Sense Of Number in Invertebrate Cognition

Project Information

SONIC

Grant agreement ID: 101111026

DOI

10.3030/101111026

EC signature date 5 May 2023

Start date 1 January 2024

End date 31 December 2026

Funded under

• Marie Skłodowska-Curie Actions (MSCA)

Total cost

No data

EU contribution

€ 298 690,72

Coordinated by

UNIVERSITE DE CAEN NORMANDIE
France

Periodic Reporting for period 1 - SONIC (Sense Of Number in Invertebrate Cognition)

Reporting period: 2024-01-01 to 2025-12-31

Summary of the context and overall objectives of the project

Understanding how numerical cognition emerges across species is a central question in cognitive science and evolutionary biology. True counting abilities — such as recognising numbers across different procedures and sensory modalities or spontaneously ordering them along a mental continuum — are often considered uniquely human. Even highly trained non-human primates show limitations when transferring numerical representations across contexts or modalities. However, accumulating evidence across different species suggests that foundational cognitive building blocks of numerical abilities are present in a wide range of animals, from insects to vertebrates. This raises a fundamental question: does numerical cognition reflect a shared ancestral mechanism, or does it represent an evolutionary solution shared by distant species, to adapt to environmental pressures and is the result of how neurons function?
The SONIC project investigates the limits of numerical cognition in invertebrates and focuses on bees and octopuses. These taxa differ profoundly in brain architecture, ecology and evolutionary history. Studying them comparatively allows us to test whether advanced numerical abilities reflect shared ancestry or convergent evolution under common computational constraints.
The project is structured around three scientific objectives:
1. Exploring the limits of symbolic representation of numbers in invertebrates.
2. Assessing the ability to translate numerical information across sensory modalities.
3. Developing computational models of conceptual numerical abilities.
At this stage, the outgoing phase, conducted at Monash University in Australia in collaboration with a French research centre to optimise seasonal data collection, focuses on bees (European honeybees and bumblebees). By combining behavioural experiments, cross-modal paradigms and computational modelling, the project aims to clarify whether abstract numerical processing can arise in miniature brains.
Overall, demonstrating abstract and crossmodal numerical processing in phylogenetically distant invertebrates would significantly reshape our understanding of the evolution of number sense. Beyond advancing fundamental knowledge, the project also contributes to broader discussions on animal cognition, welfare considerations linked to cognitive capacities, and the diversity of evolutionary solutions to complex information processing.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

During the outgoing phase of the SONIC project, experimental work focused on investigating numerical cognition in bees. Research was conducted at Monash University (Australia), and Paul Sabatier University (France). A total of 15 experiments were completed in bees, with more than 180 individual tested across two species (Apis mellifera and Bombus terrestris).
The work addressed three core scientific objectives. First, we aim at shedding light on the symbolic representation of numbers in bees’ miniature brains. In a series of experiments, the project examined whether bees can associate abstract symbols with specific numerosities and operate on these representations beyond immediate perception of a quantity. A key achievement concerns the investigation of associative symmetry (bidirectional mapping and equivalence relationship between symbols and numbers): after training bees to associate a given numerosity with a symbol, we tested whether they could reverse this mapping spontaneously. These experiments contribute to clarifying the limits of symbolic representation and relational learning in miniature brains. Second, we investigated bees’ ability to transfer quantitative information from one sense to another, i.e. crossmodal processing of numbers. A major methodological advance was the development and validation of vibrational stimuli perceivable by bees. We established experimental protocols allowing individuals to encode information in one sensory modality (tactile) and match it in another (visual). Finally, we are currently in the early stages of computationally modelling bees’ performances in learning tasks. Our initial efforts focus on modelling colour discrimination learning using reinforcement-based decision architectures. This approach aims to first capture basic associative learning dynamics before extending the framework to numerical discrimination and symbolic tasks. While preliminary, this modelling strategy lays the groundwork for a unified computational account of learning across tasks of increasing complexity.
Although no peer-reviewed articles have yet been published from the outgoing phase, several manuscripts are in preparation. Findings have been presented at international scientific conferences, enabling feedback from the research community and refinement of experimental designs. The project has also contributed to outreach activities aimed at increasing public awareness of insect cognition.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

Research on animal numerical cognition has traditionally focused on vertebrates, especially primates, rodents and birds. While bees have impressive numerical abilities on non-symbolic numbers (i.e. perceptible quantity of items), the limits of their abstract numerical cognition remain largely unexplored. Systematic comparative approaches across distantly related invertebrates are also extremely rare.
The SONIC project advances beyond the current state of the art in several ways. First, it moves beyond simple quantity discrimination by investigating symbolic number representation and associative symmetry in bees. Testing whether bees can reverse learned mappings between symbols and quantities addresses higher-order relational processing rather than perceptual discrimination alone. This contributes new empirical data to ongoing debates about the depth and limits of symbolic cognition in insects.
Second, the project establishes and validates crossmodal experimental paradigms in two bee species. Demonstrating that information acquired through tactile exploration can be matched to visual stimuli provides a methodological breakthrough for assessing abstract representations in invertebrates. Crossmodal transfer is a key criterion for identifying non-perceptual, more conceptual forms of processing.
Third, the project adopts a structured comparative perspective that will extend to octopuses during the return phase. Indeed, the project aims to clarify whether advanced numerical capacities reflect convergent evolution or specific adaptations in different taxa. This evolutionary framing goes beyond isolated species studies and contributes to a broader theoretical understanding of number sense.
The potential impact of these results is primarily scientific. If abstract and crossmodal numerical processing can be demonstrated in miniature and radically different neural architectures, this would significantly reshape current models of the cognitive and neural requirements for numerical abstraction. Conversely, identifying specific limits also provides critical constraints for theories of numerical cognition.
To ensure further uptake and long-term impact, continued efforts will focus on completing comparative datasets (including octopuses’ learning performances), finalising computational modelling frameworks and publishing open methodologies and conceptual breakthroughs to advance the debate on invertebrates’ cognitive abilities and facilitate replication and extension of this work by other research groups.