Numerosity Estimation in Accumulated Spatial Arrays: Does Anchoring Limit Accuracy?

Poster Presentation 43.326: Monday, May 22, 2023, 8:30 am – 12:30 pm, Banyan Breezeway
Session: Scene Perception: Spatiotemporal factors

Frank Durgin1 (), John Grey Crosby1; 1Swarthmore College

The estimation of temporal number tends toward increasing underestimation without calibrating feedback. In a previous study it was found that estimates for temporal number showed increasing underestimation starting at as few as five pulses. In contrast, number estimation of briefly-presented (400 ms) 2D spatial arrays tended to be linear up to about 20 without calibrating feedback. Here we consider hybrid temporal/spatial number displays in which units accumulate over time. In principle, such displays ought to be as accurately estimated as spatial arrays presented all at once, but in practice estimation is poorer, falling below linearity at about 13-15 dots. Here we tested whether underestimation could be improved for such displays if a clear end signal was given when accumulation was complete. The number of dots ranged from 2 to 58 accumulating at one of two probabilistic rates. In the base condition, the display remained present for 1 s after accumulation was complete. In the auditory condition, a click was played to signal the that accumulation was complete, followed by a final 1 s of display. In the visual condition, the accumulated display was flickered off for 100 ms and then on again prior to the final 1 s display. Estimates in all three conditions were linear only up to 13 dots, and showed the same increasing underestimation thereafter. Because the dots could accumulate over several seconds it may be that the growing estimates during the accumulation process tend to anchor later estimates to a lower value than if the display had been presented all at once. Although information theory suggests that combining information from multiple sources (e.g., temporal and spatial) should improve discrimination, it does not guarantee a reduction in bias.