Abstract
On the one hand, perceptual learning (i.e. a performance improvement in a given perceptual task following practice) has been shown to depend on low-level, “bottom-up” processes (e.g. Crist, Li & Gilbert, 2001; Gold, Bennett & Sekuler, 1999, Karni & Sagi, 1991). A recent series of experiments by Watanabe, Nañez & Sasaki (2001, 2002) even demonstrated that it could occur in the absence of high-level, presumably “top-down” factors such as awareness. On the other hand, several studies suggest that “top-down” processes can at least modulate perceptual learning (e.g. Shiu & Pashler, 1992; Ahissar & Hochstein, 1993; Ito, Westheimer & Gilbert, 1998). Nobody, however, has ever examined “top-down” perceptual learning in the absence of “bottom-up” perceptual learning. Here, we report such an experiment.
We employed the “superstitious perception” paradigm developed by Gosselin & Schyns (in press). We instructed our subjects (n = 10) to indicate the presence or absence of a noisy ‘X’ letter —that was described to them but never shown— in each of the1000 (n = 5) or 5000 trials (n = 5) that they were presented. Unbeknownst to them, the stimuli in all trials contained only white Gaussian noise. After this purely “top-down” treatment, the sensitivity of each subject was computed classically. Preliminary results indicate that perceptual learning can take place in the absence of bottom-up signal.