Abstract


= PDF Reprint,     = BibTeX entry,     = Online Abstract


Click to download BibTeX data Clik to view abstract V. Navalpakkam, L. Itti, Attention can be guided to the relevant feature category, In: Proc. Vision Science Society Annual Meeting (VSS05), May 2005. (Cited by 2)

Abstract: When the visual system is challenged by distractor heterogeneity and target-distractor similarity, can search be sped up by guiding attention to the relevant feature category, i.e., one that selectively promotes the target, and inhibits the distractors? Previous studies measuring reaction time suggest that for small feature differences between the target and distractors, search is inefficient when the target is flanked by distractors in feature space (D'Zmura, Vis Res 1991;31(6):951-966). This has been widely demonstrated in size (Treisman and Gelade, Cog Psy 1980;12:97-136), orientation (Wolfe et. al, J Exp Psy 1992;18(1):34-49) and color (Bauer et. al, Vis Res 1996;36(10):1439-1465). A widely accepted inference from these studies is that attention cannot be guided to a category, otherwise search for the medium target would be efficient. But this inference need not be true. We verified through eye tracking methods that despite inefficient search for the medium target (in size and color), a significantly higher number of fixations landed on items in the medium category than less or high categories. Our results suggest that indeed attention can be guided to a category. To reconcile previous results with our data, we propose a new computational mechanism which suggests that feature dimensions are encoded in cortex by broadly tuned ``categorical'' channels and that top-down influence can selectively boost the relevant category. Hence, inefficient search for the medium target occurs due to increased overlap between target and distractor categories, so that boosting the relevant medium category may falsely activate some distractors that belong to overlapping categories.

Themes: Model of Bottom-Up Saliency-Based Visual Attention, Model of Top-Down Attentional Modulation, Human Eye-Tracking Research

 

Copyright © 2000-2007 by the University of Southern California, iLab and Prof. Laurent Itti.
This page generated by bibTOhtml on Wed Feb 15 12:13:56 PST 2017