Counting insects

Abstract

When counting-like abilities were first described in the honeybee in the mid-1990s, many scholars were sceptical, but such capacities have since been confirmed in a number of paradigms and also in other insect species. Counter to the intuitive notion that counting is a cognitively advanced ability, neural network analyses indicate that it can be mediated by very small neural circuits, and we should therefore perhaps not be surprised that insects and other small-brained animals such as some small fish exhibit such abilities. One outstanding question is how bees actually acquire numerical information. For perception of small numerosities, working-memory capacity may limit the number of items that can be enumerated, but within these limits, numerosity can be evaluated accurately and (at least in primates) in parallel. However, presentation of visual stimuli in parallel does not automatically ensure parallel processing. Recent work on the question of whether bees can see 'at a glance' indicates that bees must acquire spatial detail by sequential scanning rather than parallel processing. We explore how this might be tested for a numerosity task in bees and other animals.This article is part of a discussion meeting issue 'The origins of numerical abilities'.

Keywords: bee; brain size; counting mechanisms; neuronal number; numerical cognition; working memory.

Figure 1.

Landmark counting by honeybees in an open field. Bees were originally trained to fly from a hive (out of view to the left) to a feeder located at a distance of 262.5 m, between the third and fourth of a series of yellow tetrahedral tents, spaced 75 m apart. In subsequent tests, spacing between the tents was systematically varied and two feeders were offered; one at or close to the distance from the hive learned during training, and a second spaced between the second and third tents, and consequently, at an altered flight distance from the hive [15]. The question was, would the bees be more likely to find the feeder at the trained distance, or would they find it by the number of landmarks passed during training flights? See text for details.

Figure 2.

Landmark counting in a laboratory flight tunnel. (a) Individual bees were trained to receive a reward after they had flown past a specified number of landmarks. During training, the landmarks were strips of evenly spaced yellow paper (upper). Spacing interval was randomly varied every 5 min, to ensure the bees could not learn the reward location by measuring flight distance. Different experimental groups were tested on the same landmarks as in training; in tunnels where the stripes were replaced by yellow disks, presenting a smaller cumulative yellow surface; or in tunnels where landmarks were arranged as baffles, so that only one could be seen at a time. (b) Results of an experiment where bees were trained on landmark 3, then tested with landmarks spaced regularly every 40 cm (upper panel) or irregularly spaced (lower panel). Modified from Dacke & Srinivasan [21], with permission.

Figure 3.

Summary of bees' choice behaviour in the experiments of Gross et al. [23]. Following training on either two or three stimuli, bees were tested in discrimination or transfer tests with a sample of numerosity either two or (illustrated here) three (in the actual experiments the correct arm of the Y-maze was randomized). (a) Exact pattern match. (b) Pattern matching by numerosity only. Size, configuration, colour were varied in extensive series of transfer tests to rule out non-numerical cues. (c) Bees were able to match to sample when distractor contained novel numerosity (four), but performance was not significantly above chance when the sample contained the novel numerosity (d). Bees were also unable to discriminate between stimuli containing four and six items (not shown).

Figure 4.

Bees cannot perceive complex visual stimuli ‘at a glance’. Bees were trained in a flight arena with six feeding platforms (blue horizontal lines in the panels on the bottom) positioned in front of a 120 Hz (8.33 ms refresh rate) gaming monitor. Separate groups of bees were trained on five tasks, ranging from simple detection to complex pattern discrimination, in four temporal presentation conditions, ranging from continuous presentation (static) to timed presentation (repetition rate randomized between 500 and 1000 ms) for 100, 50 or 25 ms. (a) Detection of oriented bar; (b) discrimination of 45° from −45° bars; (c) coarse colour discrimination yellow-blue; (d) fine colour discrimination yellow-orange; (e) discrimination of spider shape from circle (only two of six stimuli shown for simplicity). All of the bees were successful in acquiring the simple detection task, regardless of presentation duration. For fine colour discrimination, stimulus durations of at least 50–100 ms were required (d), while only a single bee learned the shape discrimination at 100 ms, even though all bees learned the task under continuous presentation (e). Modified from Nityananda et al. [47], with permission.

Figure 5.

Flight path of a bee trained, with differential conditioning, to select stimuli with two items and avoid those with four. The first 10s of the bee's scanning behaviour are shown; the path is colour-coded to show the progression from early (violet) to late (red). The bee sequentially examines two patterns containing four items, but rejects each of them after scanning three items in each. She then chooses a pattern containing the correct number of two purple crosses (even though she has not been rewarded on any other dots than yellow ones before) and finally selects another pattern with the correct number of two (yellow) dots. Dots are separated by time intervals of 33 ms. See also electronic supplementary material, video S1.