Do background depth gradients facilitate object identification?

A number of experiments have demonstrated that an object in a coherent scene can be more accurately identified than when in a jumbled scene or a display of unrelated objects. An experiment was designed to test whether the identification of an object (from a set of six) would be faster and more accurate if the objects were presented against a background gradient that conferred a unifying depth effect on the six objects. No benefit of this background was found. In fact, the accuracy of the identification of objects against a depth background was found to be lower than that of objects which were presented against either no background or a control regular grid background. It is concluded that the perceptual advantage of objects appearing in a well-formed scene derives from semantic relations that are defined between such objects.