Can bees do maths? Yes – new research shows they can add and subtract

File 20181211 76962 cfh85r.jpg?ixlib=rb 1.1
Can we have a count of all the honeycomb cells please?

Scarlett Howard, RMIT University; Adrian Dyer, RMIT University, and Jair Garcia, RMIT University

The humble honeybee can use symbols to perform basic maths including addition and subtraction, shows new research published today in the journal Science Advances.

Bee have miniature brains – but they can learn basic arithmetic.

Despite having a brain containing less than one million neurons, the honeybee has recently shown it can manage complex problems – like understanding the concept of zero.

Honeybees are a high value model for exploring questions about neuroscience. In our latest study we decided to test if they could learn to perform simple arithmetical operations such as addition and subtraction.

Read more:
Which square is bigger? Honeybees see visual illusions like humans do

Addition and subtraction operations

As children, we learn that a plus symbol (+) means we have to add two or more quantities, while a minus symbol (-) means we have to subtract quantities from each other.

To solve these problems, we need both long-term and short-term memory. We use working (short-term) memory to manage the numerical values while performing the operation, and we store the rules for adding or subtracting in long-term memory.

Although the ability to perform arithmetic like adding and subtracting is not simple, it is vital in human societies. The Egyptians and Babylonians show evidence of using arithmetic around 2000BCE, which would have been useful – for example – to count live stock and calculate new numbers when cattle were sold off.

This scene depicts a cattle count (copied by the Egyptologist Lepsius). In the middle register we see 835 horned cattle on the left, right behind them are some 220 animals and on the right 2,235 goats. In the bottom register we see 760 donkeys on the left and 974 goats on the right.
Wikimedia commons, CC BY

But does the development of arithmetical thinking require a large primate brain, or do other animals face similar problems that enable them to process arithmetic operations? We explored this using the honeybee.

How to train a bee

Honeybees are central place foragers – which means that a forager bee will return to a place if the location provides a good source of food.

We provide bees with a high concentration of sugar water during experiments, so individual bees (all female) continue to return to the experiment to collect nutrition for the hive.

In our setup, when a bee chooses a correct number (see below) she receives a reward of sugar water. If she makes an incorrect choice, she will receive a bitter tasting quinine solution.

We use this method to teach individual bees to learn the task of addition or subtraction over four to seven hours. Each time the bee became full she returned to the hive, then came back to the experiment to continue learning.

Read more:
Are they watching you? The tiny brains of bees and wasps can recognise faces

Addition and subtraction in bees

Honeybees were individually trained to visit a Y-maze shaped apparatus.

The bee would fly into the entrance of the Y-maze and view an array of elements consisting of between one to five shapes. The shapes (for example: square shapes, but many shape options were employed in actual experiments) would be one of two colours. Blue meant the bee had to perform an addition operation (+ 1). If the shapes were yellow, the bee would have to perform a subtraction operation (- 1).

For the task of either plus or minus one, one side would contain an incorrect answer and the other side would contain the correct answer. The side of stimuli was changed randomly throughout the experiment, so that the bee would not learn to only visit one side of the Y-maze.

After viewing the initial number, each bee would fly through a hole into a decision chamber where it could either choose to fly to the left or right side of the Y-maze depending on operation to which she had been trained for.

The Y-maze apparatus used for training honeybees.
Scarlett Howard

At the beginning of the experiment, bees made random choices until they could work out how to solve the problem. Eventually, over 100 learning trials, bees learnt that blue meant +1 while yellow meant -1. Bees could then apply the rules to new numbers.

During testing with a novel number, bees were correct in addition and subtraction of one element 64-72% of the time. The bee’s performance on tests was significantly different than what we would expect if bees were choosing randomly, called chance level performance (50% correct/incorrect)

Thus, our “bee school” within the Y-maze allowed the bees to learn how to use arithmetic operators to add or subtract.

Why is this a complex question for bees?

Numerical operations such as addition and subtraction are complex questions because they require two levels of processing. The first level requires a bee to comprehend the value of numerical attributes. The second level requires the bee to mentally manipulate numerical attributes in working memory.

In addition to these two processes, bees also had to perform the arithmetic operations in working memory – the number “one” to be added or subtracted was not visually present. Rather, the idea of plus one or minus “one” was an abstract concept which bees had to resolve over the course of the training.

Showing that a bee can combine simple arithmetic and symbolic learning has identified numerous areas of research to expand into, such as whether other animals can add and subtract.

Read more:
Bees get stressed at work too (and it might be causing colony collapse)

Implications for AI and neurobiology

There is a lot of interest in AI, and how well computers can enable self learning of novel problems.

Our new findings show that learning symbolic arithmetic operators to enable addition and subtraction is possible with a miniature brain. This suggests there may be new ways to incorporate interactions of both long-term rules and working memory into designs to improve rapid AI learning of new problems.

Also, our findings show that the understanding of maths symbols as a language with operators is something that many brains can probably achieve, and helps explain how many human cultures independently developed numeracy skills.

This article has been published simultaneously in Spanish on The Conversation Espana.The Conversation

Scarlett Howard, PhD candidate, RMIT University; Adrian Dyer, Associate Professor, RMIT University, and Jair Garcia, Research fellow, RMIT University

This article is republished from The Conversation under a Creative Commons license. Read the original article.


There’s insufficient evidence your sunscreen harms coral reefs

Terry Hughes, James Cook University

In the face of persistent heatwaves, Australians are reaching for the sunscreen. But you might have heard some mixed messages about its harm to the environment – specifically to coral reefs.

In July 2018, Hawaii passed a law to prohibit the future sale of sunscreens containing benzophene-3 and octinoxate, claiming these two chemicals increase coral bleaching, and have significant harmful impacts on Hawaii’s marine environment.

Read more:
Marine heatwaves are getting hotter, lasting longer and doing more damage

In October 2018, the Republic of Palau followed suit, and banned “reef-toxic” sunscreens. Like most reefs throughout the tropics and subtropics, coral reefs in Hawaii and Palau have already severely bleached multiple times during recent, unusually hot summers, causing extensive loss of corals.

Key West, in Florida, may be the latest area to follow this trend, with a proposed ban to be voted on in early February.

However, medical and skin cancer specialists have warned of the public health risks of a ban on widely used sunscreens, describing the prohibition as risky and unjustified, in part because the few studies that have addressed the environmental impacts of sunscreens experimentally “are not representative of real world conditions”.

For example, the way in which coral tissues were exposed to sunscreen in experiments does not mimic the dispersal and dilution of pollutants from a tourist’s skin (and other sources) into reef waters and onto corals growing in the wild.

Experiments that expose corals to sunscreen chemicals typically use far higher concentrations than have ever been measured on an actual reef. A recent review of the amount of benzophne-3 in reef waters found that, typically, concentrations are barely detectable – usually, a few parts per trillion. One much higher report of 1.4 parts per million, in the US Virgin Islands, is based on a single water sample.

The environmental concerns over sunscreens on coral reefs are centred overwhelmingly on just two studies. The first, published in 2008, noted that there was no previous scientific evidence for an impact of sunscreens on coral reefs.

This study exposed small fragments of corals (branch tips) to high levels of benzophenone-3 and other chemicals by incubating them for a few days inside plastic bags. The fragments in the bags quickly became diseased with viruses and bleached. The authors concluded “up to 10% of the world reefs are potentially threatened by sunscreen-induced coral bleaching”.

Bleaching is a stress response by corals, where they turn pale due to a decline in the symbiotic micro-algae that lives inside their tissues. You can make a coral bleach experimentally by torturing it in any number of ways. However, coral bleaching at a global and regional scale is caused by anthropogenic heating, not sunscreen. We know the footprint of bleaching on the Great Barrier Reef in 1998, 2002, 2016 and 2017 is closely matched to where the water was hottest for longest in each event.

Even the most remote reefs are vulnerable to heat stress. The physiological mechanisms and timescale of thermal bleaching due to global heating is very different from the rapid responses of corals to experimental exposure to high concentrations of sunscreen chemicals.

The second and most-widely cited study of sunscreen toxicity on corals is also laboratory-based. Published in 2016, it focused mainly on the responses of the day-old larvae of one coral species, as well as isolated coral cells. This study did not examine intact coral colonies.

The larvae were placed in 2-3 centilitres of artificial seawater containing a range of concentrations of sunscreen chemicals and a solvent to disperse them. After a few hours, the coral larvae became increasingly pale (bleached) with higher concentrations of oxybenzone.

Read more:
Why there’s still hope for our endangered coral reefs

This study also measured the concentration of benzophenone in sea water at six locations in Hawaii. These samples were unreplicated (one per location), and all of them had unmeasureable amounts of sunscreen chemicals. In the US Virgin Islands, the authors found higher concentrations of benzophenone at four out of ten locations, although they did not report results for any blank samples (to control for contamination). The study concluded that oxybenzone threatens the resilience of coral reefs to climate change.

In conclusion, there is actually no direct evidence to demonstrate that bleaching due to global heating is exacerbated by sunscreen pollutants. Similarly, there is no evidence that recovery from thermal bleaching is impaired by sunscreens, or that sunscreens cause coral bleaching in the wild.The Conversation

Terry Hughes, Distinguished Professor, James Cook University

This article is republished from The Conversation under a Creative Commons license. Read the original article.