Purchase this article with an account.
Nicolas Dupuis-Roy, Frédéric Gosselin; Examining the top-down component of perceptual learning. Journal of Vision 2005;5(8):862. doi: 10.1167/5.8.862.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Perceptual learning (PL) has been observed in tasks as dissimilar as pop-out detection (Ahissar & Hochstein, 2004), illusory contours shape discrimination (Rubin, Nakayama & Shapley, 2002), face identification (Gold, Bennett & Sekuler, 1999) and visual position discrimination (Li, Levi & Klein, 2004). PL can be rapid (Fahle, Edelson & Poggio, 1995) or require several months of practice (Watanabe et al., 2002). Usually the improvements are preserved after several years without practice (Karni & Sagi, 1993) but sometimes retention is only partial (Santhian & Zangaladze, 1998; Kapadia, Gilbert & Westheimer, 1994) or even null (Godde, Stauffenberg, Spengler & Dinse, 2000). A part of this remarkable diversity in PL manifestations may stem from the respective involvement of top-down and bottom-up processes in the task. Some attempts have been made to disambiguate the effects of these processes by isolating them (Godde et al., 2000; Watanabe, Nanez & Sasaki, 2002; Ahissar & Hochstein, 1993, 2000, 2002). Last year we have isolated top-down components of PL in a detection task by completely removing the signal in the practice trials (see Gosselin & Dupuis-Roy (2003) for methodological details). Here, we further examine this type of PL. Results show that the improvement is preserved after a one-year interruption. Comparison of our subjects' performance to that of an ideal observer indicates that no-signal training increases efficiency and decreases internal noise.
This PDF is available to Subscribers Only