conditioning experiment with two odors; one odor is paired with the sugar solution (CS+) whereas the other odor is never paired (CS-). CS+ and CS- trials are interleaved. Bees learn to respond to CS+ and not to CS-.

Odors delivered to the antennae do not elicit this response in naive animals. If, however, an odor precedes the arrival of the sugar solution (forward conditioning), an association is created that allows the odor to elicit the proboscis extension response (PER) in subsequent tests. This effect is clearly associative and constitutes an example of classical (Pavlovian) conditioning because the contingency learned by the bee associates two stimuli (Bittermann et al. 1983): the odor as the conditioned stimulus (CS) and the sugar solution as the reinforcing or unconditioned stimulus (US).

This preparation offers an additional advantage beyond its simplicity for studying learning: it allows the physiological bases of olfactory learning to be studied. It is indeed possible to visualize the brain of the immobilized bee in the metal tube through an opening in the cuticle of the head capsule. This capsule is made of chitin and is not innervated, so the procedure is not debilitating for the animal, which will learn in the same way as non-operated animals. It is therefore possible to visualize brain activity in vivo while the bee learns. Physiological correlates of different forms of olfactory learning can be identified at different levels, ranging from the molecular and biochemical to specific neurons or neuronal ensembles whose activity can be revealed using imaging techniques.

Conditioning of bees’ approach flight to an explicitly rewarded visual stimulus

Freely flying bees can be conditioned to visual stimuli such as colors, shapes, and patterns (von Frisch 1967). In this paradigm, each bee is individualized (for example, with colored marks on the thorax) and trained to fly to the experimental site where it is rewarded with a drop of sugar solution provided it lands on the appropriate visual stimulus. The contingencies established in this context associate visual stimuli (CS) and sugar-solution reinforcement (US), but also the animal’s response (flying toward, landing on a visual target) and the reinforcement; bees thus learn that a specific visual information (the color, for example) is associated with a sugar-solution reward and that they must land on that color in order to obtain the reward.

Studies using this behavioral paradigm have been able to identify the visual information learned by bees. Bees learn to associate all colors in their visual spectrum (from 300 nm to 650 nm) with reinforcement, but different learning rates are observed for different colors. They can also learn to discriminate different shapes, silhouettes, and visual patterns, although a larger number of learning trials is required compared with the case of color. Bees are able to recognize visual stimuli based on their position in the visual field, their spatial orientation, their geometry, their size, their spatial frequency, their depth, their motion contrast, and their symmetry. Although for obvious reasons it is not possible to perform physiological studies of brain activity in a freely flying insect, this paradigm has the advantage of exploiting the behavioral richness associated with free flight, which is reduced when the animal is immobilized in the olfactory conditioning paradigm of the proboscis extension response.

<h2>Problem solving in the context of visual discrimination</h2>

Beyond simple associations between color and reward or between shapes and reward that underlie several visual discriminations, bees are capable of visual discrimination performances that allow them to respond adaptively to stimuli they have never seen before. This discriminatory behavior relies on the ability to categorize visual stimuli. Categorization of visual stimuli is a behavioral task that has been studied and demonstrated mainly in vertebrates, in particular in those characterized as being especially apt at solving complex learning tasks (pigeons, dolphins, primates). Recently, however, it has been shown that bees are also capable of categorizing visual stimuli. Typically, a categorization experiment poses to an animal a discrimination or choice problem in which reinforcement is not signaled by a unique stimulus, but rather by a variety of stimuli sharing one or more common characteristics. The animal must be able to extract the reinforced characteristics in order to group the stimuli into the relevant categories. The experiment must also be able to show a transfer to new instances, i.e., the animal trained in this type of problem must be able to categorize stimuli it has never encountered if they present the typical characteristics of the reinforced category. It is therefore obvious that this type of performance cannot fit within the framework of simple elementary associations, because these would not suffice to explain transfer to new stimuli for which the animal has no explicit experience.

Categorization in bees has been well studied in the case of bilateral symmetry. It has been shown that bees trained with a changing succession of visual stimuli so as to reinforce bilaterally symmetric stimuli and not asymmetric ones (or vice versa) learn this abstract information (symmetry vs. asymmetry) and can transfer it to new symmetric and asymmetric stimuli they have never seen (Giurfa et al. 1996). Bees are thus clearly capable of categorizing visual stimuli because they group new objects into well-defined categories resulting from their training. Transfer to new stimuli that were very different from the stimuli used during training—symmetry or lack of symmetry aside—suggests that at some stage, abstraction processes must occur in the bee’s brain.

A more spectacular form of non-elementary learning is revealed in experiments in which bees are trained according to the scheme of the task called “delayed matching-to-sample” (Giurfa et al. 2001). In this experiment, an animal is typically confronted with a sample and then with a series of stimuli among which one is identical to the sample. The animal must learn always to choose the stimulus identical to the sample regardless of the particular physical characteristics of the sample, which also changes regularly. The animal must therefore learn the rule: “always choose what you are shown as a sample regardless of the sample’s particular properties.”

In order to determine whether bees can learn such an equivalence principle, they are trained with a changing and non-reinforced sample at the entrance of a Y-maze (the sample can be either a blue or yellow colored disk, or a disk with vertical or horizontal black-and-white bars)  (Giurfa  et  al. 2001).   Bees receive a sugar-solution reward if and only if they choose inside the maze the stimulus that matches the sample seen at the entrance. Bees trained with the yellow and blue colors are presented with vertical or horizontal bar stimuli they have never seen before (Fig. 2). Likewise, other bees trained with vertical or horizontal bar stimuli are presented with yellow or blue disks they do not know. In both cases, bees chose inside the maze the stimulus matching the sample shown, even though they were confronted for the first time with that type of sample. They are thus capable of solving the “delayed matching-to-sample” experiment and of constructing an equivalence principle that goes beyond specific stimuli.

In the same way, the ability to establish a difference principle has also been demonstrated in bees (Giurfa et al. 2001). In this case, the experiment used is “delayed non-matching-to-sample,” in which the animal must always choose the stimulus opposite to the sample presented to it. Bees can also solve this type of problem, which highlights their capacity to produce sophisticated behaviors beyond simple elementary associations.

Figure 2: Bees can learn an equivalence principle. Acquisition performance and transfer of bees in a “delayed matching-to-sample” experiment in which they are trained with colors (Experiment 1) or with patterns of black and white stripes, vertical or horizontal (Experiment 2). a) Acquisition: the curves show acquisition performance grouped in blocks of ten consecutive visits to the apparatus for each experiment. (b,c) Transfer tests: (b) In Experiment 1, bees trained with colors were tested with black-and-white bar patterns. (c) In Experiment 2, bees trained with patterns were tested with colors. In both cases, bees choose inside the maze the new stimulus matching the sample presented at the entrance, even though they have no explicit experience with these new stimuli. n: number of choices (Giurfa et al. 2001).

<h2>The bee’s mini brain: design and cognitive architecture</h2>

The behavioral richness and sophistication revealed in the bee originate in a mini brain with a volume of 1 mm3 and containing 960,000 neurons (Fig. 3). The description of the neuronal organization of this brain can be simplified by recognizing three fundamental organizational principles:

<ol>
 <li>Specialized neuropils. As in any brain, regions of nervous tissue specialized (neuropils) in processing specific sensory information (vision, olfaction, etc.) can be easily identified in the bee’s brain.</li>
 <li>
 Specialized neurons. In the bee’s brain, as in the brains of other insects, it is possible to identify single neurons that can be recognized recurrently from one bee to another and within the same bee because of their unique morphology and their unique function in defined sensorimotor routines. This is a specificity of the invertebrate brain, because such recurrent identification of several individual neurons proves technically difficult in vertebrates.
 </li>
 <li>
 Higher-order integration centers. As in other brains, centers where several sensory information processing pathways converge can be identified in the bee’s brain.  These structures are thus multimodal integration centers enabling different stimuli to be associated during learning and memorization. What makes them particularly interesting is the possibility they offer as a substrate for non-elementary forms of plasticity in which, for example, transfer between different sensory modalities could be manifested (see above).
 </li>
</ol>

Figure 3. Three-dimensional reconstruction of a bee brain in frontal view made using confocal microscopy techniques. Several neuropils are indicated: ME: medulla; LO: lobula; ME + LO + Lamina (not visible) constitute the optic lobes, the brain region where primary processing of visual information occurs; AL: antennal lobe, the primary olfactory neuropil; PL: protocerebral lobe, a neuropil whose function is poorly known; SOG: subesophageal ganglion, a brain region related to gustatory inputs; CB: a brain region related to motor responses. The two prominent and symmetrical structures occupying the central part of the brain are the mushroom bodies. Each mushroom body consists of two subunits, the calyces, lateral (LC) and medial (MC). The calyces constitute the input region of the different sensory pathways (vision, olfaction). Lobes a and b constitute the output region. Scale = 200 µm.

Different approaches can be used to study cognition in the bee’s mini brain at these three levels. Three examples are given below, one for each level.

<h3>1. Specialized neuropils</h3>

The case of the antennal lobe The antennal lobes  are  the  primary  olfactory  neuropils  in the bee’s brain. Their function is to  process  and  encode  olfactory  information  coming from olfactory receptors on the antennae (Fig. 4).

Figure 4. a) The antennal lobe is the primary olfactory neuropil in the bee’s brain. Two antennal lobes (one in each cerebral hemisphere) can be identified in the bee’s brain. Olfactory glomeruli are the functional units of the antennal lobe. These glomeruli constitute regions of convergence of the dendritic endings of olfactory receptors and projection neurons to higher centers. Lateral connections between glomeruli are also ensured by local interneurons. Each antennal lobe is made of 160 glomeruli. Calcium imaging makes it possible to record the spatial patterns of glomerular activation when an odor stimulates a bee’s antenna. Each odor determines a specific spatial pattern of glomerular activation. Colors are used to indicate the activation level, with red corresponding to the maximum and blue to the minimum. b, c) Activation patterns corresponding to pentane and 2-heptanone, respectively.

The bee brain has two antennal lobes (one per antenna or cerebral hemisphere). The antennal lobes are the equivalent of the olfactory bulb in mammals; these two structures share common principles of architecture and function. Both are composed of multiple glomeruli (160 in the case of the antennal lobe). In the antennal lobe, glomeruli are globular structures that constitute regions of convergence of the dendritic endings of olfactory receptors located on the bee’s antennae and of projection neurons that transfer information to higher centers (mushroom bodies and protocerebral lobes); lateral connections between glomeruli are ensured by local interneurons (Galizia and Menzel 2000). In order to study olfactory coding at the level of the antennal lobe, a combination of neuroanatomical studies and brain imaging was adopted (Joerges et al. 1997). Neuroanatomical studies established a precise map of the antennal lobe, allowing individual glomeruli to be identified based on their shape and position. Imaging studies made it possible to understand how odors are coded and represented at the level of the bee brain.

The basic principle underlying calcium imaging used in antennal lobe studies is the release of calcium by excited cells. Thus, fluorescent products that bind to calcium (“calcium sensitive dyes”) are used to bathe the brain of a bee stimulated with selected odors. When neurons are excited by olfactory stimulation, they release calcium that binds to the fluorescent products. This determines a change in fluorescence, detectable by a CCD camera directed at the antennal lobe. In this way, it is possible to visualize brain activity while the insect smells odors. These studies have revealed the mechanisms of olfactory coding in naive bees: each odor is coded in terms of a spatial pattern of glomerular activation (Fig. 4). When two odors are presented in a mixture, the glomerular representation is similar, but not identical, to the sum of the glomerular representations corresponding to the individual component odors. As components are added to the mixture, the spatial pattern changes and inhibitory interactions become increasingly evident.

Since calcium imaging is already established for the antennal lobe, the question that must now be answered concerns the role of experience in the potential modification of glomerular activation patterns. How does olfactory learning modify neural representations of odors at the level of the antennal lobe? Do different types of olfactory learning correspond to different glomerular representations for the same odor? A first study (Faber et al. 1999) showed that elementary differential learning (the bee must learn to react to a reinforced odor and to inhibit its reaction to a non-reinforced odor) produces qualitative but not quantitative changes in the representation of the reinforced odor. The glomerular activation pattern thus remains the same for the reinforced odor; only the activation intensity increases. No significant changes are detected in the case of the non-reinforced odor.

<h3>2. Specialized neurons: the case of neuron VUMmx1</h3>

VUMmx1 is an identifiable neuron in the bee brain whose name corresponds to its neuroanatomical location (the name corresponds to the initials for the “ventral unpaired median neuron of maxillary neuromere 1”). The structure of this neuron is particularly remarkable: the dendrites of VUMmx1 form symmetrical arborizations in the brain and converge with the olfactory processing circuit at three sites: 1) the antennal lobes, 2) the calyces of the mushroom bodies, and 3) the protocerebral lobe. The main feature of VUMmx1 is related to its excitation whenever the bee’s antennae or proboscis are stimulated with sugar solution (Hammer 1993). This property suggests that the activity of this neuron represents the sugar-solution reinforcement in the bee brain (Hammer 1993).

To test this hypothesis, Hammer (1993) carried out an olfactory conditioning experiment of the proboscis extension response in which sugar-solution reinforcement is replaced by an artificial activation of neuron VUMmx1, generated by intracellular injection of electrical current. In this conditioning experiment, the artificial depolarization of the neuron immediately follows olfactory stimulation so as to reproduce the temporal characteristics of forward conditioning, in which the CS always precedes the US. It is therefore a “virtual” olfactory conditioning experiment: the immobilized bee is stimulated with an odor followed by artificial activation of the neuron. After this pairing, one observes whether the odor alone is capable of triggering proboscis extension afterward. To avoid artifacts related to the movement of the proboscis itself, the proboscis is cut and the response is measured in terms of the activity of muscle M17, a muscle at the base of the proboscis that precisely controls its extension. If the activity of the neuron effectively represents sugar-solution reinforcement, the bee must learn to react to the odor in this experiment even though it has never received a specific reinforcement. This is indeed the result found. Bees do not react to the odor before pairing, which is logical because the odor is unknown, but show a strong reaction after pairing, i.e., after virtual conditioning (Hammer 1993). Their reaction is identical to that of a group of bees in which conditioning is normal, i.e., in which the odor has been paired with real sugar solution. These results therefore show that VUMmx1 constitutes the neuronal correlate of sugar-solution reinforcement in the bee brain. It is a specialized neuron whose function is to serve as a reinforcer for associative olfactory learning.

Figure 5. A single neuron, VUMMx1, represents the sugar-solution reinforcer in the bee brain. The diagram on the right shows a bee brain (without optic lobes) with the olfactory circuit. The diagram on the left shows, at the same scale, the morphology of neuron VUMmx1. The dendrites of VUMmx1 form symmetrical arborizations in the brain and converge with the olfactory circuit at three sites (delimited by a red dotted line): the antennal lobes (AL), the calyces of the mushroom bodies (MB), and the protocerebral lobe (PL). SO: subesophageal ganglion; PN: projection neurons; a and b: mushroom body lobes. The neuron responds to stimulation of the antennae and the proboscis with sugar solution.

The case of VUMmx1 illustrates how the electrophysiological approach makes it possible to characterize a specialized neuron. This approach similarly made it possible to characterize other specialized neurons.

<h3>3. Higher-order integration centers: the case of the mushroom bodies</h3>

The mushroom bodies are central and prominent structures occupying nearly one third of the bee brain (Fig. 3). Each mushroom body consists of 170,000 densely packed neurons, the Kenyon cells, and has two subunits fused at their base in a common peduncle. The calyces constitute the input region of the mushroom bodies; a lateral calyx and a medial calyx can be distinguished. Each calyx is divided into three compartments: the lip region, the collar region, and the basal ring region. Each of these compartments receives specific sensory afferents (lip: olfactory; collar: visual; basal ring: olfactory). The efferent region of the mushroom bodies consists of lobes a and b resulting from the fusion of the medial and lateral calyces. The efferent neurons respond to several types of sensory stimulation and are therefore multimodal. This shows that the mushroom bodies are sensory integration centers that could constitute an ideal substrate for transfer between modalities and for non-elementary forms of learning. A new method now makes it possible to study the role of the mushroom bodies in different forms of learning and to test this hypothesis. This method consists in producing adult bees with specific lesions at the level of the mushroom bodies (Fig. 6). These lesions are obtained by treating first-instar larvae with hydroxyurea, a substance that inhibits the mitotic activity of dividing cells (neuroblasts) that will give rise to the mushroom bodies (Malun 1998). Hydroxyurea is given to the larvae in the royal jelly with which they are fed during the period of division of these neuroblasts. Adults resulting from this treatment cannot be distinguished externally from normal bees. However, a post-conditioning study makes it possible to identify partial lesions of the mushroom bodies in the brains of these bees. Generally, the treatment results in the ablation of one or two medial calyces. Hydroxyurea-treated bees can be studied in learning tasks of varying complexity in order to determine the importance of the mushroom bodies in solving elementary and non-elementary problems. So far, results show that partial ablation of the mushroom bodies does not affect the ability to solve elementary problems involving a simple association between a CS and a US. This holds for tactile learning paradigms (Scheiner et al. 2001) and for olfactory differential conditioning (Malun et al. 2002). Conditioning with a reinforced odor and a non-reinforced odor delivered to a single antenna does not affect the ability to learn to react (proboscis extension) to the reinforced odor and not to the non-reinforced odor, even if the conditioned antenna corresponds to the lesioned brain hemisphere (Malun et al. 2002). These results show that the mushroom bodies are not necessarily involved in establishing elementary associations. These associations could be established earlier at the level of the antennal lobes.

Figure 6: a–e) Three-dimensional reconstructions of hydroxyurea-treated bee brains in frontal view. a) Brain of a hydroxyurea-treated bee showing no ablations; the brain is identical to that of a normal individual (compare with Fig. 3). b) Unilateral partial lesion: the medial calyx on the left side is absent whereas the lateral calyx is intact. c) Unilateral partial lesion: the medial and lateral calyces on the right side have disappeared; the left side is intact. d) Bilateral partial lesion: both medial calyces are absent. e): Total lesion: only a reduced group of cells appears in the right hemisphere; calyces and lobes are absent. Bees presenting this lesion have an extremely low survival rate and cannot be used in conditioning experiments.

<h2>Conclusion: mini brain and cognition</h2>

The combination of behavioral and neurobiological studies makes it possible to appreciate the extent to which plasticity in bees surpasses simple forms of elementary learning and to identify the underlying neuronal substrates of these abilities. The question raised in this chapter was the possible reduction of the bee’s plastic behavior to a collection of simple learning forms that would assemble as unconnected modules. Several examples show that such a reduction is incorrect. Bees are capable of non-elementary forms of learning that reproduce higher-order cognitive performances known in vertebrates, such as contextual learning, visual categorization, and learning of relational rules (Menzel and Giurfa 2001). The bee’s cognitive architecture consists of a complex network of interconnected modules (specialized neurons and neuropils, higher-order integration centers) ensuring stereotyped and flexible behaviors. Within this network, central integration centers enable consultation, comparison, and cross-talk between modules, which allows transfer of solutions from one situation to another and thus the generation of new responses. The bee brain therefore seems capable of extracting the logical structure of its world very efficiently. What are the specific limitations of this brain when compared to vertebrate brains, and what structural and functional bases are responsible for these limitations? To answer this question, it is necessary to study in depth deficiencies in learning in the bee, an area of research that is still little explored.

Studies on behavior and neurobiology in the bee allow an optimistic stance toward these questions. The bee can serve as a model for understanding intermediate levels of cognitive complexity and for identifying their neuronal substrates. It therefore deserves a justified place in the field of cognitive neuroscience. The bee’s mini brain with its 960,000 neurons has not yet revealed its full potential. The convergence of new questions that seemed unthinkable some time ago and new measurement techniques allowing access to different levels of neuronal organization makes it possible to hope that the best remains to be discovered.

<a href="?action=get_file&id=106&resource_link_id=167f" target="_blank">Download the article as a PDF</a>

See also:

► <a href="https://www.2imanagement.ch/fr/divers/liens/wwwapisavoirch/lintelligence-individuelle-de-labeille">The individual intelligence of the bee</a> 
► <a href="https://www.2imanagement.ch/fr/divers/liens/wwwapisavoirch/les-sens-loges-dans-les-antennes-de-labeille">The senses located in the bee’s antennae</a> 
► <a href="https://www.2imanagement.ch/fr/divers/liens/wwwapisavoirch/mini-cerveau-mega-performances">Mini Brain, Mega Performances</a>

References:

[1] Bitterman, M. E., Menzel, R., Fietz, A., et Schäfer, S. (1983) Classical conditioning of proboscis  extension in honeybees (Apis mellifera). Journal of  Comparative Psychology 97, 107-119.

[2] Faber, T., Joerges, J., et Menzel, R. (1999) Associative learning modifies neural representations of odors in the insect brain. Nature neuroscience 2, 74- 78.

[3] Frisch, K. von (1967) The Dance Language et Orientation of Bees. Harvard University Press, Cambridge.

[4] Galizia, C. G. et Menzel, R. (2000) Odour perception in honeybees: coding information in glomerular patterns. Current Opinion in Neurobiology 10, 504-510.

[5] Giurfa, M.,  Eichmann,  B.,  et  Menzel,  R. (1996) Symmetry perception in an insect. Nature 382, 458-461.

[6]  Giurfa,  M.,  Zhang,  S.  W.,  Jennett,   A., Menzel, R. et Srinivasan, M. V. (2001) The concepts of sameness et difference in an insect. Nature 410, 930-933.

[7] Hammer, M. (1993) An identified neuron mediates the unconditioned stimulus in associative olfactory learning in honeybees. Nature 366, 59-63.

[8] Joerges, J., Küttner, A., Galizia,  C.  G.,  et Menzel, R. (1997) Representation of odours et odour mixtures visualized in the honeybee brain. Nature 387, 285-288.

[9] Malun, D. (1998) Early  development  of mushroom bodies in the brain of the honeybee Apis mellifera as revealed by BrdU incorporation et ablation experiments. Learning et Memory 5, 90-101

[10] Malun, D., Giurfa, M., Galizia, G., Plath, N., Brandt, R., Gerber, B., Eisermann, B. (2002) Hydroxyurea-induced partial mushroom body ablation does not affect acquisition et retention of olfactory differential conditioning in honeybees. Journal of Neurobiology 53, 343-360.

[11] Mauelshagen, J. (1993) Neural correlates of olfactory learning in an identified neuron in the honey bee brain. Journal of Neurophysiology 69, 609-625.

[12] Menzel, R. et  Giurfa,  M.  (2001)  The cognitive architecture of a mini brain : the honey bee.Trends in Cognitive Sciences 5:62-71

[13] Takeda, K. (1961) Classical  conditioned response in the honey bee. Journal of Insect Physiology 6, 168-179.

Author

Martin Giurfa; 2003

L'étude en pdf

Add comment

Observations en temps réel

Register for newsletter

Behavior and cognition: what a mini brain teaches us