Fig. 4.

Correlations of the individual test performances in Experiment 3: Visual learning experiments with restrained bees. A A direct statistical correlation could not be performed, as we could only use the learner bees in the 1st phase as successful reversal of reward contingencies in the 2nd phase of reversal learning prerequisites learning the initial discrimination. We still observed that a majority of the learners in the 1st phase of reversal learning (1st RL) were also successful in the 2nd phase of reversal learning (2nd RL; 59%). 84% of the bees that failed to learn in the 1st phase, and were thus not used for the correlation, remained non-learners in the 2nd phase of reversal learning. B The individual test performances were positively correlated between the 1st phase of reversal learning and negative patterning (NP; Spearman rank correlation; n = 140, rho = 0.18, p = 0.03). Indeed, the majority of learners (57%) and non-learners (75%) in the 1st phase of reversal learning remained in their category in the negative patterning paradigm. C The individual test performances of the 2nd phase of reversal learning were not significantly correlated with negative patterning (Spearman rank correlation; n = 61, rho = 0.18, p = 0.17). While 69% of the learners in the 2nd phase of reversal learning were also successful in negative patterning, half of the bees (51%) that failed in the 2nd phase of reversal learning were nevertheless successful in negative patterning