Numerical discrimination in Drosophila melanogaster

Mercedes Bengochea, Jacobo D. Sitt, Thomas Preat, Veronique Izard, Laurent Cohen, Bassem A. Hassan

Posted on: 25 March 2022

Preprint posted on 1 March 2022

Article now published in Cell Reports at

“One, two, many, lots”: fruit flies discriminate between and form preferences about different numbers

Selected by Sophia Friesen

Why I liked this paper:

In my day-to-day work as a graduate developmental biologist, I spend a lot of time with fruit flies. They’re my model organism of choice, and I’ve learned to appreciate them for their genetic tractability, convenience, and surprisingly intricate body plan. But I’ve never thought of them as particularly smart, which is why I was surprised – and delighted – to read the latest work by Bengochea et al demonstrating that flies can discriminate between sets containing different numbers of objects. Using a conceptually simple but powerful behavioral assay, they found that flies form trainable preferences about sets of different values, and – by taking advantage of the unusually comprehensive genetic resources available in flies – determined that number discrimination appears to depend on a specific group of neurons.


Number-based thinking can be found across the animal kingdom, from minnows that regulate their shoal size to ants that orient themselves by counting steps. Neural activity monitoring can correlate brain regions with number-based behaviors, but it’s difficult to understand the mechanisms behind numerical thinking without being able to manipulate the system and see what happens. The genetic tools available to manipulate fruit flies provided the researchers with a unique opportunity to learn more about number-based thinking by directly and precisely manipulating parts of the brain.

Major findings:

Flies show an innate preference for larger numbers, which can be reversed through learning

The researchers detected numerical discrimination with a surprisingly simple behavioral assay: flies were placed in an arena where they could see two sets of visual stimuli, and researchers measured whether they spent more time near one set or the other. Flies consistently spent more time near sets of three shapes than a single shape, independent of the size of the shape or the overall size of the set. This preference for larger numbers held true for comparisons between two vs. four shapes, as well as two vs. three shapes, but distinguishing between three and four appears to be beyond the flies’ abilities.

The researchers also found that flies are able to learn new associations that require number judgments. The authors could train flies to preferentially spend time near less-numerous sets of objects by briefly exposing them to an arena with sugary food near the smaller set. Even in the absence of food, the flies spent more time near the smaller set, showing that they had formed a new association based on set size.

Flies use ratios, rather than absolute values, to compare numbers

In general, animals understand numbers through two cognitive systems: the Object Tracking System, which makes simultaneous mental representations of each object to precisely measure small numbers, and the Approximate Number System, which relies on the ratio between two sets to make less-precise judgments of indefinitely large numbers. To identify which cognitive system flies use to compare numbers, the researchers ran the behavioral test comparing sets of large (ish) numbers with large ratios. Flies don’t show a preference for four vs. three objects, but they do show a preference for eight vs. four objects. This means that their number cognition doesn’t “max out” at four – rather, they can use ratios between larger numbers to determine which is bigger.

A specific group of neurons, LC11, is required for numerical judgments

To find out which parts of the brain are involved in numerical preference, the researchers genetically expressed Tetanus Light Chain in different brain regions, which “silences” neurons by preventing signaling across synapses. They could then test if flies still showed number-based behavioral preferences when a particular brain region was silenced. After testing several different brain regions, the researchers identified lobula columnar neurons 11 (LC11) as required for both innate preference and number-based learning. These neurons are involved in fly social behaviors that are modulated by group size, hinting at a way that numerical cognition could be advantageous for flies.


Questions for the authors:

  1. Could you provide more information on how you silenced specific brain regions or sets of neurons? I assume you used a Gal4 or LexA genetic driver line, but I couldn’t find it in your methods.
  2. I’m a little surprised that flies are able to function somewhat normally when large regions like the mushroom body are silenced – do you see other changes in behavior or viability in these conditions?
  3. As far as I can understand, your results demonstrate that flies can use the Approximate Number System to make judgments about large numbers, but you don’t rule out the possibility that flies use the Object Tracking System for smaller numbers. How can you test for the presence of an Object Tracking System, given that the small numbers it would presumably be used for all have large ratio differences?



Read preprint (No Ratings Yet)

Have your say

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here