| Associative learning and memory | Apis mellifera | Harnessed bees given an odour paired with a sugar reward on the antennae, then tested with the odour alone (proboscis extension conditioning) | Bees extend the proboscis to the trained odour after one or a few pairings, and memories pass through short-, mid- and long-term phases. | Classical conditioning, with memory phases resembling those of other animals. | A restrained-bee assay; it measures one response under laboratory conditions, which free-flying foraging only partly resembles. |
|---|
| Sameness and difference | Apis mellifera | Free-flying bees in a Y-maze rewarded for choosing the arm that matched (or differed from) a sample, then tested with new stimuli — including from another sense | Bees transferred the matching or non-matching rule to stimuli they had never seen. | Learning of a relational rule, not only of particular stimuli. | Shown for this task and these stimuli. It does not show that bees hold a general concept of sameness outside the apparatus. |
|---|
| Two rules at once | Apis mellifera | Bees trained to choose by a spatial relation (above or below) and by sameness together, then tested with new images | Bees applied both rules to new stimuli in the same choice. | Combination of two learned relational rules. | One experimental design; the stimuli and training define what was learned. |
|---|
| Ordering an empty set | Apis mellifera | Bees trained to choose the stimulus with fewer (or more) elements, then offered a blank card alongside cards with elements | Bees trained on 'fewer' tended to place the blank card at the low end of the order. | Consistent with treating an empty set as a low quantity; contested as a response to low-level cues such as total area or edge length. | Whether this reflects a numerical representation at all is unresolved. It is a result about one task, not a claim that bees grasp zero as a number. |
|---|
| Learning from a demonstrator | Bombus terrestris | Bumblebees shown a ball being moved to a target for reward, by a live or model demonstrator, then given the task | Bees learned the task by observation and often improved on the demonstrated solution rather than copying it. | Social learning with flexibility beyond imitation. | Bumblebees, not honey bees, and a task with no counterpart in their natural foraging. |
|---|
| Learning a signal from others | Honey bees (Apis); the abstract read does not name the species | Colonies in which young bees could or could not follow dances before they first danced themselves | Bees that could not follow dances danced with larger angle errors and encoded distance wrongly; the angle errors improved with experience, the distance error did not. | Correct waggle dancing depends partly on social learning. | Shows that experience shapes the signal; it does not settle how much information recruits extract from a dance in natural foraging. |
|---|
| A state that biases choices | Bombus terrestris | Bumblebees given an unexpected sugar reward, then tested on ambiguous cues between trained rewarded and unrewarded ones | Rewarded bees approached ambiguous cues faster; a dopamine antagonist abolished the effect. | An 'emotion-like' state in the authors' phrase — a reward-induced bias in decisions. | The qualifier matters: the result shows a behavioural and pharmacological state, not felt experience. |
|---|