Fish tend not to be renowned for their intellect. Their memories weren't thought to be great and until recently no one thought that they could assess quantity. But now we are having to reconsider these opinions. ‘Fish are capable of processing both small and large numbers with a performance similar to that described in mammals and birds’, says Christian Agrillo from the University of Padova, Italy. However, all of the fish species that have been tested to date were able to see. Could blind fish process numbers using other senses? Wondering whether vision is essential for fish to evaluate quantity, Agrillo and his colleagues from Padova and the University of Ferrara, Italy, set about testing the numeracy of blind cavefish (p. 1902).
According to Agrillo, Somalian cavefish (Phreatichthys andruzzii) are the ideal species to test as they have been deprived of light for two million years, leading to loss of eye function. However, the fish have compensated for the deficiency by increasing the sensitivity of their lateral line sensors, which allows them to discriminate between different 3D shapes.
Having trained the fish to associate groups of six objects with food until they automatically gravitated towards the groups – even when food was no longer present – the team then tested whether the fish could distinguish between clusters of six and two objects; which they did. However, the team needed convincing that the fish weren't relying on other cues (such as differences between the surface areas of the two clusters or the different volumes occupied) to differentiate between groups with different numbers of objects.
This time the team tested the fish's ability to discriminate between groups of two and four objects, but sometimes they made sure that the two groups had the same surface area, occupied the same volume or were distributed with the same density, while on other occasions they allowed the surface area, volume and density to differ between the two groups. As soon as the team removed the differences between the additional cues, the fish were no longer able to differentiate.
Puzzled, the team wondered whether changing the test conditions (groups of two objects versus groups of four) from the training conditions (groups of two objects versus six) had flummoxed the fish, or whether the fish may simply not have the processing power to differentiate between a small difference of two. The team trained the fish using clusters of two and four objects – where the two clusters had the same surface area, volume and density characteristics – and this time the fish successfully discriminated between the two clusters. ‘This represents the first evidence of non-visual numerical abilities in fish’, the team says. However, there is a limit to the blind fish's numerical abilities; they were unable to distinguish between groups of two and three objects, suggesting that vision may improve fish's numerical accuracy.