Abstract
The classic finding of a basic-level advantage in categorization (Rosch et al., 1976) has been challenged by recent findings in ultra-rapid categorization tasks (Thorpe et al., 1996). Rapid decisions made in a glance suggest that superordinate (e.g., “animal”) representations are accessed as fast as or faster than basic (e.g., “bird”) representations (Macé et al., 2009; Mack & Palmeri, 2010). The conclusions from classic and rapid categorization are in conflict. That could be because the two different paradigms underlying these conclusions differ on several critical factors.
We attempted to make sense of these contradictory conclusions by investigating both the effect of stimulus exposure and blocking of target categories on the speed and accuracy of categorization decisions. We crossed two different exposure durations (25 ms versus 250 ms) and two different category orders (blocked versus randomized) across four category verification experiments where animals and plants were categorized at the superordinate, basic, and subordinate levels. As expected, with brief exposures and blocked target order - conditions typically used in the rapid categorization paradigm - superordinate categorization was as fast and accurate as basic categorization. In contrast, we observed a basic-level advantage in the three other conditions. With a random target order and brief exposures and with longer exposures responses were fastest and most accurate for basic categorization.
These results suggest that the experimental context of rapid categorization may evoke specific processing strategies not found when using what may arguably be considered default categorization strategies. Classic and rapid categorization results may not be in conflict per se. Instead, these two paradigms characterize different aspects of the perceptual categorization system. You may spot the animal faster than the bird, but only at a glance and when finding animals is the only thing on your mind.