The absence of formal perspective in most or all Roman painting is often read as “denial” of space (so explicitly La Rocca 2008, but essentially already in, e.g., Panofsky or van Blankenhagen) or (superficially more positive) as actively engaging the viewer to “oscillate” between “contradictory” and/or “ambiguous” features (e.g. Bergmann 1992, cf. Platt/Squire 2016). Substantively, I argue that both views are mistaken, but that considering the reason such views could arise in the first place leads us not just to refutation but to a more explanatory account. Methodologically, I suggest that this case helps us think through an important distinction between social ambiguity (different readings across a population) and cognitive ambiguity (different readings by a single agent).

(A) Both views stem from the correct observation that no unique “solution” maps any twodimensional Roman work of art into three-dimensional space. But the expectation of such mathematical/algorithmic regularity is an anachronistic constraint. Instead, in a case study mainly of sacro-idyllic images, artists offer varying sets of features—e.g. adjacency and overlap, grounding lines, narrative-associated characters, multiple color patterns (in fresco), thickness and incision (in stucco)—which serve as affordances by which viewers may exploit in various combinations to articulate individual elements and thus to construct spatial understandings. The complexity of the mechanism shows these works are very much spatial. Works by different artists work in objectively different ways, and the subjective readings of even a single work presumably varied from viewer to viewer. But those readings could still have been individually stable, rather than being fundamentally about ambiguity itself. That they not only could be but typically were so is suggested by the diversity of their “sets of affordances” that all deliver the same results. That is to say, the artists are united not by genre features that could be mechanistically (and so arbitrarily) transmitted, but by the fact that their differing systems all appear to serve the same (spatial) ends.

(B) The same preference for stability is also suggested by looking at plays on the framing (here in a roughly literal sense) of these images. Plays on the potential ambiguity of frame and image are surprisingly rare and typically quite modest (cf. Platt 2016). I offer as an “exception that proves the rule” two stucco reliefs from a single room under the Farnesina which exhibit a much more complex visual jeu d’esprit. The specific contrast in this respect between these works and otherwise similar objects in near-by rooms is paralleled by differences in the overall style of their decor. The more elaborate perspectival games are in a room in which the overall stability of pictorial illusionism is weaker (see Squire 2016). That is to say, the style of each room’s decorative scheme serves as a frame (in the theoretical sense) that orients the viewers’ interpretation of the individual components – including, crucially, the represented frames (now again in the literal sense).