Contrast discrimination, non–uniform patterns and change blindness

• Published in: Proceedings of the Royal Society. B, Biological sciences Vol. 265; no. 1411; pp. 2159 - 2166
• Main Authors: Scott-Brown, K C, Orbach, H S
• Format: Journal Article
• Language: English
• Published: England The Royal Society 22.11.1998
• Subjects: Adult; Attention; Attention - physiology; Blindness; Blinking; Change Blindness; Contrast Increment Thresholds; Contrast Sensitivity - physiology; Discrimination (Psychology) - physiology; Humans; Luminance; Memory; Memory disorders; Mental stimulation; Modeling; Noise prediction; Pattern Recognition, Visual - physiology; Pattern Vision; Psychophysics; Saccades; Set Size Effect; Signal noise; Space life sciences; Visual Memory; Visual Perception - physiology
• ISSN: 0962-8452; 1471-2954
• DOI: 10.1098/rspb.1998.0553

Change blindness-our inability to detect large changes in natural scenes when saccades, blinks and other transients interrupt visual input-seems to contradict psychophysical evidence for our exquisite sensitivity to contrast changes. Can the type of effects described as 'change blindness' be observed with simple, multi-element stimuli, amenable to psychophysical analysis? Such stimuli, composed of five mixed contrast elements, elicited a striking increase in contrast increment thresholds compared to those for an isolated element. Cue presentation prior to the stimulus substantially reduced thresholds, as for change blindness with natural scenes. On one hand, explanations for change blindness based on abstract and sketchy representations in short-term visual memory seem inappropriate for this low-level image property of contrast where there is ample evidence for exquisite performance on memory tasks. On the other hand, the highly increased thresholds for mixed contrast elements, and the decreased thresholds when a cue is present, argue against any simple early attentional or sensory explanation for change blindness. Thus, psychophysical results for very simple patterns cannot straightforwardly predict results even for the slightly more complicated patterns studied here.