<p>Participants were trained and tested on their ability to discriminate among geometric angles presented to them as freestanding three-dimensional objects located within a larger experimental room. Three different angle sizes were used during training and participants were grouped based on their assigned training angle (110°, 135° or 160° angle), which they learned to discriminate from the remaining two angles. Following a learning phase, participants experienced test trials in which their training angle was paired with a novel test angle, and they made a choice between the two angles. Participants in the smallest angle group (Group 110) biased their choices toward test angles that were smaller than their training angle; conversely, participants in the largest angle group (Group 160) biased their choices toward test angles that were larger than their training angle. When it came to trials involving fine discrimination between angles, both groups showed a high degree of precision when needed. Results support cognitive strategies of geometric estimation that include relational rule learning and more fine-grained judgements to maximize overall accuracy.</p>