Older theories of context effects in brightness perception assume that the effect of context can be mapped into a scalar which is somehow used to compute a target's brightness in relation to the context. This implies that for every complex context pattern there exists a homogenous or uniform context field which is equivalent to the complex context in the sense that it results in exactly the same context effect for every possible target.
We use a new method to find equivalent homogenous context intensities: A constant target patch is shown within a modulated context field. The context pattern has two limiting states. One is a high contrast complex pattern with 2, 4, or 8 different luminance patches. The other is a uniform field. The context smoothly changes from one state to the other. The two context pattern states are said to be equivalent if the context pattern change does not induce any change in the appearance of the target patch. The subject's task is to adjust the intensity level of the uniform field state until there is no longer any change in the appearance of the target patch.
There are 4 different types of complex context patterns for two factors of variation: The first factor is the harmonic mean of the context pattern intensities which is varied in 3 levels. The second factor is the range of context pattern luminance values. This factor is varied in 2 levels. The third factor investigated is the target patch intensity level. The target intensity ranges from below the context pattern intensities to above the context pattern intensities with 9 possible targets. We test whether harmonic mean or maximum of the context pattern intensities is a good predictor for the equivalent context.
The results show statistically significant effects of (1) average context pattern luminance, (2) context pattern range, and (3) target intensity. In general the equivalent homogenous context has an intensity which is slightly above the harmonic mean of the context pattern intensities. It also is strictly below the maximum context pattern intensity. The equivalent context, however, seems not to be unique for highest luminance targets. In this case there is only a very small effect of context modulation.
The effect of target intensity reveals the presence of a much stronger crispening effect for uniform context fields as compared to complex context patterns. This is confirmed in a follow up study which explicitly compares the crispening effect (edge enhancement for low but nonzero contrast) for uniform and complex patterns.
Dieser Eintrag ist Teil der Universitätsbibliographie.
Search Authors in