Purchase this article with an account.
Jeffrey S. Johnson, Bruno A. Olshausen; Early target related processing in the discrimination of natural objects. Journal of Vision 2002;2(7):690. doi: 10.1167/2.7.690.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Human ERP studies have suggested that visual processing of natural scenes occurs rapidly, with differences between object categories under discrimination emerging by 150 ms after presentation. These early differences have been suggested to be related to the target status of a given image, which implies that they represent an upper limit for the time required to perform object recognition processes. Previous studies, however, leave open the possibilities that these differences could be the result of pre-object recognition visual processing or post-target decision motor processes. Target and nontarget images in previous studies have differed in spatial frequency content and background contextual cues which might affect pre-object recognition visual processing. To address this issue, we prepared a set of images consisting of cutout natural objects of widely varied categories placed against a pseudonatural background. We presented these images to volunteer subjects and asked them to perform a forced choice target/nontarget discrimination. A variable target cue before each image ensured that, across subjects, the images seen in target and nontarget conditions were identical. To address whether the differences might be due to motor processes, a separate set of subjects performed a go/no-go version of the task above, in one block pressing a button only for target images, in another pressing a button only for nontarget images. The results of these experiments demonstrate clear target-related differences. However, the differences arise later (∼170 ms) and more gradually than those seen in previous experiments. These results confirm that the visual processing of natural objects occurs very rapidly but leave open the possibility that the earliest differences reported previously do not represent an upper limit on the time required for object recognition.
This PDF is available to Subscribers Only