Why do ambiguous figures reverse?

For much of this century the dominant theory of reversal of ambiguous figures has been neural satiation. We suggest a very different kind of theory based on the consequences of the typical instructions given to subjects in experiments on reversal. In analyzing these instructions in preliminary experiments we conclude that what matters is the subject's knowledge that the figure is reversible and specific knowledge of how its alternative construals appear. This leads to an intention to reverse the figure. In one experiment we demonstrate that recent experience with the ambiguous figure and its two modes of appearance does not lead to its reversal if the subject does not know the figure is reversible. In another experiment we demonstrate that when the subject does know the ambiguous figure is reversible, it nonetheless will not tend to reverse if the intention to do so is contravened by a different intention. We suggest that the intention to reverse cannot succeed unless the subject "knows" how to carry it out and, further, that the intention may become involuntary and automatic.