Abstract
We have all seen a human face in a cloud, a pebble or in blots on a wall. Evidence of “superstitious” perceptions have been documented since classical antiquity (Gombrich, 1960; Janson, 1973) but have received little scientific attention. Here, we used superstitious perceptions in a new principled method to reveal the properties of unobservable object representations in memory. We presented 20,000 complete or partial white noise fields to our observers. Three observers saw an ‘S’ letter in a significant proportion of the noise fields in Experiment 1, and three observers saw a smile in Experiment 2. We extracted the first and second Wiener kernels (Wiener, 1958) subtending these perceptions. The first Wiener kernels were similar to an ‘S’ and a smile in Experiment 1 and 2, respectively. Moreover, they possessed spectral properties compatible with those reported in the psychophysical literature (e.g., Bayer, Schwartz & Pelli, 1998; Solomon & Pelli, 1994). These results suggest that purely top-down processes are responsible for our subjects perceptions. The possibility remains, however, that some bottom-up structure infiltrated our noise fields. In Experiment 3, the same sequence of 20,000 noise fields was presented to two observers: One saw a ‘Y’ letter and the other an ‘H’ letter, and, as expected, the corresponding first Wiener kernels were similar to a ‘Y’ letter and to an ‘H’ letter, respectively.