The importance of information localization in scene gist recognition.

Abstract

People can recognize the meaning or gist of a scene from a single glance, and a few recent studies have begun to examine the sorts of information that contribute to scene gist recognition. The authors of the present study used visual masking coupled with image manipulations (randomizing phase while maintaining the Fourier amplitude spectrum; random image structure evolution [RISE]; J. Sadr & P. Sinha, 2004) to explore whether and when unlocalized Fourier amplitude information contributes to gist perception. In 4 experiments, the authors found that differences between scene categories in the Fourier amplitude spectrum are insufficient for gist recognition or gist masking. Whereas the global 1/f spatial frequency amplitude spectra of scenes plays a role in gist masking, local phase information is necessary for gist recognition and for the strongest gist masking. Moreover, the ability to recognize the gist of a target image was influenced by mask recognizability, suggesting that conceptual masking occurs even at the earliest stages of scene processing.

The role of higher order image statistics in masking scene gist recognition

“…Fully phase-randomized scenes lack higher order scene statistics (Thomson, 1999(Thomson, , 2001a and are unrecognizable (Joubert et al, 2009;Loschky & Larson, 2008;Loschky et al, 2007). Thus, testing our hypothesis required comparison stimuli that, in contrast with phaserandomized images, did contain higher order statistical relationships such as those in natural scenes (Thomson, 2001a(Thomson, , 2001b, but, similar to phase-randomized images, were not recognizable.…”

“…Several studies have shown that incrementally destroying the higher order image statistics of scenes by incrementally randomizing their phase spectra, while preserving as masks, as compared with normal scenes, is that phaserandomized scenes are less recognizable; that is, more recognizable masks produce more masking-a phenomenon known as conceptual masking (Bachmann, Luiga, & Põder, 2005;Intraub, 1984;Loftus & Ginn, 1984;Loftus, Hanna, & Lester, 1988;Loschky et al, 2007;Potter, 1976). However, masking produced by unrecognizable texture masks could not be explained that way.…”

“…They used a scene gist masking paradigm that systematically disrupted the higher order image statistics (by incremental phase randomization) of scene image masks, which incrementally reduced gist masking strength. Going a step further, Loschky et al (2007) found that masking a target scene by a fully phase-randomized version of itself (which thus had identical second-order statistics to the target) caused no more scene gist masking than when the target scene was masked by a fully phase-randomized version of a scene from a different category (thus having different second-order statistics). These results suggest that (1) higher order image statistics carried in the phase spectrum are important for categorizing scenes, and (2) the differences in the second-order image statistics of scenes from different scene categories are not particularly useful for recognizing scene gist.…”

“…their second-order image statistics, reduces gist recognition from near perfect to chance levels (Joubert, Rousselet, Fabre-Thorpe, & Fize, 2009;Loschky & Larson, 2008;Loschky et al, 2007). Furthermore, Loschky et al (2007) provided converging evidence by building on the long history of using masking to study the structure of spatial vision (Carter & Henning, 1971;de Valois & Switkes, 1983;Henning, Hertz, & Hinton, 1981;Legge & Foley, 1980;Losada & Mullen, 1995;Sekuler, 1965;Solomon, 2000;Stromeyer & Julesz, 1972;Wilson, McFarlane, & Phillips, 1983).…”

“…Furthermore, Loschky et al (2007) provided converging evidence by building on the long history of using masking to study the structure of spatial vision (Carter & Henning, 1971;de Valois & Switkes, 1983;Henning, Hertz, & Hinton, 1981;Legge & Foley, 1980;Losada & Mullen, 1995;Sekuler, 1965;Solomon, 2000;Stromeyer & Julesz, 1972;Wilson, McFarlane, & Phillips, 1983). They used a scene gist masking paradigm that systematically disrupted the higher order image statistics (by incremental phase randomization) of scene image masks, which incrementally reduced gist masking strength.…”