Semantic Priming effects on the Kanizsa Illusion: a case for cognitive penetrability

Abstract Visual illusions are considered key examples for cognitive impenetrability, as they are held not to be affected by non-perceptual processes. We revisit this claim in five experiments (N=1148; four preregistered) focused on the Kanizsa illusion, where a nonexistent shape is experienced within illusory contours. Pac-Man-like shapes inducing the illusion were presented after primes that were either semantically related to the Pac-Man game or not. We hypothesized that semantic primes would promote interpreting the shapes as individual Pac-Man characters, thus biasing participants away from the holistic Kanizsa illusion. Indeed, we found that the Kanizsa shape was detected less when participants were primed with Pac-Man-related stimuli. We then also demonstrated the opposite effect: a prime indexing the illusory shape (‘Triangle’) enhanced the probability of seeing the illusion. Together, our results suggest that semantic priming can both reduce and increase the probability of experiencing the Kanizsa illusion, thus supporting claims of cognitive penetrability.