Fish can count as well as college students, according to a study published in a recent issue of the journal PLoS ONE.
Building on an earlier study that demonstrated that the Eastern mosquitofish (Gambusia affinis) had the ability to count up to four and possessed a rudimentary mathematical ability to visually count items if the number is small, Christian Agrillo, Laura Piffer and Angelo Bisazza of the University of Padova used a series of three experiments to investigate how mosquitofish perceive large numbers.
The first experiment examined the raw numerical ability of mosquitofish by testing if they were able to discriminate between four and eight objects with or without the continuous variables that co-vary with number (area, space, density, total brightness).
The authors first trained ten female mosquitofish to discriminate between four and eight objects.
This was done by isolating the test females from their shoals and allowing them to rejoin the shoals through doors that had been labeled with either four or eight random geometric shapes (female mosquitofish show a strong preference to shoal, so rejoining their shoals is a reward for them).
After the fish had been trained (half were trained to go through the door with four symbols and the other half the door with eight), the authors then removed the non-numerical variables that the fish may have been using to discriminate between the stimuli by matching the cumulative surface area, density, overall space occupied by the arrays of figures and total brightness between the two groups of numbers.
Again, the fish were able to discriminate between the two sets of numbered objects.
The authors next trained six fish to distinguish numbers in a 1:2 ratio (four versus eight, five versus ten and six versus 12). Finding no significant difference in the discriminatory abilities of the fishes regardless of quantity (i.e. the fish trained to distinguish between four and eight objects performed as well as those trained to distinguish six and 12), the authors then presented the fish with a choice consisting primarily of larger quantities (four versus eight, 15 versus 30, and 100 versus 200).
The fishes again were able to distinguish between the larger numbers presented when the stimuli were made of identical geometric figures.
In the third set of experiments, the authors used fishes trained to discriminate between two sets of numbers (five versus ten and six versus 12) and tested if they could discriminate numerical contrasts in the ratios of 1:2, 2:3 and 3:4 (seven vs. 14, eight vs. 12 and nine vs. 12 respectively).
The authors found that the fishes could discriminate numbers presented in the ratios of 1:2 and 2:3, but not in the ratio of 3:4.
Lastly (and most interestingly), the authors applied the same tests used in the second and third experiments on human subjects (25 undergraduates), allowing the undergraduates to view the opposing sets of stimuli for 150 milliseconds (this prevented the human subjects from actually counting the geometric shapes) and asking them to choose the larger number.
In this set of experiments, half the tests were corrected for the continuous variables that co-vary with number and the other half were not.
The subjects were also required to intone 'abc' continuously to prevent verbal processing of the stimuli. The authors found that the performance of the undergraduates was equivalent to that of the mosquitofish, implying that the mosquitofish's ability to discriminate between large numbers of objects is similar to that observed in humans in almost all respects.
According to the authors, two lines of reasoning can explain this similarity.
Firstly, the cognitive abilities and behavioural complexity of fishes have previously been greatly underestimated and it is now well documented that fishes possess many of the cognitive functions that were once believed to be associated with evolution of a large brain in mammals and birds.
Secondly, analyses of neural circuitry in species with a simple nervous system (e.g. insects) have shown that only a very small number of neurons is needed for apparently complex cognitive functions such as the ability to enumerate. This means that the evolution of a complex brain may not be required to possess this ability.
For more information, see the paper: Agrillo C, L Piffer and A Bisazza (2010) Large number discrimination by mosquitofish. PLoS ONE 5, e15232. doi:10.1371/journal.pone.0015232