Abstract
In repeated viewing of ambiguous patterns, perception is biased by positive and negative priming effects. In a companion abstract, we report that positive priming lasts 20–30 times longer than negative priming. Here, we exploit this difference to disentangle positive and negative priming by a given pattern. To establish generality, we compare five kinds of patterns (kinetic-depth-effect or KDE, two instances of binocular rivalry, Necker cube, Anstis dots). Nine observers participated.
For all kinds of ambiguous patterns, an unambiguous version negatively primes the ambiguous pattern: the two are rarely perceived in the same way (Psurvival ≈ 0), if no pause intervenes. When a pause is introduced, negative priming is no longer evident (Psurvival ≈ 0.5). When a pause separates two ambiguous patterns, positive priming is found in all cases (Psurvival ≈ 1.0). Thus, only ambiguous patterns appear to leave a “perceptual memory” which may bias subsequent perceptions.
To further study this particularity of ambiguous patterns, we generated KDE patterns with varying degrees of ‘bias’ for one percept or another (quantified separately for each observer). The dominance fraction under continuous viewing served as a measure (50% for unbiased and 100% for completely biased). Next, we asked how a biased pattern (“prime”) affects perception of a subsequent ambiguous pattern (“probe”)? With no pause between probe and prime, negative priming (Psurvival ≈ 0) gradually gives way to positive priming (Psurvival ≈ 1.0) as prime bias decreases. When a pause between probe and prime allows negative priming decay, statistically identical positive priming (Psurvival [[gt]] 0.5) obtains for prime bias 50% to 97%. No priming (Psurvival ≈ 0.5) occurs for 100% bias (unambiguous KDE).
We conclude that ambiguous patterns engage additional levels of processing where they leave a “perceptual memory”. Unambiguous patterns do not engage these levels and leave no “memory”.