Jessica Cantlon

Shared system for number representation in monkeys and adult humans

Although current evidence suggests that humans and non-human animals exhibit qualitatively similar performance on numerical tasks, few studies have directly compared the numerical abilities of humans and non-humans using the same task and stimuli. In this study, monkeys and adult humans were tested on the same ordinal numerical comparison task and we found qualitative and quantitative similarities in the psychophysical functions of their performance. Participants from both species were tested according to the same task parameters, on the same numerical values (2 to 30), and with the same stimuli.  Accuracy and reaction time for monkeys and humans were modulated by the ratio of the numerical values being compared, the pattern predicted by Weber’s law.  The qualitative similarity in performance between monkeys and humans substantiates claims that both species invoke a common psychological process to perform this numerical task.

Numerical representation across sets with high perceptual variability in monkeys

Rhesus monkeys were tested with a ordinal numerical task in which they had to choose the smaller numerical value from two arrays of dots.  The elements within each visual array systematically varied in color, size, shape and all combinations of these dimensions.  Both monkeys had prior training only with numerical stimuli that were homogeneous in the color, size, and shape of the elements.  Despite this previous training with exclusively homogeneous arrays, monkeys showed no evidence of a cost in accuracy or reaction time when tested with heterogeneous stimuli.  Furthermore, there was no evidence of improvement in accuracy on heterogeneous stimuli over the testing period supporting the conclusion that monkeys did not have to learn to ignore the perceptual variability of elements but did so spontaneously. 

Nonverbal numerical abstraction in preschool children

Children between 3 and 5 years of age were tested on their ability to estimate number when the sizes, shapes, and colors of the elements in an array were varied (heterogeneous condition) versus when they did not vary (homogeneous condition).  One group of children was tested on an ordinal task, identical to that tested in monkeys (above), in which the goal was to select the smaller of two arrays while another group of children was tested on a match-to-sample task in which the goal was to choose one of two visual arrays that matched the sample in number.  Children performed above chance on both homogeneous and heterogeneous stimuli in both tasks.  However, while children showed no impairment on heterogeneous relative to homogeneous arrays in the ordering task, performance was impaired by heterogeneity in the matching task.  Children’s performance on these tasks suggests that nonverbal numerical abstraction emerges early in development but, is task-dependent. 

A common numerical comparison algorithm in monkeys and adult humans.

Monkeys performed a numerical comparison task in which they had to choose the smaller of two numerical values when the background color of the task screen was red, but they were required to choose the larger of two values when the screen was blue.  Monkeys’ performance was affected by the semantic congruity between the ordinal direction of the comparison and the size of the numerical values being compared: monkeys were faster to choose the smaller of two small numerical values (e.g., 1 vs. 2) than two large values (e.g., 8 vs. 9) but they were faster at choosing the larger of two large values than two small values. Adult humans show similar semantic congruity effects when they compare stimuli from a variety of classes along an ordinal continuum, including numbers.  The finding of a semantic congruity effect in monkeys is evidence that monkeys and humans use the same mental algorithm for comparing numerical values.

Nonverbal addition in monkeys and adult humans

To investigate whether monkeys share a capacity for arithmetic reasoning with humans, monkeys and adult humans were tested on the same nonverbal addition task. Subjects were presented with two sets of dots on a touchscreen monitor separated by a delay.  Following the presentation of these two sets, monkeys were required to choose between two arrays: one with a number of dots equal to the sum of the two sets and a second, distractor array that contained a different number of dots. Monkeys performed this task at a high level of accuracy and their performance was qualitatively similar to that of adult humans.  These results offer a definitive demonstration of common mental arithmetic process shared by humans and non-human primates.

Functional imaging of numerical processing in adults and 4-year-old children

Using functional magnetic resonance imaging (fMRI) and an event-related fMRI adaptation paradigm, we found that adult subjects showed a greater IPS response to visual arrays that deviated from standard stimuli in their number of elements than to stimuli that deviated in local element shape. These results support previous claims of a neurophysiological link between non-symbolic and symbolic numerical processing in adulthood. In parallel, we tested 4-year-old children with the same fMRI adaptation paradigm as adult subjects to determine whether the neural locus of non-symbolic numerical activity in adults shows continuity in function over development. The IPS responded to numerical deviants similarly in 4-year-old children and adults. This is evidence that the neural locus of adult numerical cognition takes form early in development, prior to sophisticated symbolic numerical experience.