Abstract
Snowden & Milne (1997, Curr. Biol., 7, 717–722) reported evidence for the existence of higher level — complex motion — detectors. They used a radial stimulus divided in quadrants where only two quadrants (e.g. quadrant 1 and 3) contained moving dots. Apart from the classic motion aftereffect (MAE) in the quadrants that actually contained moving dots, they showed that motion adaptation could be measured for (locally) non-adapted locations also. This phenomenon is known as a phantom MAE. It is likely that this indicates a global gain. Interestingly, the threshold elevation for the non-adapted locations was smaller than for the locally adapted regions.
The starting point of this study was to gain more insight in the mechanisms underlying this difference. We tested a linear model where we assumed that the coherence threshold for the directly adapted regions is a product of the setting of the gains of the local motion sensors plus the gain of the global system. This opens up a way to quantify the contribution of the local gains: in the areas that were not directly stimulated, the threshold elevation is due to the setting of the global gain only, whereas the threshold in the stimulated part is due to both the local as well as the global gain. As a result the value of the local gain is the coherence threshold for the adapted region minus the threshold value in the phantom area.
This model opens up possibilities to quantify the contribution of even higher order mechanisms like attention. For example, by paying attention to only one of the quadrants (either containing motion or not).