Abstract
We have created outline versions of a standardized set of line drawings of everyday objects (Snodgrass, J.G., & Vanderwart, M., 1980, Journal of Experimental Psychology: Human Memory and Learning, 6, 174–215) by removing all interior details and spline fitting the extracted contours. Because we have a full curvature graph of each of these contours, we could create 40 different fragmented versions of each contour: 20 levels of increasing fractions of the total contour, either starting from the extrema of curvature (positive maxima and negative minima) or starting from the inflections (where curvature goes through zero when it changes from positive to negative). Based on a pilot study we ensured that these fragmentation levels yielded equally increasing identification rates for both the extrema (E) and the inflection (I) condition. The experiment consisted of two phases, a “study” phase and a “test” phase, each consisting of a number of trials in which a complete contour was gradually built up in 20 steps of 500 msec each. Subjects pressed a button as soon as they identified the picture. In the test phase, one third of the objects were new (“unprimed”), two thirds had also been shown in the study phase (“primed”). Priming was measured as the difference in identification level between study and test phase. The primed stimuli could either be built up from the same fragments (E in both phases or I in both phases) or from different fragments (E in study and I in test, or vice versa). The first of these conditions reflects both conceptual and perceptual priming (same object and same fragments), whereas the second reflects only conceptual priming (same object but initially different fragments). Results showed strong overall priming effects and a small but significant advantage for the additional perceptual priming when the same fragments were used. We will discuss these findings in terms of the processes and representations involved in picture identification from contour fragments.