Abstract
Cognitive flexibility, the capacity to adapt to changing conditions, is often assessed with reversal learning, in which a learned association must be updated after reward contingencies change. Trials-to-criterion (TTC) is a widely applied learning threshold, but it can misrepresent performance; some individuals improve steadily but fail to reach the criterion due to variability (false negatives), while others meet it through a spike without sustained learning (false positives). We evaluate TTC limitations and demonstrate learning curve analysis as a more nuanced approach to investigate learning dynamics. We tested wild striated caracaras (Phalcoboenus australis) using a two-choice discrimination task followed by a reversal task and compared TTC with trial-level modelling. Although the group showed overall improvement, individual trajectories varied widely. TTC both over- and underestimated learning, misclassifying inconsistent performers and overlooking gradual improvers. In contrast, learning curves captured trajectory, stability and consistency of change. We argue that continued reliance on binary thresholds obscures the dynamics of learning, and that slope- and trajectory-informed analyses provide a more accurate and ecologically valid framework for assessing learning in the wild.