Abstract
It is generally agreed that when we look at the world, some aspects of the visual scene are encoded consciously while others are encoded unconsciously, or without our awareness that they were encoded. Although several studies have looked at unconscious encoding and its effect on subsequent behavior, it has thus far been impossible to determine which features of a given scene are encoded in this gray zone between seen and unseen. For the first time, we isolate the unconsciously perceived features of a stimulus by using “Bubbles” (Gosselin & Schyns, 2001), a technique that reveals the information in an image that is useful (‘diagnostic’) for a given task. Faces (n = 10) are presented in an unconscious repetition priming paradigm similar to that of Dehaene et al. (2001). The presentation sequence mask-blank-prime-blank-mask-target leads to conscious repetition priming. The inversion of the blanks and the masks abolishes the awareness of the prime but not the priming effect, thus generating unconscious repetition priming. In both conditions, the primes are presented under “bubble masks”. These masks only allow a proportion of the prime to be seen in the two-dimensional image plane and also selectively reveal information on the third dimension of spatial scale. Observers (n = 5) are asked to indicate by pressing the appropriate key, as precisely and as rapidly as possible, the gender of the target over 2000 trials. Then, a regression is run on the bubble masks and the response times and statistical analyses are conducted on the difference between the two conditions' regression coefficients. Preliminary results indicate a complex pattern of differences. Isolating the features of the visual scene that are unconsciously perceived is a crucial step towards identifying the neural substrates of unconscious perception.