Associative learning: does performance in a simple discrimination task predict success in a more complex one?

Understanding why individuals differ in their cognitive performance is a central goal of cognitive ecology. A major challenge, however, lies in designing cognitive tasks that accurately assess cognitive performance and cognitive diversity. Here, we investigated whether individual performance metrics in a simple associative learning task predicts metrics in a more complex one, using zebra finches ( Taeniopygia guttata ). We performed two visual discrimination tasks: a simple task (one colour always providing five seeds, the other always empty) and a more complex task (one colour providing five ± one seeds, the other providing two ± one seeds). We assessed three performance metrics: number of trials to reach a learning criterion (learning score), behavioural reaction norm (learning speed), and best linear unbiased prediction (learning strength) of individual choices. Birds learned faster and achieved greater learning strength in the simple compared to the complex task, confirming that trial-by-trial changes in reward-magnitude increased discrimination difficulty. Yellow and pink cues significantly enhanced learning speed in both tasks, irrespective of reward contingencies. More importantly, interindividual differences in learning score, speed and strength were significantly greater in the complex task, and simple-task performance metrics poorly predicted complex-task performance metrics. Our findings demonstrated that differences in reward magnitude and cue properties jointly influence associative learning performance, with task complexity emerging as a major determinant of interindividual differences in learning performance. Since individual choices result from the interplay of cognitive abilities, reward structures and stimulus properties, these findings emphasize the importance of manipulating task complexity in experimental designs to better capture interindividual cognitive diversity.