Invariance is a property of perception. This is the property of perception that allows us to recognize an object immediately, even though we see it from many different angles or perspectives. For example, if you placed five identical elongated boxes around a room along with one square box, even if the various boxes are placed in different positions so their objective images differ, an observer would quickly pick out the box that was the exception. The perception of the pattern of the elongated boxes does not vary (is invariant) with variances in objective images.
The invariance principle functions in conjunction with other perceptual properties. Those properties are called emergence, reification and multistability. Emergence means that a complex pattern emerges whole, not as a calculation of its parts. A child sees a dog, and instantly recognizes a dog, not paws, head, tail, etc. Reification is the experience of a pattern even when it is not all there. You look at the leaves on the ground, see only the tip of a tail and a portion of a head, and you recognize a snake. Multistability is bouncing back and forth between more than one perception of the same image. The classic example is the black and white picture that looks like a vase or, after studying it, looks like two silhouettes of heads facing each other.
Gestalt theory does not explain how invariance works; it only describes it. The actual activity of the brain during perception is incompletely understood. Scientists know that an area called the visual-ventral pathway, a neuronal network that somehow encodes information, is activated during perception. This activity is stimulated by observation, and brain researchers have identified two principles at work in this complex process: selectivity and invariance, the latter concept being shared by Gestalt psychologists.
Australian neurologist David Rail has incorporated the Gestalt principle of invariance in a model of language, particularly in the use of metaphor. Language and perception, according to Rail, are interdependent. Just as the elongated boxes in the earlier example have the same "meaning" viewed from different angles, similar meanings constructed linguistically can emerge from descriptions of different things. Instead of visual perception of relations between objects in the case of the boxes -- where the meaning does not vary -- metaphors rely on recognition of patterns in linguistic wholes called tropes.