The attraction of Africanized honey bees to soft drinks and perfumes

The attraction of Africanized honey bees to soft drinks and perfumes – Apis mellifera L

Charles I. Abramson

The proboscis-extension reflex has become invaluable in studying a wide range of behavioral, genetic, and neurobiological phenomena in European honey bees (EHBs). The reflex is studied by confining bees in small metal harnesses. One or more presentations of an odor and sucrose feeding increases the frequency of background emissions of proboscis extension to odor (Smith & Abramson, 1992). The development of this Pavlovian conditioning procedure has made possible a variety of sophisticated experiments on honey-bee learning (Abramson, 1994). The experiments discussed in this article were designed to test the feasibility of using the proboscis extension reflex to study preferences in Africanized honey bees (AHBs).

In designing the present experiments, we selected the proboscis-conditioning method over the more naturalistic free-flying method – in which bees fly from the hive to the laboratory, where they can distinguish targets based on color, odor, or position – because the proboscis method permits better control of training variables known to influence behavior. The intertrial interval (ITI) and stimulus duration, for example, are better controlled in harnessed bees than in bees trained to fly to the laboratory on their own accord. Moreover, if large numbers of subjects are to be examined, the proboscis technique is more cost-effective in terms of time. Typically, only 1 or 2 free-flying bees can be trained each day. With harnessed bees, an individual researcher can test as many as 12 at a time. Because the bees are restrained, the proboscis technique is also better suited for quantitative physiological and biochemical analysis of compounds contained in perfumes, colognes, cosmetics, soft drinks, and other consumer goods that may attract bees. The technique also lends itself nicely to testing the attractiveness of other commercial products such as hair spray and shampoo.

The stimuli selected to serve as the initial models for the proboscis-conditioning preference tests were perfumes and soft drinks. We selected those stimuli because they represent common consumer products with which bees readily come in contact.(1) In addition, perfumes and soft drinks can be accommodated easily into the existing proboscis-conditioning method as conditioned stimuli (CS) and unconditioned stimuli (US), respectively.

Studying the attractiveness of consumer products to honey bees is important from both a medical and a psychological perspective. Approximately 0.4% of people in the United States have severe or even fatal allergic reactions to bee stings (Reisman, 1992). Even among some nonallergic individuals, the thought of a bee sting is enough to produce a fear of bees, technically known in the psychological literature as Melissophobia (Halohen & Santrock, 1996; Matchett & Davey, 1991). Of the 12.5% of Americans who are stricken with some type of phobia (Regier et al., 1988), 5% to 15% suffer from animal phobias, including a fear of insects (Zimbardo, 1992). In 1995 alone, 17,874 persons required some type of medical treatment following attacks by bees, wasps, and hornets. That number increases to 72,582 when other invertebrates are considered (Litovitz, Felberg, White, & Klein-Schwartz, 1996). Anecdotal evidence suggests that the threat of a bee sting is reduced by wearing light colors; not wearing perfume, cologne, cosmetics, or hair spray; and avoiding quick movements when around bees (Free, 1961). However, no one has reported whether AHBs prefer some perfumes or cologne to others or demonstrate preferences to particular soft drinks, cosmetics, hair spray, or other consumer products. Such information would provide health-care professionals specializing in insect probias or allergic reactions to insect venom with useful information.

Studying the preferences of bees to soft drinks is also interesting from a methodological perspective. Most of what is known about proboscis conditioning in honey bees has been learned from studies using only sucrose solution as a US (Abramson, 1994; for an exception, see Smith, Abramson, & Tobin, 1991). We were unable to find any reports of proboscis-conditioning studies that involved naturally occurring US. The absence of research with naturally occurring US leaves open the possibility, however unlikely, that proboscis conditioning is only a laboratory phenomenon that depends heavily on high concentrations of sucrose for successful conditioning.

Informal observation of trash cans and recycling bins in the Stillwater, Oklahoma, area suggests that honey bees are attracted not only to perfumes but also to residual soft drinks left in discarded cans. We predicted that bees will learn to associate antecedent stimuli with the consumption of a soft drink in much the same way that a bee associates the odor of a flower with nectar feeding.

The necessary first step in the study of AHB preferences was to determine if the proboscis technique developed for the study of behavior in EHBs could be used. We focused on three questions: (a) Can AHBs associate a perfume CS with a sucrose feeding? (b) Do AHBs have soft-drink preferences? (c) Can AHBs learn to associate an olfactory stimulus with a soft-drink feeding?

Experiment 1: Classical Conditioning With Perfumes

Our purpose in this experiment was to determine whether perfumes could serve as a CS. Perfumes are usually composed of one or more floral essences (Bauer, Garbe, & Surburg, 1990), the number of which depends upon the type of fragrance. Other, nonfloral essences are also included in perfumes. Perfumes are classified according to their overall smells (also known as notes), such as woodsy, spicy, floral, sweet, and green (Bauer et al., 1990). Because designer perfumes are composed of costly floral essences and other ingredients, most tend to be expensive. Some perfume manufactures have attempted to offer what are known as “designer impostor” perfumes, which smell similar to the designer brands but are much less expensive. Although they share the same notes, the originals and their “impostors” contain completely different ingredients. The main reason for the differences in price is that the inexpensive perfumes contain synthetic ingredients rather than the costly floral oils. Most people can tell only a slight difference between real fragrances and their impostors. Although not the object of the present study, this difference may be detected by bees, because the synthetic chemicals may be sensed differently than the floral oils in fragrances.

We chose two perfumes: Realm for Men and Realm for Women (manufactured by EROX Corporation, Fremont, CA). We chose those two because of their popularity (they are reported to attract members of the opposite sex) and because they are the only perfumes containing synthesized human pheromones. Realm for Women is a mixture of spices, fruits, florals, and woods with a faint scent of honey and vanilla. Realm for Men is a musky scent composed of spices, fruit oils, woods, and patchouli. To control for nonassociative effects, we used a differential conditioning procedure in which the two perfumes served as CS – one of which was paired with a sucrose feeding. Discrimination is one procedure used to assess the contribution of sensitization and has the virtue of requiring fewer animals per group because each animal serves as its own control.


Subjects. In mid-July 1996, 24 ABHs were captured in the apiary of the Laboratorio Apicola of the Universidade Federal da Paraiba (UFPB), Bananeiras, Brazil. The Bananeiras UFPB campus is an agricultural university located near the Atlantic forest in northeastern Brazil. The university maintains lemon and orange groves about 400 m from the hives. The vicinity of the laboratory is heavily populated with roses, daisies, and sunflowers in addition to the citrus groves.

Africanized worker bees were collected in glass vials as they departed from the hive at about 3 p.m. on the day before we were to conduct the study. Collection of subjects in this manner yields a mixture of bees of different behavioral specializations. Individual subjects were rendered unconscious in the laboratory by placing the glass vials in an ice-water bath. When the bee became inactive, it was immediately removed from the vial and put into the restraining harness. To secure the bee in the harness, we placed a strip of duct tape between the head and the thorax and fastened the tape to the sides of the harness. After regaining consciousness, subjects were fed a 2.9-M sucrose solution until full, after which they were left until the next morning. Training began at 7:30 the following morning (for details, see Abramson, Aquino, Silva, & Price, 1997).

Apparatus. Materials consisted of the metal harnesses in which the bees were held, a ventilation chamber to prevent the accumulation of the CS scent in the testing area, and a plastic 20-cc syringe used to present the olfactory CS. In addition, filter paper strips dipped into a jar of 2.9-M sucrose solution were used to apply the unconditioned stimulus. The CS delivery device was prepared by spraying a 1-s burst of perfume onto a piece of 1-[cm.sup.2] filter paper secured to the tip of the syringe’s plunger with an uncoated thumbtack. Unfortunately, it was not possible to express the concentration of Realm for Men and Realm for Women in terms of molarity, because the composition of the individual components is not publicly available.

Procedure. We divided the 24 animals into two groups of 12. One group of animals received a CS+ of Realm for Men, followed by a feeding, and a CS- of Realm for Women, not followed by a feeding. For the second group, the CS+ consisted of Realm for Women, followed by a sucrose feeding, and a CS- of Realm for Men, not followed by a feeding. CS presentation was pseudorandom. For half of the animals, the presentation of the CSs consisted of three successive sequences of CS+ CS- CS- CS+ CS- CS+ CS+ CS-. The remaining animals were exposed to three successive sequences of CS- CS+ CS+ CS- CS+ CSCS- CS+. Each animal received 12 CS+ and 12 CS- trials. The CS-US interval was 2 s.

Upon termination of the CS, we presented the US by first touching sucrose to the antenna and then to the then-extended proboscis for a 3-s feeding. The intertrial interval was 5 min. We selected that interval because we found it to be effective in previous experiments (Abramson et al., 1997), and it has the virtue of maintaining the interval between CS+ and CS- at approximately 10 min. The 10-min intertrial interval is commonly used in honey bee proboscis-conditioning experiments (Abramson, 1994).

To control for calendar variables, we ran the bees in squads containing animals from both groups. The number of bees in each squad varied from day to day. Conditioned responses to the CS were visually categorized into one of two states during each trial. If a subject extended its proboscis after the onset of the CS but before its antennae were touched with the sucrose, a response was registered. Otherwise, a nonresponse was recorded.

Members of each squad were treated sequentially. Bees in their individual tubes were placed at the entrance of an exhaust duct designed to remove the olfactory stimulus from the training area. Following a trial, the bee was returned to the general holding area, where the second bee was placed in front of the fan. This continued sequentially, so that all subjects had received a trial before the first bee received a second trial. The number of animals making up a squad varied from day to day, with the only requirement being that an equal number of animals from both groups be included in the squad. The experiment was over when all animals had received 24 CS presentations. To discourage the animal from forming an association between placement in front of the exhaust fan and feeding, we varied the onset of the CS between 1 and 10 s from the time of placement.

Results and Discussion

We plotted the results in terms of the proportion of animals responding to the CS+ and CS- for each of the 24 CS presentations (12 CS+ trials and 12 CS- trials; [ILLUSTRATION FOR FIGURE 1 OMITTED]). Comparison of the performance on CS+ and CS- trials clearly demonstrates the effect of discrimination training. When the perfume CS was followed by a feeding, the proportion of animals responding to the CS rose; when a different perfume CS was not followed by a feeding, the proportion of animals responding to that CS remained low. An analysis of variance (ANOVA) in which the performances on CS+ trials and on CS- trials were compared yielded a significant group effect (CS+ vs. CS-), F(1, 22) = 34.62, p = .0001; a significant trial effect, F(11, 242) = 2.79, p = .002; and a significant Group x Trial interaction, F(11, 242) = 4.42, p = .0001. There was no significant difference in performance between the bees who received Realm for Men as CS+ and those who received Realm for Women. ANOVA yielded no significant group effect, F(1, 22) = 2.49, p = .1289; a significant trial effect, F(11, 242) = 5.03, p = .0001; and no Group x Trial interaction, F(11, 242) = .70, p = .7354. There was also no significant difference between Realm for Men and Realm for Women when they served as CS-. ANOVA yielded no significant group effect, F(1, 22) = 0.60, p = .4483; no significant trial effect, F(11, 242) = 0.92, p = .5226; and no Group x Trial interaction, F(11, 242) = 0.50, p = .9036.

Experiment 2: Preferences for Soft Drinks

Our informal observations in the surrounding Stillwater, Oklahoma, area suggested that bees regularly visit trash sites and recycling centers. We were especially concerned to see the large numbers of bees visiting garbage sites located near schools. Upon closer inspection, it appeared that bees were attracted to those sites because of the residual sugar content of discarded soft drinks. A subsequent survey of 117 people revealed that only 44% of recyclers removed residual juice and other soft drinks before placing bottles in the receptacles (Aquino, Barker, Abramson, & Eikenbary, 1997). We decided to determine which soft drinks were attractive to AHBs. Groups of harnessed bees were given access to 10 different soft drinks. Two other groups received spring water or sucrose as a control.


Subjects. In mid- to late July 1996, 216 Africanized bees were captured as they left a UFPB hive. Workers were collected as they departed the hive at about 3 p.m. on the day before use. The subjects were placed in the conditioning harness as in Experiment 1, with training beginning at 7:30 the following morning.

Apparatus. We used the same apparatus as in Experiment 1, except that no CS-delivery devices (syringes) were used.

Procedure. The 216 subjects were divided into 12 groups containing 18 bees each. We used the following commercial soft drinks: Diet Coca-Cola, Diet Guarana, (a Brazilian soft drink), Diet Pepsi, Coca-Cola, Fanta Grape, Fanta Orange, Guarana, Pepsi, Sprite, and Sukita Orange (a Brazilian soft drink). Two other groups of bees were given sucrose or spring water as a control. The soft drinks, all obtained from local suppliers in Brazil, were selected based on their availability and popularity. When possible, we attempted to include the diet version of the soft drink. We also attempted to find soft drinks that were bottled by the same company. The makers of Pepsi, Guarana, and Coca-Cola provided diet and nondiet versions of their product; the Coca-Cola company provided the greatest range of soft drinks (Coca-Cola, Diet Coca-Cola, Fanta Orange, Fanta Grape, and Sprite).

The percentages of natural juice and sugar content were obtained from the labels. Sprite contained 2.5% lemon-lime juice; Fanta Orange and Fanta Grape each contained 10% natural juice. The labels of the remaining soft drinks did not reveal the amounts of natural juice used.

The spring water and the diet drinks contained no sugar. Diet Coke, however, contained .28 mM of saccharine, .32 mM of aspartame, and 2.5 mM of cyclamate; Diet Pepsi contained 1.9 mM of aspartame; Diet Guarana contained 0.89 mM of sodium saccharine and 3.4 mM of sodium cyclamate. The sugar content of the sucrose solution was 2.9 M. The sugar contents of Coca-Cola, Pepsi, Sprite, Fanta Orange, Fanta Grape, and Guarana were not printed on the labels, and attempts to obtain that information from local bottlers were unsuccessful. Coca-Cola contained an unspecified vegetal extract, and both Guarana and Diet Guarana contained Guarana vegetal extract. Vegetal extract was not present in the remaining beverages.

It should be noted that an obvious limitation in working with consumer products is the inability to obtain detailed information regarding the ingredients. The sugar content of soft drinks, for example, is seldom noted on the label, nor is the type of sugar specified (soft drinks often contain an unspecified blend of sugars such as corn sweetener and sucrose). Moreover, the problem is compounded when working with products manufactured outside the United States, because the rules and regulations governing the informational content of labels differ from country to country.

Approximately 2 weeks before the experiment, the soft drinks were purchased and the C[O.sub.2] was removed. Subjects were captured as in the previous experiment and tested in squads. Each squad contained bees from all groups. The soft drinks, sucrose, and spring water were poured into individual plastic cups. Strips of filter paper strips (1 x 3 cm) were dipped into the beverages. The beverage-impregnated filter paper was touched first to the antennae and then to the extended proboscis. We used a stopwatch to time the length of contact between the proboscis and the saturated filter paper. If an animal did not respond to the beverage or if it stopped responding to a beverage, we reattempted to stimulate the antennae and mouth parts. When the animal ceased feeding, it was permitted to drink its fill from a 2.9 M sucrose solution. The rationale behind the additional sucrose feeding was to ensure that animals who consumed little of the test solution were indeed capable of drinking. Following the sucrose feeding, there was no further participation of the animal in the experiment.

Results and Discussion

Diet sodas and spring water elicited the least contact, and those containing natural flavors elicited progressively more contact time. Sucrose elicited the most contact [ILLUSTRATION FOR FIGURE 2 OMITTED]. There was less than 10 s of contact with Diet Pepsi, Diet Guarana, Diet Coca-Cola, and spring water, 40-60 s of contact with Coca-Cola, Sprite, Pepsi, Guarana, and Sukita Orange, 75-95 s of contact with Fanta Orange and Fanta Grape, and 140 s of contact with sucrose. ANOVA yielded a significant effect for soft drinks, F(11, 204) = 18.74, p = .0001.

Experiment 3: Pavlovian Conditioning With Soft Drinks

In the previous experiment, we tested the ability of various soft drinks to serve as US. Many potential environmental cues – including colors, shapes, odors, and locations – are used by honey bees as signals for food sources. In addition to the unique odors associated with trash cans and recycling centers, the individual letters and colors identifying these sites consist of complex and contrasting shapes favored by bees (Ribbands, 1953). In the present experiment, we assessed the ability of a representative sample of soft drinks to support Pavlovian conditioning, by using the discrimination paradigm shown to be effective in Experiment 1. We were also interested in testing the ability of bees to form appetitive-conditioned reflexes with a US consisting of something other than sucrose.


Subjects. In late July 1996, 72 bees were captured from a UFPB hive. Workers were collected as they departed the hive at about 3 p.m. on the day before testing. The subjects were placed in the conditioning harness, as in Experiment 1; training began at 7:30 the following morning.

Apparatus. The apparatus was the same as in Experiment 1. The CS delivery devices were prepared by placing a drop of odorant (obtained from Sigma Chemical, St. Louis, MO) onto a piece of 1-[cm.sup.2] filter paper secured to the tip of the syringe’s plunger with an uncoated thumbtack. The drop was applied by dipping a wooden toothpick into the bottle and smearing the odorant onto the filter paper.

Procedure. The 72 bees were divided into six groups containing 12 bees each. Diet Pepsi, Pepsi, Guarana, Fanta Orange, Fanta Grape, and sucrose were used as US. The CS consisted of Citral (Sigma Chemical product number C-1645, 5.6 M concentration) and Geraniol (Sigma Chemical product number G-5135, 5.6 M concentration). Within each group, 6 of the 12 animals received a CS+ of Geraniol and a CS- of Citral. The remaining 6 animals received a CS+ of Citral and a CS- of Geraniol. A small amount of CS odorant was placed undiluted onto the filter paper by first dipping a wooden dowel into a vial containing one of the two odors used in the study. The dowel was then rubbed briefly onto the filter paper to transfer the odor from the dowel to the paper. We selected these odorants because they had been found to be effective in a previous AHB experiment (Abramson et al., 1997). Half of the animals in each group were exposed to three successive sequences of CS+ CS- CS- CS+ CS- CS+ CS+ CS-. The remaining animals were exposed to three successive sequences of CS- CS+ CS+ CS- CS+ CS- CS- CS+. The intertrial interval and the CS and US durations were the same as in Experiment 1 (5 min, 2 s, and 3 s, respectively). The sucrose concentration was 2.9 M. The experiment was concluded after an animal had received 24 CS presentations. The bees were placed on the conditioning stand and were run in squads, as in Experiment 1.

Results and Discussion

Except with Diet Pepsi, there was a gradual increase in the number of CS+ responses, with a plateau at about Trial 6 (see Figure 3). In contrast, there was no increase in responsiveness to the CS- (see Figure 4). With the exception of Diet Pepsi, the difference between CS+ and CS- performance was significant in all the soft drinks tested. Individual ANOVAs revealed a significant main effect for Fanta Grape (CS+ vs. CS-), F(1, 11) = 21.14, p = .0008; no significant trial effect, F(11, 121) = 1.34, p = .2087; and a significant Group x Trial interaction, F(11, 121) = 2.20, p = .0185. We found a similar pattern of results for Fanta Orange (CS+ vs. CS-), F(1, 11), = 12.28, p = .0049; with a significant trial effect, F(11, 121) = 3.52, p = .0003; and no significant Group x Trial interaction, F(11, 121) = 1.11, p = .3358. The other results were the following: Guarana (CS+ vs. CS-), F(1, 11) = 16.18, p = .002; no significant trial effect, F(11, 121) = 1.30, p = .2349; and a significant Group x Trial interaction, F(11,121) = 2.40, p = .0099; and Pepsi (CS+ vs. CS-), F(1, 11) = 11.13, p = .0086; a significant trial effect, F(11, 121) = 2.57, p = .0058; and no significant Group x Trial interaction, F(11, 121) = 0.91, p = .5366. Results indicated no conditioning with Diet Pepsi as the US (CS+ vs. CS-), F(1, 11) = 1.94, p = .1911; no trial effect, F(11, 121) = 1.14, p = .3348; and no Group x Trial interaction, F(11,121) = 0.86, p = .5803. As expected, conditioning with a sucrose US revealed a significant effect (CS+ vs. CS-), F(1, 11) = 13.28, p = .0039; a significant trial effect, F(11, 121) = 2.31, p = .0130; and a significant Group x Trial interaction, F(11, 121) = 2.16, p = .0212.

A comparison of the CS+ performance across groups revealed a significant main effect for groups (Fanta Grape, Fanta Orange, Guarana, Pepsi, Diet Pepsi, and sucrose), F(5, 66) = 4.48, p = .0014; a significant trial effect, F(11, 726) = 12.24, p = .0001; and no significant Group x Trial interaction, F(55,726) = 1.18, p =. 1847. Post hoc comparisons with Tukey’s method revealed that Diet Pepsi was different from all the rest. A comparison of the CS – performance across groups revealed no significant effects for groups (Fanta Grape, Fanta Orange, Guarana, Pepsi, Diet Pepsi, and sucrose), F(5, 66) = 1.42, p = .2273; no trial effect, F(11, 726) = 1.36, p = .1848; and no Group x Trial interaction, F(55,726) = 0.80, p = .8443.

General Discussion

The successful use of the proboscis-conditioning method opens the door to a host of behavioral and physiological experiments investigating the preferences for consumer products of Africanized and European honey bees. The Pavlovian discrimination task, in particular, has much to recommend it for the study of behavioral preferences in bees. Because the first presentation of a CS in any Pavlovian experiment must not elicit the conditioned response before training (i.e., the CS must be neutral), the presence of such a response indicates a behavioral preference for that CS. In addition, the discrimination procedure has the virtues of providing information regarding the modification of preferences when the US is never presented (CS-), serving as a sensitive within-subject control for nonassociative effects such as sensitization, and providing information about how similarly a bee perceives two CS scents by the degree of generalization. The more trials needed to show a separation between CS+ and CS- curves, the greater the perceived similarity between the two CSs. The proboscis technique is also easy to use, and many subjects can be tested rapidly, thereby making preference tests with large numbers of subjects possible.

Despite the advantages of the proboscis-conditioning method, a limitation should be noted. An analytical study of the individual ingredients (and a precise specification of the concentration of an individual ingredient) is difficult to obtain unless a researcher has access to the complete chemical formula of the product. It was our experience that manufacturers of consumer products are reluctant to provide such detailed information. We believe that this limitation will hinder the “molecular” study of the attractiveness of individual ingredients but will not influence the “molar” study of a consumer product in which the central question is whether the product attracts bees. It would be in the best interest of a manufacturer that markets a product found to be highly attractive to stinging insects to provide detailed information on the product or, alternatively, to form a partnership with a laboratory that studies the molar effects.

Experiment 1 demonstrated that perfumes can serve as discriminative stimuli in the same way as naturally occurring floral scents. The low proportion of animals responding on the first trial indicates that both Realm for Men and Realm for Women, although perhaps attractive to humans, are not strong attractors of Africanized bees. In addition, the clear and early separation of the CS+ and CS- curves suggests that these two perfumes are easily discriminable and perceived differently by the animals.

In a small pilot experiment patterned after Experiment 1, Bowe (1995) showed that European bees were unable to discriminate between Red perfume by Giorgio of Beverly Hills and the impostor version manufactured by Mystic Impressions. Bees also failed to discriminate Poison fragrance by Christian Dior from the impostor version manufactured by Mystic Impressions. However, an examination of the learning curves revealed a higher asymptotic level of performance in animals receiving either the designer or the impostor version of Poison. On the second CS presentation, for example, over 90% of the bees were extending their proboscises when exposed to either version of Poison. In contrast, only 50% of the bees exposed to the designer version of Red responded, and only 10% of the bees responded to the impostor version of Red. There were also differences between Poison and Red with regard to the number of animals extending their proboscises to the first presentation of the scents. Seven of 16 animals responded to the first presentation of either Poison, whereas only 2 of 16 bees responded to Red and its impostor version. This is not surprising. Poison is a sweet, fruity perfume, whereas Red contains spicy notes. Although the sample size in this pilot experiment was small, the results suggest that EHBs prefer sweet (Poison) over spicy florals (Red).

The results of Experiments 2 and 3 indicate that bees have preferences for soft drinks and that consumption of nondiet soft drinks can support classical conditioning. The latter finding is the first in which a soft drink US was used to demonstrate classical conditioning in honey bees. Aquino and Abramson (1995) reported a preliminary experiment based on the design of Experiment 2 that investigated soft drink preferences of EHBs. The results indicated that of the soft drinks tested, Welch’s strawberry drink (containing 10% natural juice) was the most attractive to bees, no less attractive to them than sucrose alone. Coca-Cola and Pepsi were the least attractive to bees of the nondiet soft drinks. Presumably, this was because Coca-Cola and Pepsi contain bitter compounds repellent to honey bees. Like Africanized bees, European bees also did not feed on Diet Coke or Diet Pepsi. European honey bees did, however, feed for a few seconds on Diet Dr. Pepper. It also was observed that the amount of sugar contained in the soft drinks was not proportional to the contact time. Sprite, for example, containing 93.6 mg/ml of sugar, elicited more contact than either Coca-Cola (96.0 mg/ml) or Pepsi (101.0 mg/ml). Moreover, the contact time with a 1.5-M sucrose solution was not significantly different from that recorded for Welch’s strawberry soda (126.0 mg/ml of sugar).

One of the more interesting findings in the present series of experiments was that except with diet soft drinks, the amount of contact time was not related to the efficacy of conditioning. For example, animals trained with a sucrose US performed as well as animals trained with a Fanta Orange US. Why this was so remains unclear. However, several possibilities can be suggested, the most straightforward of which is that following food deprivation, bees will associate a CS with any appetitive US. A second possibility is that the soft drinks contain natural flavors or odors that the AHBs have previously experienced. The AHB colonies were located relatively near citrus groves, and the vicinity of the hives was heavily populated with various types of flowers. A US containing familiar natural flavors and odors may be perceived by the AHB as a potent US despite the low sugar content relative to a 2.9-M sucrose solution.

A third possibility to account for the failure of AHBs to relate contact time with the efficacy of conditioning is that the effect might be revealed only when bees stop receiving the US. Extinction is one measure of persistence in conditioning experiments (Abramson, 1994), and previous work with EHBs has indicated that some conditioning effects are expressed only in extinction (Buckbee & Abramson, 1997). The role of extinction was not tested in Experiments 1 and 3, because we focused on what attracts AHBs to a consumer. After a consumer is stung, the bee dies and extinction does not have the opportunity to develop.

Our results lead to several recommendations. Health care professionals counseling clients suffering from Melissophobia or allergic reactions to bee stings should recommend that the clients’ perfume or cologne be tested for its attractiveness to bees. If the scent is found to attract bees, the client may decide to use another scent, such as a spicy floral or its designer impostor. Clients who are afraid to wear perfume or cologne and cosmetics because they may attract bees, wasps, hornets, and fire ants could be given a list of “safe” scents. University laboratories specializing in insect behavior could collaborate with allergy clinics to create empirically based lists of perfumes, cosmetics, and other consumer products that are Hymenoptera-safe.

We also recommend that bee traps be located near recycling centers and other public areas such as elementary schools, to monitor the amount of honey-bee activity. These traps could be maintained by local beekeepers’ associations or extension entomologists and could provide an excellent training opportunity for students. Establishing these monitoring stations would also provide a fine opportunity to educate the public about the importance of honey bees and other foraging insects.

Our results demonstrating the ability of bees to feed on soft drinks and to associate neutral stimuli with a soft-drink feeding strongly suggest that a greater effort should be made to remove appetitive-unconditioned stimuli from these containers, either by regularly washing out the containers or designing “bee proof’ containers. The negative publicity generated by a honey bee or “killer bee” attack at a recycling center, for instance, would certainly adversely influence the recycling effort.

1 Mention of a commercial product in this article does not constitute an endorsement of this product by the authors. No part of this article may be used in any form or by any means for advertisement purposes.


Abramson, C. I. (1994). A primer of invertebrate learning: The behavioral perspective. Washington, DC: American Psychological Association.

Abramson, C. I., Aquino, I. S., Silva, M. C., & Price, J. M. (in press). Learning in the Africanized honey bee: Apis mellifera L. Physiology & Behavior.

Aquino, I. S., & Abramson, C. I. (1995, March). The attractiveness of soft drinks to honey. bees (Apis mellifera L.). Paper presented at the annual meeting of the Oklahoma Psychological Association, Oklahoma City, OK.

Aquino, I. S, Barker, R. W., Abramson, C. I., & Eikenbary, R. D. (1997, February). The honey bee: A potential problem for the recycling centers in Oklahoma? Paper presented at the annual meeting of the Entomological Society of America (Southwestern branch), Oklahoma City, OK.

Bauer, K., Garbe, D., & Surburg, H. (1990). Common fragrance and flavor materials: Preparation, properties, and uses (2nd ed.). New York: VCH.

Bowe, K. (1995, April). Differential conditioning in honey. bees (Apis mellifera): Perfumes as conditioned stimuli. Paper presented at the annual meeting of the Southwestern Psychological Association, San Antonio, TX.

Buckbee, D. A., & Abramson, C. I. (in press). Identification of a new contingency-based response in honey bees (Apis mellifera) through revision of the proboscis extension conditioning paradigm. Journal of Insect Behavior.

Free, J. B. (1961). The stimuli releasing the stinging response of honey bees. Animal Behaviour, 9, 193-197.

Halonen, J. S., & Santrock, J. W. (1996). Psychology: Contexts of behavior. Dubuque, IA: Brown & Benchmark.

Litovitz, T. L., Felberg, L., White, S., & Klein-Schwartz, W. (1996). 1995 Annual report of the American Association of Poison Control Centers toxic exposure surveillance system. American Journal of Emergency Medicine, 14, 487-537.

Matchett, G., & Davey, G. C. (1991). A test of a disease-avoidance model of animal phobias. Behaviour Research and Therapy, 29, 91-94.

Regier, D. A., Boyd, J. H., Burke, J. D., Rae, D. S., Myers, J. K., Kramer, M., Robbins, L. N., George, L. K., Karno, M, & Locke, B. Z. (1988). One-month prevalence of mental disorders in the United States. Archives of General Psychiatry, 45, 977-986.

Reisman, R. E. (1992). Stinging insect allergy. Medical Clinics of North America, 76, 883-894.

Ribbands, C. R. (1953). The behaviour and social life of honeybees. London: Bee Research Association.

Smith, B. H., & Abramson, C. I. (1992). Insect learning: Case studies in comparative psychology. In L. I. Nadel & J. Byrne (Eds.), Encyclopedia of learning and memory, (pp. 276-283). New York: Macmillan.

Smith, B. H., Abramson, C. I., & Tobin, T. R. (1991). Conditional withholding of proboscis extension in honey bees (Apis mellifera) during discriminate punishment. Journal of Comparative Psychology, 105, 345-356.

Zimbardo, P. G. (1992). Psychology and life (13th ed.). New York: HarperCollins.

COPYRIGHT 1997 Heldref Publications

COPYRIGHT 2004 Gale Group