Aversive learning in honeybees revealed by the olfactory conditioning of the sting extension reflex

Abstract

Invertebrates have contributed greatly to our understanding of associative learning because they allow learning protocols to be combined with experimental access to the nervous system. The honeybee Apis mellifera constitutes a standard model for the study of appetitive learning and memory since it was shown, almost a century ago, that bees learn to associate different sensory cues with a reward of sugar solution. However, up to now, no study has explored aversive learning in bees in such a way that simultaneous access to its neural bases is granted. Using odorants paired with electric shocks, we conditioned the sting extension reflex, which is exhibited by harnessed bees when subjected to a noxious stimulation. We show that this response can be conditioned so that bees learn to extend their sting in response to the odorant previously punished. Bees also learn to extend the proboscis to one odorant paired with sugar solution and the sting to a different odorant paired with electric shock, thus showing that they can master both appetitive and aversive associations simultaneously. Responding to the appropriate odorant with the appropriate response is possible because two different biogenic amines, octopamine and dopamine subserve appetitive and aversive reinforcement, respectively. While octopamine has been previously shown to substitute for appetitive reinforcement, we demonstrate that blocking of dopaminergic, but not octopaminergic, receptors suppresses aversive learning. Therefore, aversive learning in honeybees can now be accessed both at the behavioral and neural levels, thus opening new research avenues for understanding basic mechanisms of learning and memory.

Figure 1

View of a honeybee in the experimental set-up. The bee is fixed between two brass plates (E1, E2) set on a plexiglas plate (pp), with EEG cream smeared on the two notches (N1, N2) to ensure good contact between the plates and the bee, and a girdle (G) that clamped the thorax to restrain mobility. The bee closes a circuit and receives a mild electric shock (7.5V) which induces the sting extension reflex (SER). An originally neutral odorant is delivered through a 20 ml syringe (S) placed 1cm from the antennae. Odorant stimulation lasted 5 sec. The electric shock started 3 sec after odorant onset and lasted 2 sec so that it ended with odorant offset. Contamination with remains of odorants used for conditioning or pheromones is avoided via an air extractor (AE) which is on continuously.

Figure 2

Associative olfactory conditioning of the sting extension reflex (SER) in honeybees. a) Responses (SER) of bees trained with an odorant explicitly paired with an electric shock (black squares; n = 38) and with odorant and unpaired electric shock (white squares; n = 39) during 6 trials. Only the bees in the paired group learned the association and extended their sting as a response to the odorant. One hour after conditioning an olfactory aversive memory was present in the paired (black bar), but not in the unpaired, group (white bar). b) Responses (SER) of bees (n = 48) trained to discriminate an odorant reinforced with an electric shock (black squares) and a non-reinforced odorant (white squares) during 12 trials (6 reinforced and 6 non-reinforced). Bees learned to discriminate between odorants as a result of conditioning. *: p<0.0001.

Figure 3

Simultaneous aversive and appetitive learning in honeybees. The same group of bees (SER-PER group; n = 80) was trained in a double discrimination task with an odorant (‘A’) paired with an electric shock that elicited the sting extension reflex (SER) and with another odorant (‘B’) paired with sucrose solution delivered to the antennae and proboscis that elicited the proboscis extension reflex (PER). a) Bees responded significantly with a SER to the odorant associated with the electric shock (black dots and bar), but not to that associated with sucrose (white dots and bar). b) The same bees responded significantly with a PER to the odorant associated with sucrose (white dots and bar), but not to that associated with electric shock (black dots and bar). One hour after the last conditioning trial, bees still responded correctly to the odorants (bars), even if their respective USs were absent in the tests. As a result of training, bees exhibited the appropriate response to the appropriate odorant. Appetitive and aversive learning can thus be mastered simultaneously. c,d: Control groups trained to discriminate odorants A and B, one of which was non-reinforced and the other reinforced either with electric shock (c: SER group; n = 80) or with sucrose solution (d: PER group; n = 80). c) Bees responded significantly with a SER to the odorant associated with the electric shock (black squares and bar), but not to the non-reinforced odorant (grey squares and bar). The performance of this group did not differ from that of the SER-PER group [compare with a)]. d) Bees responded significantly with a PER to the odorant associated with the sucrose solution (white squares and bar), but not to the non-reinforced odorant (grey squares and bar). The performance of this group was significantly better than that of the SER-PER group [compare with b)]. *: p<0.0001.

Figure 4

The effect of electric shock on appetitive olfactory conditioning of the proboscis extension reflex (PER). Responses (PER) of bees trained to associate an odorant with sucrose solution along six conditioning trials and experiencing six electric shocks (‘shock group’; black dots and bars; n = 40) or six placements in the conditioning setup (‘placement group’; white dots and bars; n = 40) interspersed pseudorandomly. Bees in both groups learned to respond to the rewarded odorant irrespective of the presence or absence of shock. One hour later, both groups behaved similarly in the retention test (bars) and responded significantly more to the conditioned odorant than to the novel odorant. Thus, repetitive stimulation with the electric shock did neither affect appetitive olfactory learning nor olfactory memory one hour after conditioning. *: p<0.0001.

Figure 5

The effect of octopaminergic and dopaminergic receptor antagonists on olfactory conditioning of the sting extension reflex (SER). Responses (SER) of bees trained to discriminate an odorant reinforced with an electric shock (black squares) and a non-reinforced odorant (white squares) during 12 acquisition trials (6 reinforced and 6 non-reinforced). A retention test was conducted 1 h after the last acquisition trial (black bar: odorant previously reinforced; white bar: odorant previously non-reinforced). a) Responses (SER) of control bees injected with Ringer into the brain (n = 40); b) Responses (SER) of bees injected with the octopaminergic antagonist mianserine 3.3 mM into the brain (n = 40); c) Responses (SER) of bees injected with the dopaminergic antagonist flupentixol 2 mM into the brain (n = 40). Ringer-and mianserine-injected bees learned to discriminate the reinforced from the non-reinforced odorant and remembered the difference one hour later. Flupentixol-injected bees did not learn to discriminate the reinforced from the non-reinforced odorant, nor did they respond appropriately in the retention tests. Similar results were obtained with other concentrations of octopaminergic and dopaminergic antagonists (see Fig. 6). These results show that dopamine, but not octopamine receptors are required for aversive olfactory learning in honeybees.

Figure 6

The effect of different octopaminergic and dopaminergic antagonists on acquisition (differential conditioning with two odorants, one reinforced and the other non-reinforced) and retention (memory test 1h after conditioning) of olfactory aversive learning in honeybees. Values correspond to the mean±S.E of a discrimination index (responses to reinforced odorant–responses to non-reinforced odorant) in the last acquisition trial and in the retention test. They are expressed in%. Sample sizes are indicated in parentheses. Colors correspond to the groups that were performed in parallel. Groups sharing the same color share the same Ringer control. Pairwise comparisons between drug-and Ringer-injected (control) groups were performed using a Mann-Whitney test. Z-adjusted values and significance level (p) are given for each comparison. NS: non-significant; *: p<0.05 in blue row; *: p<0.0125 in yellow rows (α/4); *: p<0.008 in pink rows (α/6). (^†): Significance in this case is due to an unusually low performance in the control (Ringer-injected) group, and not to an incremental effect of epinastine.

Figure 7

The effect of vertebrate D1-like and D2-like dopamine receptor antagonists on olfactory conditioning of the sting extension reflex (SER). Responses (SER) of bees trained to discriminate an odorant reinforced with an electric shock (black squares) and a non-reinforced odorant (white squares) during 12 acquisition trials (6 reinforced and 6 non-reinforced). A retention test was conducted 1 h after the last acquisition trial (black bar: odorant previously reinforced; white bar: odorant previously non-reinforced). a) Responses (SER) of control bees injected with Ringer into the brain (n = 40); b) Responses (SER) of bees injected with the D1-like dopamine receptor antagonist SCH23390 3 mM into the brain (n = 40). c) Responses (SER) of bees injected with the D2-like dopamine receptor antagonist spiperone 2.5 mM into the brain (n = 40). Ringer-and SCH23390-injected bees learned to discriminate the reinforced from the non-reinforced odorant and remembered the difference one hour later. Spiperone-injected bees did not learn to discriminate the reinforced from the non-reinforced odorant, nor did they respond appropriately in the retention tests. These results suggest that AmDOP receptors could contribute differently to aversive learning in bees.