[1] Around 1425 Filippo Brunelleschi displayed a small painted panel of the Florentine Baptistery, and viewers who studied its reflection through a peephole found it nearly indistinguishable from the building itself. [2] The demonstration is usually treated as the founding moment of linear perspective, the technique by which a painter projects three dimensions onto a flat surface as though that surface were a pane of glass held upright between the eye and the world. [3] A decade later Leon Battista Alberti set the method down as a teachable procedure, and perspective soon carried the prestige of geometry: it seemed not a mannerism but a discovery of how seeing actually works. [4] On this account the Renaissance painters had recovered the true grammar of vision, and earlier art looked like the work of people who had not yet learned to see.
[5] That confident story has been unsettled by a line of criticism descending from the art historian Erwin Panofsky, who argued that perspective is not a transcription of sight but a "symbolic form," a convention as particular to its culture as a style of handwriting. [6] The physiological case is easily stated. [7] The perspective construction assumes a single motionless eye fixed behind an unmoving frame, whereas people look with two eyes that are never still, and the surface on which the image finally lands is curved, not flat. [8] Straight lines the construction keeps straight are, on a curved retina, projected as gentle arcs. [9] So the "window" of perspective is no window but a grid laid over experience, which no viewer from another tradition need accept as natural.
[10] Yet the revisionist claim can be pushed too hard. [11] To call perspective a convention is not to call it arbitrary, for the construction still predicts relationships that any observer can check: objects of equal size do dwindle with distance at a measurable rate, and a tiled floor does foreshorten in just the way the method specifies. [12] A code that anticipates such measurements is doing something a style of handwriting never does. [13] Perspective is therefore best understood as a disciplined selection — a decision about which features of ordinary seeing to keep on the panel and which to leave off — rather than as either a neutral window onto the world or a merely local habit answerable to nothing beyond itself. [14] The Renaissance achievement was genuine, but it was an achievement of art, not a theorem borrowed from optics.
