Slow-fast differences in cognition among individuals have been proposed to be an outcome of the speed-accuracy trade-off in decision-making. Based on the different costs associated with acquiring information via individual and social learning, we hypothesized that slow-fast cognitive differences would also be tied to the adoption of these different learning modes. Since foragers in honeybee colonies likely have both these information acquisition modes available to them, we chose to test them for inter-individual differences in individual and social learning.

Individual honeybees foragers were presented with a learning task: making a choice between two types of flowers one rewarding (offering sucrose) and one unrewarding (offering water). This task was presented twice: 1) the bee was alone (individual learning) and 2) there was a social cue (model bee) on the correct color. For every individual, the raw data consisted of 20 successive choices between rewarding and unrewarding flowers in each of the two learning tasks. From these data, learning curves were constructed by calculating an accuracy index as the proportion of correct choices for rewarding in the 4 visits up to and including that visit, and a logisitc function was fit to the accuracy index. From this logistic function we calculated individual learning parameters: maximum individual learning score, individual learning rate, maximum social learning score, social learning rate. 

Our results support the existence of a speed-accuracy trade-off in both the individual and the social learning contexts. However, the trade-off is steeper during individual learning, which was slower than social learning but led to higher accuracy. Most importantly, our results also show that bees which attained high accuracy on the individual learning task had low accuracy on the social learning task and vice versa.

Individual foragers were measured for both indivdual and social learning ability. The learning task consisted of presenting foragers with two different colored flowers: one rewarding and one non-rewarding. Individual decisions (correct or incorred) were recorded for a total of 20 deceisions. During the individual task, foragers completed the learning task alone, while during the social learning task, a social cue (model bee) was placed on the correct color morph. Scores were then fit to a logisitc function to extract two parameters: learning rate (the slope) and maximum accuracy (asymptote) in both contexts. This data set includes indivdual scores (learning rate and maximum accuracy), age and colony of origin.